Regular readers of The Odin Times will recognise many of these papers from past articles. The directory is part of our goal to take this further, encouraging individuals to examine the literature and use it to feed their own unique and objective model of the world.

There’s a vast and criminally underutilised body of academic research looking at venture capital and the associated disciplines.

This research is valuable, because academics are unlikely to have vested interests and their reasoning and methodologies are transparent. The work is observational, but it is also objective and anchored in reality.

It’s a profound contrast with the output and consumption of private industry, where almost everything is an anecdote built around a narrative. This is the overriding state, whether that’s a result of calculated behaviour or simple self-conditioning.

This preference for short-form insight is mostly the result of processing fluency, where simple ideas travel faster, but two other factors amplify the problem:

1. Opacity - It’s hard to make a complex argument when the facts are thin.

2. Insecurity - The desperation for confidence creates an aversion to complexity.

If you were to draw these two factors out as a venn diagram, venture capital would be in the dead centre.

This is why, as an industry, venture capital is uniquely vulnerable to the seductive coherence of simple ideas, rather than complex truths. That’s unfortunate, because venture capital is also uniquely dependent on intellectual diversity, as evident in the damage done by group-think versus the extreme profitability of contrarianism.

This stacks on existing research which illustrates how social media creates echo chambers that amplify consensus ideas while filtering out unconventional or contrarian views — which in turn builds on existing theory that describes how individuals self-censor opinions when they suspect they are in the minority.

“To the individual, not isolating himself is more important than his own judgement. […] This is the point where the individual is vulnerable; this is where social groups can punish him for failing to toe the line.”

Elisabeth Noelle-Neumann, Spiral of Silence (1974)

As the world has become louder and more connected, the outsized power of ideas that can cut through the noise, regardless of their merit, has become more obvious. This wielding of memetic influence is now a field of study that crops up in everything from marketing playbooks to industrial strategy and geopolitics.

It’s no coincidence that venture capital’s propaganda era began around the same time as a paper discussing the concept of “proximal mass opinion leaders” which essentially sets out the theory behind programmes like a16z’s New Media Fellowship.

More than ever, people are keen to tell you how they see the world; to tell the bigger story. To share their model of the world with you, as if it were a map that you could also follow to success. And, thanks to social media, it’s getting harder to resist.

This is an existential problem for venture capital, which relies on people having the courage and curiosity to build their own uniquely idiosyncratic models of the world. And the courage to hold onto those models in the most difficult conditions.

The Stars Can’t Lie

“Last of all he will be able to see the sun, and not mere reflections of him in the water, but he will see him in his own proper place, and not in another; and he will contemplate him as he is.”

Plato, “Allegory of the Cave” The Republic (380 to 375 BC)

Consider the lesson of early maritime explorers, who are often cited as an allegory for venture capital with many obvious parallels.

Like venture capitalists, they searched for unclaimed territory. The unbounded opportunity of new frontiers. They used maps that approximated the known world and stars as the source of truth which allowed them to navigate beyond it.

These explorers, who made their own maps by necessity, offered a significant advantage for their patrons. For example, from 1488 to 1499, only the Portuguese knew how easy it was to navigate around the Cape of Good Hope. The more widely available maps tended to exaggerate the difficulty of the journey, which deterred others.

The same is true for Russia’s deception about the Siberian longitude in the 1760s, designed to discourage European merchants from using the Northeast Passage. Or Captain Cook’s maps of New Zealand in the 1770s that obscured harbours and safe passage around the islands. Or the Dutch East India Company guarding their knowledge about passage through the Le Maire Strait.

So, maps were also used to deter or mislead, increasing the risk of voyages or discouraging exploration entirely. A precursor to today’s information warfare; the goal was to be seen as an authoritative source while preserving some asymmetric advantage — the ability to exploit a frontier before it became competitive.

So, either you embrace the risk of exploration, following the stars into uncharted waters, or you just go where a map tells you to go — knowing that whoever produced it was, for one reason or another, happy for you to do so.

Simulacra and Simulation

“Today abstraction is no longer that of the map, the double, the mirror, or the concept. Simulation is no longer that of a territory, a referential being, or a substance. It is the generation by models of a real without origin or reality: a hyperreal.”

Jean Baudrillard, Simulacra and Simulation (1981)

Maps, therefore, have power over humanity, by dictating reality to those who don’t know better. In fact, when maps are sufficiently influential, they become our reality.

This is the perspective of French philosopher, Jean Baudrillard.

Imagine a map that showed a canal traversing a peninsula. An arriving ship would be so convinced that the canal must exist, because their course runs through it, that they might end up digging it out themselves. The determination would essentially be that, of the two possibilities, reality was the incorrect party.

The same is true for less literal maps, whether they relate to the hierarchy of ecosystems, the flow of capital, or the focus of attention. A sufficiently authoritative map can organise the world around whatever reality it asserts. And wherever reality is uncomfortably complex or uncertain, people are keen to embrace simple abstractions that offer confidence.

Consider, for example, the fundamental truth of venture capital; there is no formula for success. The biggest winners are rarely the expected winners. There are literally no easy answers in the pursuit of alpha. The industry’s response to this is to manifest an abstract reality of categories, heuristics, and hierarchies around which people and capital can organise themselves. To pretend that there are such things as “obvious winners”, when the term is an oxymoron in the context of investing.

Inside this hyperreality, concepts emerge and embed on the merit of how well they match up with expectations and preferences, rather than any connection to reality. This is fertile ground for memes; simple narrative devices that shape expectations and preferences more easily than a person might alter reality. If you want a boom of enterprise software, sometimes it helps to say there is a boom of enterprise software.

This is evident in the many ideological artifacts of venture capital, most of which have no empirical basis, or are even contrary to empirical findings. These nuggets of common wisdom are not a distillation of reality, but a means by which reality is replaced with something more convenient. Bite-sized and reassuring.

While this was initially the natural response of human psychology, it is increasingly wielded with purpose to shape behaviour and modify outcomes.

It is, effectively, a narrative wargame within a hyperreality of capital allocation.

World Models

“If the organism carries a ‘small-scale model’ of external reality and of its own possible actions within its head, it is able to try out various alternatives, conclude which is the best of them, react to future situations before they arise, utilise the knowledge of past events in dealing with the present and future, and in every way to react in a much fuller, safer, and more competent manner to the emergencies which face it.”

Kenneth Craik, The Nature of Explanation (1943)

The final evolution of the map is the world model; an all-encompassing reflection of reality in which ideas are tested, simulations are run, and complexity is erased. Everyone has a world model, as a reflection of their life experiences and the information they have consumed. But, again, there is danger.

It’s difficult and uncomfortable to try fitting the whole world in your head. To travel widely, to debate, to scrutinise facts and consume data. So, inevitably, our world models are streamlined simulations. Like a video game that lowers the quality of assets when they’re too far in the distance, or around a corner.

Where we lack confidence, we borrow from the models of others. We accept the CNN model of politics, or the Fox News model, whichever fits our priors. We don’t know what to make of AI, so we assimilate Ed Zitron’s scepticism or Brad Gerstner’s optimism. Like a celebrity endorsement of an otherwise lazy world view.

A consequence of this is that building and selling world models becomes a seriously appealing proposition. It is effectively the commercial playbook for 5th Generation Warfare, applied to corporate interests rather than national interests. And, thanks to our base human nature, we are inclined toward the simplest and most affirming option; the GPT-4o of world models. Tell us we’re the good guys, that we’re correct, and that the people we’re secretly jealous of are evil.

For anyone seeking power, learning to pull these levers is obviously attractive. So, like the explorers and their patrons, they create authoritative-seeming world models that serve a hidden agenda, and market them as morally correct, financially lucrative, or more prestigious. These models promise all the benefits of intellectualism, with none of the hard work required, so their users are not inclined to test their validity.

For the venture capitalist, this is death, as characterised by the generation of prestige-seekers that joined the industry during ZIRP. They all adopted variations of the same world model, chased the same opportunities, and experienced the same disbelief and confusion when there was no pot of gold at the end of that rainbow. They did not even have the instinct to reflect on why their approach was doomed from the start.

But venture capital is also where this proposition is most tempting. The long horizons of uncertainty, battling peer pressure and market volatility, are uniquely conducive to insecurity. And what better way to reduce insecurity than to absorb a world model of seemingly successful peers through endless podcasts and posts. The comfort of the herd, and the complete annihilation of alpha.

Ultimately, this may be why emerging managers and outsiders demonstrate aggregate outperformance. They are far enough from the herd to preserve a unique and idiosyncratic model of the world. The kind of model which allows them to recognise the unique and idiosyncratic opportunities which produce outsized returns.

Anchoring in Reality

“What important truth do very few people agree with you on?”

Peter Thiel, Zero to One (2014)

All value that an investor can hope to capture will be a consequence of how they model the world. The ability to recognise potential in a company, which may be little more than a few slides at the present moment, is dependent on their ability to simulate the outcome in their mind.

Like all models, volume of data is one component, but the quality of that data is far more important. And even when you have a large amount of quality data, if it’s not proprietary and differentiated it will leave you stuck on par with everyone else. And par in venture capital is a seriously mediocre place.

So, success requires two things:

First, it requires being in the world, interfacing with reality, literally or figuratively. Collecting data that feeds a world model that is uniquely your own. Absorbing information, whether that’s from visiting a factory, analysing a paper, or reading science fiction to understand the infinite possibilities of the future.

But it’s not the simple act of visiting or reading that produces the benefit. What matters is what is recognised in it, plucked out and mentally filed away, and for each person that will be slightly different. Experiences are not fungible.

Second, it requires epistemic hygiene. In a world where everyone is promoting their own world model, you must carefully protect your own. Like the recursive nightmare of training an LLM on the output of other LLMs, quality will be continually degraded as an investor absorbs the opinions of their peers into their model.

Which is not to say that venture capitalists should avoid interacting with each other, but, as the data shows, the best networks are broad, diverse, and arms-length. Productive collaboration that broadens horizons, not incestuous commingling.

“Remember, your goal in investing isn’t to earn average returns; you want to do better than average. Thus, your thinking has to be better than that of others — both more powerful and at a higher level... You must be more right than others… which by definition means your thinking has to be different.”

Howard Marks, The Most Important Thing (2011)

Indeed, returning to the question posed by Thiel; to hold an important truth which few others agree with, you implicitly have a different view on the world — which will have emerged from a unique body of facts and experiences.

This is why research repeatedly shows that venture capital benefits from greater intellectual diversity, even as the forces within the market drive concentration which pulls more capital behind generic, myopic, corporatised models of the world.

It’s why the ceiling on effective allocation, and therefore human progress, has always been the number of good investors, not the volume of capital; the variety of world models that are being applied to seed opportunity in the most unlikely places.

Against common belief, which favours dull uniformity and credentialism, looking for powerful idiosyncrasy in world models may be how the best venture capitalists operate, and how the most effective allocators identify their potential.

"Worldly wisdom teaches that it is better for reputation to fail conventionally than to succeed unconventionally."

John Maynard Keynes, General Theory (1936)

Watch the latest episode of Going Solo, with Sarah Drinkwater of Common Magic

Run your investment firm from your phone, with Odin

Sarah came to venture from the outside. Journalist, club night runner, vintage clothing shop owner, operator, angel. Her journey looks nothing like a straight line. As you often find, that’s kind of the point.

Her fund, Common Magic, backs products with community at their core, building on her experience creating products with community at their core, from music nights to Google Maps.

There are a few things she said that resonated strongly for me:

1. Companies don’t usually die because they run out of money. They die because emotional capital runs out first. Co-founder misalignment, a loss of belief, or something else that’s a drain on vibes. Across 68 investments, Sarah said the pattern is consistent. What I took away from this is something you hear frequently, which is that true passion and a deep “why” behind what you are building are vitally important. I think this is especially valuable to bear in mind if you are investing in the “current thing” during the peak of a hype cycle. The “current thing” attracts tourists, who aren’t always building for the right reasons. When the going gets tough, the energy reserves necessary to keep pushing deplete rapidly unless the belief and core values in the founding team are strong and consistent.
   

2. Somewhat related to (1): the hardest problems in building a company are social, not technical / rational. Organisations are just groups of people. They make decisions emotionally, they are subject to politics. This touches every aspect of entrepreneurship.

   1. Product: especially when you can vibe-code things rapidly, building a deep understanding of the human behaviours and needs for whatever you have in mind is vital. It’s about understanding how people think, and what they want - even if that’s different to what they’re telling you.

   2. Sales: if the champion loves the product but the buyer isn’t there yet, you’re not going to sell anything.

   3. People: if the team are running on fumes, underpaid and overworked (as is always the case in startups), but absolutely aligned on where they’re going, you can achieve a remarkable amount. If they’re well paid, not stretched, but misaligned, you can achieve remarkably little. As Elon puts it, the outcome of any given company is the vector sum of the people within it.
      

3. Every investment is a vote for the world you want to see.
   This is something we really believe strongly at Odin, and is the reason we decided to build this company. More people should be voting with their money on the world they want. For Sarah, this shapes everything from how she sources to how she runs LP coffees and how she thinks about her portfolio’s go-to-market.

We also got into the state of the European venture ecosystem and what she’d tell someone thinking about raising their first fund right now.

Hope you enjoy it!

PR

P.S. Subscribe to listen to future Going Solo episodes via Spotify/Apple Podcasts/YouTube.

One of the worst things that digital technology has done is to disconnect us from writing.

If you think in terms of competing content delivery systems, then the idea of sitting for hours looking at a blank page or a blinking cursor, with an empty brain and a string of words you already don’t like, seems pointless. Inefficient.

This word, “efficiency”, will be the death of us.

Many people who have found success, of one kind or another, say it came down to building a tiny model of the world in their minds. A simplifying exercise where you decide that only two or three things matter, require constant attention, so expend all resources towards that. Anything else encountered, is to be ignored, avoided or attended only in spirit.

We shape our buildings and then they shape us.

The issue arises when people start believing that their model train set, a rough schema of the world, is in fact a shared reality for everyone else. “Where is that train headed, it’s on a loop right?”

Simplified, outdated or unfounded models of other people’s behaviour or motivations get us in trouble. The bigger the model, the harder the fall.

Increasingly we hear spectacular claims about machine intelligence in competition with human intelligence. A nation of PhD-level researchers, an army of agents, at the cost of a single high-level professional, a well-paid job. A sprawling maze of data centres, doing all our homework, drinking up lakes and oceans. An economy of tireless,

dispassionate actors replacing real people first in commerce, and eventually in the role of their lives. Every vision that has been publicly articulated is an acid trip, if a bit two-dimensional and bewilderingly short-sighted.

Examined closely, it reveals quite a reductive conception of intelligence and human consciousness. It posits that intelligence is simply a function of the volume and speed of information that can be processed.

All of life and its ambiguity, reduced to an optimising equation. Scale it. Keep adding compute. Super Intelligence emerges. That’s the prevailing theory that many have bet billions on. Let’s see.

There is a profound difference between data and wisdom, as anyone with a teabag but without boiling water or a cup can tell you. As for consciousness, how can you reconstruct or improve on an elegant structure if you do not truly understand its foundations or the materials it is composed of ? Perhaps this is why some now talk of “growing” rather than “building” machine intelligence.

I recently had the privilege of participating in a half marathon. A grey chubbing bushel of wheat, an upright seal, washed up from the waters of Ireland, competing en France. À Paris!

Totally brained it.

I stayed up till 3 the night before creating a playlist, lacked hydration, forgot suncream. The second I saw people amassed in their hundreds, all pointed towards a positive end, in the Paris sunlight, some long-forgotten chemical diffused into my brain, informing me that I was Usain Bolt.

I discarded 3 months of gentle, incremental pace training from a couch potato position. Two weeks out, we ventured from Milltown to Sandymount strand, a first 15k run. I sported a slow, swaggering, Mr Burns gait with accompanying grunts and feeble whistles. I just about made it home that evening.

The adrenaline will carry you on the day, they say.

I gassed my fat ass sprinting at the speed of light for one minute. I hadn’t experienced that level of exciting group exercise or positive community in aeons. I came to, only to realise that I was entirely depleted. There was 20k to go.

Hell.

We are social animals. As my rasher slowly fried, I scanned the flock to identify similar, ready-made boil in the bag zombies. Discards from the assembly line. Delirium. Then you enter a kind of meditation.

Every 5 kilometres, I’d approach a set of tables, a brass band, cheering crowds, without adequate refuelling gear.

Parched, burnt and blinded. An ominous line of volunteers would direct a hose at my face and mouth area. Then spray intensely for a few seconds. Like I was a horse. Or a donkey.

When your resolve plummeted, you’d catch sight of a father running in honour of a lost child, a smiling cherub emblazoned on his t-shirt, someone wheeling a partner, or people just struggling who were carrying more wheat than me. As we dipped and panted below bridges, you’d constantly track people in your wheelhouse.

I won’t give up. If they don’t give up.

There were twins. One was a serial runner. The other was on 3% battery, with a long way to go.

She staggered erratically, like a wounded deer, stunned from a gunshot.

I was moving so slowly that they could perform an opera around me. They would run vigorously for a stretch, the athletic sister half carrying the other. Then they’d overtake me. I’d pass them a few moments later sitting by the side of the road, imploring, crying, motivating. Drinking water.

Me shuffling past, a discarded wine bottle bobbing up the Seine. Then they would rise, regain inspiration and surge past. At the finishing line, with no discussion, we all embraced. Then parted.

As much as I am not a runner, there is something spiritual in that process.

The last stretch is disorientating. Early finishers start running backwards. You meet streets without barriers where ordinary people are just getting on with their lives. Bemused eyes look on from bars. Emotional thoughts rise to the surface. You picture family members, forgotten friends and acknowledge unresolved sadness.

Your body keeps the score, so a physical ordeal after a great slumber, exhumes some of life’s ghosts. But your systems are in critical. Just don’t stop moving. Lights flashing. “Snake!”

The feeling upon finishing and embracing my other half. The post-race meal, fully deserved, delivered slowly, savoured between bloated, cathartic chats. Mission accomplished, I can see why they do it.

We are still capable of coming together and coordinating 50,000 strangers from anywhere, towards a clear goal, for purely positive incentives. Remember that.

Clips have appeared recently of a marathon in China where the people run alongside robots. Can you think of anything more demoralising ?

I’d say, if you are manufactured in an advanced factory, and composed of metal, yet can’t run faster than me, then let’s scrap the entire technological project. In these rare uplifting moments of shared human perseverance, solidarity, that people still enjoy publicly, why is it framed against the backdrop of the evolutionary chain or technological progress ?

You’ve battled against the many seductive, destructive forces that ensnare our minds and bodies, to do something physically healthy, only to be passed by a dispassionate little tin man whizzing along without a soul, coldly repeating something like “stand clear, luggage doors operating”.

There are myriad future worlds we could live in, with varying degrees of technological entanglement and oversight.

Don’t let people convince you that there is just one outcome, or that it is inevitable, where the mass of men live a hollow, haunted existence, severed from nature and one another, strangers in their own skin, while an anointed few live insulated from it all, as petulant, demonic kings.

Ronan Hayes is an entrepreneur and writer with a background in psychology, additive manufacturing and sustainable enterprise. He focuses on the influence of technology on civilisation, group behaviour and the way in which our societies seem to be unravelling. His work strives to make you reflect deeply on a topic and occasionally snort with laughter.

Deploy capital to vote on the future you want, with Odin

The truth, as usual, is somewhere in the middle; Marc may be right about David, and David may be right about Marc.

This relates to the divergence into large generalists and small specialists, and what AI means for these two sides of venture capital.

An LLM can be viewed as a library of human achievement. Everything that we have theorised, discovered, calculated and ultimately recorded. What we know about life on Earth, and the little we know about what lies beyond.

Venture capital exists to fund the expansion of that frontier. Never looking backwards, always looking forwards for the next opportunity.

Through this lens, AI and venture capital should exist in harmony.

As VCs take pitches about anything from asteroid mining to plutonium batteries, AI is the expert partner that costs $20 a month and doesn’t disappear for ski trips or try to fire founders. Agents can break cold inbound down into the most practical format, handle reporting for LPs, synthesize investment memos and package market research.

In many ways, it’s the perfect teammate.

“Individuals working with AI showed a substantial 0.37 standard deviation performance increase over the baseline of working alone without AI. This finding suggests that AI can effectively substitute for certain collaborative functions, acting as a genuine teammate by granting individuals access to the varied expertise and perspectives traditionally provided by team members.”

The Cybernetic Teammate: A Field Experiment on Generative AI Reshaping Teamwork and Expertise

So, the era of AI is also likely to be the era of the high-agency capitalist. It will empower an army of individuals and small firms.

This is a stark contrast to the development of “mega-funds”, who have consuming a growing share of total allocation in recent years. But a “barbell distribution” is supposed to have two equal ends; there should be a small number of very large firms, and a large number of very small firms.

Sixteen years after Seth Levine first wrote about the bifurcation of venture capital, the market may finally reach a more productive equilibrium.

Flight from the Centre

To understand this notion of “barbelling”, consider the framing in Oliver Williamson’s seminal work, Markets and Hierarchies, from 1975.

• In environments where transaction costs are high, with more focus on diligence, monitoring and execution risk, hierarchies are the natural solution.

• In environments where transaction costs are low, with more focus on speed, optionality and diversity with smaller investments, broader markets emerge.

Essentially, the larger multi-stage platform end of the venture market has drifted further towards centralised and hierarchical firms that service larger pools of capital and scaled capital needs.

Equally, the smaller early-stage segment should drift in the opposite direction, towards a decentralised market of many independent actors, better suited to engage with early-stage founders.

This process is easier to observe in mature industries that are further along in their transformation. There we can see how bifurcation is not only reflected in scale, but also in economics and incentives that reflect two very different products and end-users.

Consider asset management:

• On one end of the spectrum are a handful of mega-managers like BlackRock and Vanguard. Here, fees have been driven down to 0.03–0.10% by scale and systematisation.

• On the other end of the spectrum is a variety of smaller active asset managers, like Capital Group or Baron. Here, fees are in the 0.5–2%+ range, reflecting performance.

Essentially, as scale erodes returns, fees must come down. Otherwise, large funds would produce a huge fee drag, which wouldn’t be appealing to investors. So, scaled firms use systematic and organisational efficiency to bring down costs and compete on cost.

In other major fund strategies, management fees have fallen towards the marginal cost of providing the product thanks to competition and scale. As funds have expanded, firms have been able to compress their fee % to remain competitive. Venture capital is the only strategy where fees have increased, even as funds have gotten larger.

This is where venture capital has been stuck, as the amount of capital the large firms managed has grown by an order of magnitude, yet they maintain the same boutique manager fee structure. It’s an unwieldy one-ended barbell at the moment.

The result has been underperformance, stagnation, and a weaker product for both founders and LPs, because venture returns do not scale.

In order to correct this behaviour, the firms and economics at both ends of this spectrum must be well-understood by LPs.

Unreasonable People

“The reasonable man adapts himself to the world: the unreasonable one persists in trying to adapt the world to himself. Therefore all progress depends on the unreasonable man.”

Man and Superman, by George Bernard Shaw

Consider that boutique venture capital is fundamentally a hunt for the purest source of alpha; outlier founders and companies, the weirder the better.

Implicitly, it’s a deeply insecure strategy. The best investments are non-consensus, with no peer validation. Success is revealed on distant horizons, incremental metrics are mostly useless. Investors never really know how well they are doing at any given moment.

The result of this insecurity is herd behaviour. Because performance across the industry is understood on a relative basis, there is an incentive to do the same thing as everyone else, even when participants recognise it’s likely to be wrong. So, venture firms become enmeshed with each other, literally and ideologically, and alpha decay pulls down returns.

The survivors are “unreasonable” individuals, by definition. Angel investors or solo GPs with conviction in a perspective that precludes herd behaviour. Their idiosyncrasy is a shield, and their eccentricity prevents them from ever fitting into the group too comfortably. As a result, they tend to struggle in larger firms with more bureaucracy and louder peer effects.

This archetype is the inversion of an LLM. They are outliers. They spend their time thinking about what could be, rather than what is. What people should be doing, rather than what they are doing. They may or may not study history, but they certainly have strong opinions on the future.

Their success is reliant on recognising other outliers, so origination and selection are vital skills which amplify the idiosyncrasies of the individual. These investors provide the critical interface between capital and talent, where potential is least obvious. It’s inherently an unscalable human job, best achieved by a larger field of independent actors.

So, these investors are best positioned to benefit from an AI teammate that handles everything other than the interface. That benefit, in theory, is that AI allows these investors to take bigger swings and spend more time on investing, producing even stronger investment performance.

“The future isn’t written in training data. Early-stage investing lives in what hasn’t happened yet. Humans are still the best at this and I can’t imagine AI being better on any alarmist timeline.”

Cyan Banister

For the best investors in this category, there is room to increase fee income without a meaningful drag on performance. Indeed, by making life with a smaller fund much more manageable, it incentivises raising right-sized vehicles that can deliver attractive multiples. It also allows for increased compensation without shifting strategies in a way that erodes returns.

The Firms

To the extent that an AI can predict the future, it will do so by extrapolating the past; assuming a continuation of the status quo. Ask an AI how to deploy capital in the stock market and it will recommend buying the index because markets are mostly efficient and everything knowable, everything in the model, is already priced in.

Definitionally, this is a beta-generating strategy that supports today’s incumbents. There is no variance, positive or negative. For almost everyone, almost all of the time, it’s the right thing to do.

Apply the same logic to venture capital, and the result is something like the mega-fund platforms. Their strategy is to index market themes, reduce the brand-related risk of missing “obvious winners” and juice fund metrics with momentum. Implicitly, they are incentivised to support a stable status quo that produces clear winners and predictable outcomes, rather than messy disruption.

“Smart Beta are large multi-stage platforms that “index” across legible signals so that they may plow capital into companies that break out. We define legible companies as those with recognizable signals (i.e., well-formed businesses, founder spinning out of Big Tech, repeat founders, meaningful product/commercial traction, etc.).”

Smart Alpha vs Beta in VC

This strategy inherently sacrifices outliers; the startups building novel solutions in obscure categories. It’s an acceptable compromise, as the bigger firms can always pay up for access later as the opportunity becomes more legible.

Here, AI can create huge gains in efficiency by informing screening and investment decisions according to a clearer set of target parameters. The benefit is not better performing investments but rather lower costs, and therefore lower fees and stronger net performance.

Indeed, early experiments with applying LLMs to venture capital deal selection do appear to improve the overall rate of success. It comes with the same sacrifice of outliers, as the model’s inherent biases in the training data, but this could be reduced over time and isn’t a deal breaker for the scaled strategy.

“We evaluated nine state-of-the-art LLMs and found that several outperform not only the market index but also the leading VC firms, with GPT-4o achieving the highest F0.5 score. These results show that anonymized founder profiles are sufficient to surpass human-expert baselines in early-stage venture forecasting.”

VCBench: Benchmarking LLMs in Venture Capital

A similar result comes from studies looking at how AI can be used to categorise and filter opportunities. The LLM-based approach was 537 times faster than the human analyst without any loss in quality. Again, outliers will be lost in the margins of this process, but that is already accounted for in the scaled venture strategy.

Better Economics

“The fee drag on gross-to-net returns is estimated at 0.1x to 0.7x, equivalent to an annualized impact of 5% to 8% on returns. This impact is notably higher for buyout funds (7.9%) and venture capital funds (8.5%) than previously documented. […] This drag raises concerns that high fees may significantly erode investors’ returns, potentially outweighing the benefits of investing in private capital markets.”

Accessing Private Markets: What Does It Cost?

In practice, a reduction of management fees to ~0.5–1% would significantly lift the performance of this part of the market, making it even easier to draw in LP capital, pulling down the cost of capital that gets passed on to the founders and their companies. It would also resolve some of the incentive-related issues at the larger end of the market.

Today this cost of capital often becomes the killer over time, as it jacks up valuations beyond any reasonable appetite for exit markets.

Essentially,

1. a high fee-drag means LPs expect significant multiples on their capital which often aren’t possible with a large-fund...

2. so, capital increasingly concentrates into inflated “winners” in an attempt to hit an unreasonable target returns…

3. and the outcome is the weak performance of companies post-exit, the favouring of consensus, the stagnation of technology, etc.

It is rational that scaled firms should operate with reduced management fees compared to smaller funds. This makes sense both in terms of expected returns and in terms of fee income relative to workload. As described above, it’s also true of every other fund strategy.

Correcting this holdover of previous eras would ensure a healthier exit market for venture investments, stronger performance for LPs and a better product for founders at both ends of the spectrum.

A Wedge for Innovation

“Ironically, innovations in venture capital haven’t kept pace with the companies we serve. […] As chips shrank and software flew to the cloud, venture capital kept operating on the business equivalent of floppy disks.”

Roelof Botha

For a long time, and for reasons related to everything fromfund structure to economics, it has been clear that venture capital is bizarrely behind the times. As an industry that supposedly obsesses over innovation, it clearly doesn’t spend nearly enough time on introspection and has become top-heavy and stagnant.

As the market has changed since the turn of the century, few people have asked what this means for the “2 and 20” structure, which dates back to Alfred Winslow Jones in 1949.

That lack of critical thinking about compensation is a red flag for market capture; conditions that incumbents have clearly grown comfortable with. Indeed, new GPs remain reluctant to innovate on fund strategy or economics for fear of scaring off risk-averse LPs, having been warned it would induce adverse selection.

“Consistent with the literature on institutional behaviour, we find that the continuous use of boilerplate provisions is a form of ‘collective conservatism’. We argue that these results help explain why venture capitalists typically maintain using suboptimal boilerplate provisions. Venture capitalists may be hesitant to abandon these provisions out of concern that their peers and major rivals prefer to use them in their partnership agreements. This suggests that the often ineffective ‘2 and 20’ rule may persist for this reason.”

Conservatism and Innovation in Venture Capital Contracting

There is a clear argument that AI breaks this status quo.

For scaled, systematic allocation there are obvious gains to be made in efficiency and productivity, which would allow for reduced fees and a lower cost of capital for entrepreneurs. This would help keep the late-stage market more rational, and preserve the viability of paths to exit and liquidity for LPs.

For small firms and solo investors, there is reason to feel more bullish than ever. Adding a cybernetic teammate to their arsenal will help to scale their impact on the world. An expanded appetite for moonshots from much larger group of investors.

Watch the latest episode of Going Solo, with Adam Besvinick of Looking Glass Capital

Run your investment firm from your phone, with Odin

This has remained broadly true despite the venture industry growing almost 20x over the last two decades. This expansion has appropriated growth from public markets for a similar number of outcomes, rather than increasing the volume or scope of innovation.

So, how can the venture capital industry actually find and fund more important companies each year, without diluting the overall pool of potential?

As it happens, the friction is a feature.

Scaling Innovation

Imagine you have just been put in charge of your country’s national innovation strategy, with the goal of expanding the technology sector.

You can measure your success along a number of dimensions:

1. Number of startups created

2. Volume of venture capital available

3. Number of startup exits

4. Total market cap of new technology IPOs

Each one of these metrics, in isolation, creates perverse incentives.

• Focusing on the number of startups may raise the number of marginal companies created each year, raising the net cost of failure.

• Looking to raise venture capital volume will produce sloppy allocation and weaker outcomes in the long run.

• Both number of exits and market cap produce incentives to encourage either faster exits or inflated exits, both of which are a drag in the long term.

Inevitably, you need to take a full systems thinking approach that can proxy the behaviour of a healthy and competitive market.

“Strong interrelationships among the ecosystem elements also reveals their interdependence and need for a systems perspective.”

Entrepreneurial ecosystem elements, by Erik Stam and Andrew van de Ven

Essentially, your goal should be to raise the total market cap of new technology companies → by increasing the number of good exits → which will require more venture capital → which is downstream from the volume of good opportunities.

The key word here is “good”, reflecting the unavoidable nature of qualitative assessment to ensure that marginal companies and poor investment decisions are consistently screened out of the process.

Multidimensional Markets

Venture capital faces the same problem; there are no easy answers to scale if your goal is driving progress rather than expanding fees. Research has consistently illustrated the distortion created when one theoretically positive metric outpaces others.

Lowering the bar for startup formation to increase the number of opportunities does not inherently increase the number of quality companies.

“Policies that reduce frictions to entry can broaden participation and encourage experimentation, but do not automatically generate high-quality outcomes. Achieving top-tier performance appears to continue to depend on the recombination of diverse human knowledge and experience.”

Generative AI Fuels Solo Entrepreneurship, but Teams Still Lead at the Top, by Hyunso Kim, Hyo Kang and Jaeyong Song

A glut of capital tends not to result in greater levels of valuable innovation, but rather more fuel shovelled into the molten core of the market.

“Increases in venture fund-raising which are driven by factors such as shifts in capital gains tax rates appear more likely to lead to more intense competition for transactions within an existing set of technologies than to greater diversity in the types of companies funded.”

Short-Term America Revisited? Boom and Bust in the Venture Capital Industry and the Impact on Innovation, by Josh Lerner and Paul A. Gompers

A surge of exits often doesn’t end up producing enduring value, and may actually represent a distressed environment dumping toxic waste.

“The IPO market has significant time variations over time and IPOs issued during the high-volume period are more likely to have worse performance on average than those issued during other periods.”

Essays On Ipo Cycles And Windows Of Opportunity, by Meng Chen

As an illustrative parallel, research makes the case that startups must scale appropriately along all dimensions to ensure long-term success. It appears that the same is true for the markets they exist in; poorly-balanced scale creates inefficiency that may be exploited by opportunists.

This points to one of the many paradoxes of venture capital; the desire to increase the surface area for opportunity while focusing intensely on the few outliers.

Incumbent venture capitalists and LPs will often warn against expanding the market by introducing more capital or supporting the entry of new firms, on the basis that it simply dilutes the limited set of outcomes.

However, it’s inevitable that great companies are missed each year because the venture market lacked coverage, conviction or originality. Most of venture capital’s great outcomes involve stories of struggling to find willing investors in the early years.

Indeed, as more emerging managers continue to prove their ability to find outliers, it’s harder to make the case that alpha in private markets is running dry.

So, how might these two opposing ideas be reconciled?

Building the Base

One of the findings of research into venture industrial strategy is that there may be easy gains if a government takes on the role of LP to support the entry of emerging managers who invest at the earliest stages.

Institutional LPs are burdened by career-related insecurity, and so bias towards larger established firms that are typically less focused on early-stage companies. This persists despite clear evidence of emerging manager outperformance.

Additional fee-related incentives drive many successful early-stage managers to grow their funds and move to later-stage investing with larger rounds and lower career risk.

As a result, the early-stage foundation of venture capital is often undercapitalised and under-resourced. At the point where it seems most important that investors are prolific, experienced and patient, there is often a shortage of all three.

The interface between capital and talent must be broad. It should cover every corner of the economy and penetrate every sector, in reach of anyone with entrepreneurial ambitions. Fundamentally, this is why early-stage VC should be the domain of many solo GPs or small partnerships who each operate in a variety of ecosystems.

However, as a result of today’s market structure, an increasing amount of early-stage capital is held by a declining number of larger firms. Additionally, each of these firms has narrowed their focus on certain consensus features in order to increase capital velocity. This is entirely opposite to the direction that the market should be moving; the resulting interface is weaker, with worse coverage. It is built on the self-serving idea that founders should pursue investors, rather than vice-versa.

One way to test the benefits of a wider distribution is to analyse years where a larger than usual share of venture capital went to first-time funds, and see if that correlates with increased exit value in the relevant ‘age at exit’ adjusted year.

Essentially, are emerging managers a real driver of future value, through their ability to find a greater volume of generational companies at the earliest stages?

If you exclude the single outlier year of 2021 (where market conditions encouraged a wave of exits), there is a clear correlation between these two factors; a greater number of first-time funds appears to have a positive association with subsequent exit value.

Despite this, research continues to suggest that expansion in the venture market continues to disproportionately favour the already well-capitalised incumbent firms.

Allocators are sacrificing long-term outcomes and technological progress for nearer-term confidence in their decisions and limiting their perceived career risk.

Scaling Conviction

The common resistance to a market with many small firms is the resulting coordination frictions for more capital-intensive companies.

Once again, there is nuance in the market and how it influences investment behaviour.

In hot markets, venture capitalists will often complain about fighting tooth and nail to get allocation. While this sounds desirable from the perspective of coordinating capital, it’s actually reflective of gatekeeping, cronyism and narrower interest in fewer companies that will not necessarily provide better outcomes.

If you talk to venture capitalists in cooler markets, or those who deliberately avoid hotter parts of the market, they tend to have larger networks of coinvestors and more productive relationships. This aligns with research that shows a broader network is more beneficial than a deeper network.

Indeed, analysis by has found that smaller venture firms have more connections per employee than larger firms, meaning that the greater capital coordination frictions may be less of an impedance.

Research also suggests that there isn’t a huge amount of organisational capital in venture capital, with most value residing in the qualities of individual partners. This has been viewed as an inherent restriction on the size of VC firms, as larger firms dilute the influence of individual partner ability.

“Our findings are consistent with the idea that the organizational capital inside a venture capital firm is limited, in turn limiting firm size. Given this, it appears that brand, process, deal flow, etc. are not critical firm level characteristics, but rather they are partner attributes.”

Is a VC Partnership Greater than the Sum of Its Partners?, by Michael Ewens and Matthew Rhodes-Kropf

Essentially, the argument is that an early-stage market with more individual investors or small partnerships would be significantly more efficient at applying that valuable human capital to find and fund outliers.

The same paper also suggests that larger firms have been formed as a means to reduce their own fundraising friction, rather than in pursuit of stronger performance. In essence, the flawed interface between GPs and LPs is chiefly responsible for the agglomeration that has eroded returns.

“Overall, the results suggest that venture capital partners are often significantly different from each other, but high quality firms are those with a group of better partners. What then is a VC firm? We suspect that the benefits of joining together may relate to fundraising.”

Is a VC Partnership Greater than the Sum of Its Partners?, by Michael Ewens and Matthew Rhodes-Kropf

For growth investment, this dynamic changes. Traditionally, as companies develop more technical proof or traction, they become more legible to investors managing larger pools of capital who are inherently more risk-averse. This is a different class of venture investing, which can operate at a larger scale and with different economics.

The graduation from early-stage to growth capital is a milestone in the maturity of a company, which is later followed by a similar graduation from growth capital to public markets as companies develop strength and become more trusted and predictable. At each step there is pricing tension that keeps the market honest.

However, that chain of events doesn’t always play out so rationally.

The Crooked Pyramid

Disrupting this tidy market logic is the rise of scaled multistage capital.

The first problem is that multistage investors get to mark their own homework. There is effectively no mechanism to keep pricing rational if the same firm can keep leading rounds for as long as an investment remains viable.

“Insider pricing rounds, when financing is led by existing investors, creates a natural conflict of interest; as many times, it’s in the insiders’ best interest to inflate the values. The inflated values allow the investors to show paper gains to their LP’s on their existing investment, which makes their LP’s happy, thereby making the GP happy. This also makes the founders happy as it’s generally easier to raise money from existing investors, and most entrepreneurs feel a higher valuation is better.”

The Growing Ponzi Scheme In Venture Capital, by Jay Levy of Zelkova Ventures

Secondly, and more relevant to the question of scaling the venture market, is the multistage firm’s ability to opportunistically allocate across any stage. Imagine a manufacturer in the middle of the hardware value chain adjusting component volumes based on market vibes rather than predictable order volume.

• If a multistage GP feels like the best fundraising story is Seed capital for an emerging wave of AI startups, that’s what they will do.

• If a multistage GP feels like the best fundraising story is a growth fund for a more legible set of market winners, that’s what they’ll do.

As a result, the market for capital at certain stages will occasionally expand or contract, leading to the creation or sustainment more marginal companies, or the death of otherwise deserving companies.

All of this creates a more lumpy market with more friction for founders and worse returns for LPs, because those stories are not reality.

In a perfectly rational world, the venture market is formed like a pyramid.

• At the base, there are many small firms that pursue diverse idiosyncratic risk by operating on the fringes and investing in the most innovative ideas.

• At the top, there’s a small group of more cautious managers with larger pools of capital, who want to help mature businesses transition to public life.

• Between the two, the base narrows evenly and proportionally up to the top through intermediate capital providers. It is able to do so efficiently because success at each level creates opportunity for the next.

As a result, there is always a healthy interface layer, and there is always capital to help the best of those companies graduate up through the stages towards an exit.

Unfortunately, venture capital is neither efficient nor competitive. It is needlessly opaque and overwhelmingly driven by shallow signals and cyclical herd behaviour.

So, exits come in waves, with painfully long waits in between, as investors cling to narrative moments that may be exploited for LP attention.

Platform Distortion

Outside of the market structure, there’s a question of the startups within it and how they are shaped by the different categories of investor.

• In a market with a larger number of emerging managers, solo GPs and small partnerships, there is likely to be a wider array of more novel and interesting startups in the world.

• In a market with fewer, larger firms there is likely to be a much narrower and more internally competitive set of startups that will be centrally located and consensus-themed.

The former will graduate up through the layers of the market more slowly, as novel technology adds risk which may require more patient capital. The latter can move up much more quickly, although that’s largely a function of capital appetite and hype rather than business maturity or fundamentals.

On top of that, the scaled platform firms offer a range of support functions which further warp the outcome by essentially making venture capital less Darwinian.

Consider two rival propositions:

• A deeptech company in some frontier category that has just received seed funding from a $5M early-stage solo GP. Each subsequent round will be a battle as the company has to prove itself to each new investor, although the list of GPs in its corner will grow.

• A SoMa-based AI startup that has been “kingmade” by a platform firm that will provide them with network connections, support for core functions, social media astroturfing, media connections, a “tier 1” fund brand and the promise of downstream capital.

Which of those two environments is more meritocratic?

Which is likely to ensure capital is consistently allocated to the most important and deserving companies over time?

To repeat a consistent theme in all honest venture market analysis, alpha is implicitly found in the hardest places and least popular ideas. For venture capital to scale effectively, it must do so without losing its Darwinian nature.

It may become larger and more accessible, but the bar must not be lowered.

The Case for Scaling Venture Capital

This all makes the case that scaled platform firms (vertical scaling) are not the answer to expanding the opportunity of venture capital, as they inherently erode the process that incrementally selects for great outcomes.

A more ideal approach would be to scale the number of firms (horizontal scaling) proportionally across all levels. If you want more great exits, you need a greater number of investors at the initial interface layer, and up through each layer above.

Historically, this interface layer has been limited by information friction with LPs, incentive issues related to fee income, and cyclical markets that systematically wash out younger firms rather than the weakest performers. In order to scale venture capital properly, to accelerate the development of technology, these constraints must be addressed through innovation to the venture model itself.

It’s easy to see a future where AI significantly reduces the reporting and compliance frictions between LPs and GPs. In a world with agents, it shouldn’t be significantly more difficult to manage 20 smaller fund commitments rather than 1 larger fund commitment. At some relatively trivial breakpoint the benefit to returns will pay for whatever token burn is required, and it’s a workflow LPs will likely adopt anyway.

The fee incentives in venture capital are plainly broken, protected only by the market capture of the largest firms. It’s reminiscent of how often you’ll hear “top quartile” as the benchmark, when top quartile funds haven’t produced compelling performance since the mid-90s. The industry has changed radically over the last 25 years, and the compensation economics (along with much else) must evolve as well.

The excessive cyclicality of private markets is likely self-correcting if the underlying information frictions and incentives are taken care of. Indeed, many of the problems of venture capital today may clear themselves up with a few structural fixes.

Ultimately, the friction inherent in coordinating capital across subsequent rounds is a feature of private markets, not a bug. It is the mechanism by which marginal companies are systematically and iteratively screened out of the market.

Which is not to say that the market otherwise functions perfectly today, but the remaining issues cannot be properly diagnosed if they are masked by multistage capital.

“While the law [of competition] may be sometimes hard for the individual, it is best for the race, because it ensures the survival of the fittest in every department.”

The Gospel of Wealth, by Andrew Carnegie

Run your investment firm from your phone, with Odin

As a result, a growing share of the expansion is siphoned out as management fees.

But market expansion also pushes startup exits further into the future. Increasingly delayed exits drive venture capital’s appetite for capital to support larger, later rounds.

So, exits continue to slow, the venture market snowballs, and the fees keep growing.

To begin, consider the hypothetical journey of a successful venture-backed company from a pre-ZIRP era:

In this scenario, a startup is formed around some novel proposition that disrupts a market and promises outsized returns if it works. There’s obviously extreme uncertainty, so raising capital is challenging and dilutive.

After bootstrapping for as long as possible, perhaps raising from angels or friends and family, the founders approach a venture capital firm to lead their Seed round. A syndicate of other investors is assembled, and over the next few years the company is carried through subsequent Series A and B rounds as it continues shedding risk.

For their larger Series C, the company brings in a venture growth firm. This capital helps the company expand into a couple of new markets and build out the required compliance functions to go through an IPO. Their new capital partner understands public market dynamics, and acts as a steward in that journey.

In year 8, the company goes public at a valuation of $1.2B. It’s comfortably a fund-returning outcome for the early investors, and a solid multiple on an excellent timescale for growth equity. All within a decade, as designed.

Most importantly, the company has healthy economics and will be well received by public market investors. Access to deeper capital markets helps to fund compounding growth and innovation. Over time, the company may earn access to the major index funds and become a truly generational outcome, especially for those who held onto their shares through the IPO. There is no ceiling on future growth.

However, this scenario only justifies around $10M in total management fees, across all funding rounds and the various firms involved.

In 2010, as the interest-rate crash settled into the market, venture capitalists got a taste of a bright new future. The growth of capital inflows would change the industry forever.

Over the next decade, a new form of venture capital emerged which can be referred to as “selling allocation at scale”.

A common trajectory for successful venture-backed companies in the modern era:

A startup is formed around an innovative proposition that fits some current theme of opportunity, maximising investor appeal. The founders are well credentialled, centrally located in SF, and quickly scouted by a large venture platform.

Rather than the modest Seed round the founders target, the price-insensitive investor cranks it up by a multiple to signal their conviction and discourage competition; they are a “kingmaker”. In return for this generosity, the company promises future growth at an incredible pace.

To hit the required metrics, the substantial balance sheet is used to heavily subsidise unit economics and accelerate growth. Advertising budgets climb, burn rate surges, but all that matters is that revenue keeps accelerating.

Underpinning this is the modern VC’s valuation tool of choice; the revenue multiple. If a fast growing tech startup is valued at 20x revenue, it can spend $1 on winning 50c of revenue and still print $10 of incremental value for every $1 invested.

Downstream of this, startups will fail more often as they prioritise subsidised growth over building stronger economics or creating more value. Failure is compressed into the early years, allowing capital to concentrate faster. The survivors will be those that hit the growth metrics to attract further investment, filtering for companies that are quick to revenue, easy to scale and “legible to capital”, rather than any deeper quality.

As a result, venture platforms have developed a huge appetite for inexpensive early stage “call options”, although they are explicitly looking for scalability and obvious investment appeal over the typical outlier profile of frontier technologies and ideas.

This dynamic solves one problem, while creating another:

On the one hand, it allows the venture platform to show strong incremental metrics to their own investors. Short-term IRR becomes more predictable, and they’ll be known for their investments in the fastest growing companies. This checks all of the boxes for allocators at the largest LPs who optimise for short-term career objectives.

On the other hand, the resulting companies are cash-incinerators and fundamentally incapable of reaching an exit. Their economics are so deeply negative that acquirers won’t want to pick them up (except perhaps a “reverse-acquihire” talent raid), and the transparency required by public markets will only expose their weakness.

However, even here there is opportunity. If you have a fast-growing company that could continue absorbing allocation and printing paper growth, why exit?

If you can raise a $400M fund and earn $8M in fees each year investing in Series A companies, why not raise a $2B fund and earn $40M a year by scaling private companies further and investing in Series E, F and G+ rounds?

So, in subsequent years, the investor makes further investments in the company, putting more capital to work at terms set within that original syndicate. First tens of millions, and then hundreds of millions, at spurious valuations.

While these investments initially come from smaller early-stage funds, later they’ll come from larger growth vehicles. This allows the venture platform to keep allocating to the company from new funds that are within the 10(+2) year fee window. It even allows early investments to be cashed out at a premium in later rounds, if desired.

This behaviour exploits the opacity of private markets, where capital provides the most obvious signal. When it’s difficult to see how well a company is actually doing, investors will look to the behaviour of their peers. This gravitational effect encourages accumulation while destroying alpha and beginning to resemble a Ponzi scheme.

In this modern scenario, a hypothetical company that gets held for 14 years in private markets could easily absorb $1B from venture capitalists across subsequent rounds. This is capital that previously would have been pulled from deeper public markets, where the average fee is closer to 0.5%, rather than 2%.

Horowitz’ Law: “Capital flows to wherever the fees are highest.”

Now, in this scenario this one company provides the basis for an impressive $250M of fee income, rather than $10M from the previous scenario.

So, what about exits?

To the limited extent that exits still matter to the scaled venture platforms, they are somewhat achievable. Take an “exciting” company with hype and revenue at a huge scale and ship it like a meme stock when the market feels amenable. Most of the time, it will sink like a stone at lock-up expiry, but everyone will have celebrated the day-one pop and will lose interest in subsequent months anyway.

The later-stage funds that catch more downside from this behavior may be stapled to the more appealing early-stage funds, so LPs don’t necessarily have a choice but to participate in both. One offers the inflated metrics and occasional liquidity, the other solves for the allocation at scale. It’s a reasonable solution for careerist allocators who don’t plan to stick around to see the rotten fruits of their labour anyway.

It seems intuitive that the companies which experienced the greatest enthusiasm in private markets would continue to be strong performers later in their lives. Unfortunately, this isn’t the case. After all, they are selected on their ability to scale rapidly, not on any deeper quality or sustainable competitive advantage.

Indeed, of the 12 companies that raised more than $3B in private markets, only Robinhood has a (slight) positive return versus the S&P 500 today. Only Lucid and Airbnb were positive at lock-up expiry, though neither are today.

Overall, the returns from this group of venture capital’s giants are miserable. A portfolio based on the theory that raising more $3B+ from venture capital was a positive signal would be sitting at –117% today relative to the index.

This outcome is aligned with a large body of research on the interface between venture capital and public markets. Venture-funded growth, beyond a certain point, becomes a major negative indicator for post-IPO prospects. This finding is well grounded in data and financial theory, but the simple explanation is that public markets want strong companies while private markets want big ones.

Despite that, the venture industry keeps growing larger, and pumping more capital into the hottest companies.

In the end, the driver of all of this is that generating fee income is a much better business than generating carry. Scaling allocation is a better business than shaping good exits. Firing money into consensus is a better business than finding outliers.

In 2005, the five venture firms that raised the most money in that year each generated around $150M in subsequent fee income. For 2025, it’s closer to $1.6B. Which would you prefer?

A company-level case study of this phenomenon can be found in the example of Rivian versus Tesla. Both consumer EV startups from similar eras, who emerged in different fundraising environments, with investors that had fundamentally different motivations.

Tesla might have had some brief headwinds from the cleantech wave in the early 2000s, but on the whole the company faced a challenging time raising capital. As a result, it raised minimally, found great efficiency, went public relatively quickly, and has gone on to incredible success in the years since.

Rivian was a beneficiary of the low rates environment, and absorbed huge amounts of private capital prior to its IPO in 2021. It never felt the same capital constraint as Tesla, and had no need to develop the same capital efficiency. This has contributed to Rivian’s struggle and a share price that is now down by more than 80%.

Indeed, Rivian raised $10.5B in private capital prior to IPO, compared to Tesla’s $200M. As a vessel for allocation, it supported a far greater volume of fee payments, despite a clearly weaker outcome.

The development of venture capital as a business of selling allocation, scaling the 2% rather than the 20%, is responsible for a number of unintuitive or seemingly contradictory phenomena in the venture market:

• The way in which venture investors have increasingly embraced “consensus” in recent years, seeking to lower fundraising friction instead of generating alpha.

• The institutional hostility toward antitrust regulation, despite evidence that antitrust is beneficial for nascent startups, innovation and early-stage returns.

• The shift of the venture capitalists’ role from one of monitoring and stewardship, to one of choosing to overlook or even incentivise questionable behaviour.

• The invention of functions like platform teams or media arms, designed to support the idea that venture firms have advantages that accrue to scale, which is not reflected in returns.

And the greatest paradox of all, that while the venture capital market has scaled, it has increasingly failed on the two fronts where it should be leading; generating returns for limited partners and driving technological innovation. The era of the greatest abundance, from 2016 to 2021, produced a bonfire of capital in the name of a glut of enterprise software and crypto slop.

Ultimately, this is a failure of LPs who have allowed capital to be assigned to something labelled “venture capital” that is actually just the beta of private market hype cycles. This product is designed to serve the myopic career goals of mercenary capital allocators and the continued bloat of venture platforms.

It has been a radical failure for everybody else. The result is a market with weaker returns that is less tolerant of original ideas. A deadzone for alpha.

A Handful of Companies

The common refrain is that today’s winners are bigger, and the rewards more concentrated into fewer firms, but the evidence is clear that these giant outcomes are not positive in the long run.

• The most capacious vessels for venture allocation generally do not continue to grow at a healthy rate after they exit, nor do they continue producing meaningful innovation or other positive externalities.

• The screening process which selects for these rapidly scalable companies is crude and unidimensional. It is not well designed for catching the typical outliers which drive attractive venture returns.

The Magnificent 7, which remain the most important companies in the world today, were almost all reared in capital-constrained environments that required relatively early IPOs (a median of 5.9 years). Clearly, this did not limit their future growth. In fact, those constraints helped forge truly durable, powerful businesses.

Where would AI be today without foundational work done by Google, the hardware provided by NVIDIA, and the cloud infrastructure from Google, Amazon and Microsoft?

Indeed, private markets are feasting on the gains of investments made decades ago, while failing to consider what the next crop ought to look like.

What future progress is being sacrificed to the goal of turning venture capital into a game of flipping jacked-up companies onto public markets?

Odin: SPVs & Fund Administration for truly adventurous capital

“Over 25 years, I’ve learned that you can’t predict the future. You can only decide what kind of future you want to live in and do your best to build it.”

Noubar Afeyan, Co-founder of Moderna and CEO of Flagship Pioneering

The world first encountered Moderna in 2020, when the inconspicuous Massachusetts biotech announced that it was developing a vaccine to combat the pandemic.

That vaccine, mRNA-1273, would be administered hundreds of millions of times across more than seventy countries. It introduced an entirely new class of medicine, where a synthetic strand of genetic code tells the body to manufacture a protein that the immune system can then learn to recognise.

As with most overnight success stories, the technology had been in development for more than a decade before it was launched onto the world stage.

“There was no ‘aha’ moment when the mRNA breakthrough happened. The Moderna platform was built on a constellation of technologies, methods, and know-how that evolved over time.”

What Evolution Can Teach Us About Innovation, by Noubar Afeyan and Gary P. Pisano

Moderna was founded in 2010 and went public in December 2018, one year before SARS-CoV-2 emerged. Their initial public offering raised approximately $604 million, the largest biotech IPO in history at the time.

The size of the offering was extraordinary given how little the company had to show for it commercially. There was no approved product, no late-stage clinical results and no obvious near-term revenue. The company was built on the promise of the mRNA platform. The central idea, that messenger RNA could be used safely and reliably, was viewed with skepticism by much of the pharma community.

So, Moderna had remained a “non-consensus” investment for most of its life, even as a listed company. The company’s existence can largely be credited to the investor and entrepreneur who decided that the question of mRNA was important enough to deserve a real answer. That investor was Noubar Afeyan, the co-founder and chairman of Moderna and the CEO of Flagship Pioneering.

Flagship Pioneering has now originated or fostered more than 120 ventures with an aggregate market value approaching $100 billion, with about $52 billion in aggregate exit value. Moderna is the most famous, but is just one output of a process that Afeyan has demonstrated is scalable and repeatable.

Indeed, Afeyan’s view is that Moderna was not the typically idiosyncratic success story of a “venture bet”, but the predictable result of a process that he had designed to build test and develop ideas that were overlooked for being too risky or unreasonable.

“We have to be willing to embrace unreasonable propositions and unreasonable people in order to make extraordinary findings. Because the notion that utterly reasonable people doing utterly reasonable things will produce massive breakthroughs, it doesn’t compute to me.”

Noubar Afeyan, in conversation with Tyler Cowen

A Stranger in a New Land

Noubar Afeyan was born to an Armenian family in Beirut in 1962. In 1975, as the Lebanese Civil War erupted, his family fled to Canada, which gave them citizenship and a chance to begin again. He grew up in Montreal, studied at McGill University, and then arrived at MIT to work on biochemical engineering, becoming part of one of the very first PhD cohorts to be educated at the hybrid frontier of engineering and biology.

In 1987, at the age of 24, he started a company. This was an unusual move in itself, since venture funding at that time tended to flow to executives from Merck or IBM rather than to fresh PhDs without a network in the tech industry. He had no obvious right to be doing what he was doing, which he would later recognise as a strength.

“I had no connections. I was an immigrant. I came originally from Lebanon, escaped the war, went to Canada, kind of grew up in a foreign place. So I had zero expectations. Zero entitlement. I was not even American, I was Canadian.”

Noubar Afeyan, in conversation with Karim Lakhani of Harvard Business School

PerSeptive Biosystems, the company he founded, made instruments for the protein side of the emerging life sciences industry. Over ten years, five as a public company, it grew to roughly 880 employees (more than 200 of them with advanced degrees) and became the largest player in its category. In 1997, PerSeptive merged with Perkin Elmer in a stock swap valued at $360 million. Afeyan, by then in his mid-thirties, spent his last working years at Perkin Elmer’s successor, Applera, where he helped initiate Celera Genomics, the company that raced the public consortium to sequence the human genome.

While running PerSeptive, he had also been co-founding other companies on the side, partly because he kept noticing that good ideas were not finding their way into the world. The conventional view was that entrepreneurship involved a founder dedicating themselves to solving just one problem, but he didn’t see why that had to be the case.

Rejecting the Lottery

By the late 1990s, Afeyan’s experience with founding multiple companies had him pondering a Drucker-esque question that would shape the rest of his career.

“Why isn’t entrepreneurship a profession? And if it was going to be a profession, how could it be a profession?”

Noubar Afeyan, in conversation with Sarah Guo of Conviction

What bothered him was the valorisation of chaos, the heroic creation myths, and the prizes and competitions that created a lottery-like mindset about startup success. Founders were expected to live or die on a single giant outcome while investors sensibly diversified their exposure across many.

In 1999, he co-founded a firm with Ed Kania, a seasoned venture capitalist from Morgan Holland Ventures. Initially they called it NewcoGen, for “new company generation.” Two years later, after a number of people warned him it sounded like a disease, it was renamed to Flagship.

The premise was to manufacture companies, rather than to invest in them. In its earliest years the firm blended traditional venture investing with company creation, generating a series of exits that built serious credibility. Adnexus Therapeutics sold to Bristol-Myers Squibb, Accuri Cytometers sold to Becton Dickinson, Morphotek sold to Eisai, and Hypnion sold to Eli Lilly. There were also public listings such as AVEO Pharmaceuticals, Agios Pharmaceuticals and BG Medicine, and Color Kinetics.

The exits generated by the firm to date have comfortably funded the experiment that mattered most to Afeyan, which was to see whether the act of company creation could become a discipline.

“I realised that becoming a co-founder is an interesting concept relative to a solo founder and that you could do more than one thing in parallel. That led to the founding of Flagship.”

Noubar Afeyan, in conversation with Gary Pisano

Venture Hypotheses

Flagship’s operating system has four stages, as it takes ideas through to full commercialisation.

In the exploration phase, small interdisciplinary teams of scientists generate “venture hypotheses,” speculative conjectures about what might be possible. Crucially, the firm runs many of these in parallel, on the order of fifty to a hundred a year, and explicitly avoids domains that are already crowded with other companies.

“We’re trying not to operate in a way that’s grounded in the present. We’re trying to say, ‘Okay, what could exist?’ And then we work backward.”

Noubar Afeyan, in conversation with Gary Pisano

Those early hypotheses are then exposed to a punishing external review process. Flagship presents them to working scientists, large pharmaceutical companies, academic experts and informed amateurs, knowing that most will react with scepticism. Indeed, the “organised scepticism” of science is being deliberately weaponised to surface flaws and force the idea to evolve.

The hypotheses that come through that process without being killed by a fatal objection advance to the prototyping phase, which Flagship calls a ProtoCo.

“In that phase, it’s the founding team who worked on the exploration, plus maybe three or four people that have expertise in a particular scientific area. This team of maybe six or seven will engage in answering literally killer questions. We want to falsify our hypotheses experimentally.”

Noubar Afeyan, in conversation with Gary Pisano

The prototyping budget, in line with the firm’s broader philosophy, is intentionally modest, typically $1 million to $1.5 million over nine to twelve months, designed to maximise what Afeyan calls the “learn-to-burn ratio.”

Of the hundred or so explorations begun each year, only ten or fifteen become ProtoCos, and roughly six to eight of those become NewCos, the dedicated companies that Flagship founds, funds and grows. The rest are killed. The discipline of killing weak ideas quickly is what makes it safe to test all kinds of outlandish-seeming hypotheses to begin with.

"Reasonableness in my view is the gravitational force that keeps innovation from happening. Because people enforce reasonableness in the scientific world on what you are allowed to say."

Noubar Afeyan, in conversation with Karim Lakhani of Harvard Business School

Indeed, the reason the system can pursue unreasonable ideas, Afeyan argues, is the scale at which they operate. With fifty or a hundred parallel explorations, abandoning one is not a tragedy. Within a single company, killing the idea often might mean losing the company.

Paranoid Optimism

The ingredient that holds this system together is what Afeyan calls “paranoid optimism”. This matches his description in a recent Boston Magazine profile, titled, “Moderna Cofounder Noubar Afeyan Is Afraid of Everything, And That’s His Genius.”

“I can see the scary part of just about everything, because we have to imagine it to try and go after it.”

Noubar Afeyan, in Boston Magazine

He describes the mindset as similar to learning to switch between the gas pedal and the brake pedal. A completely optimistic founder makes reckless commitments, while the totally paranoid founder never makes it to launch. Toggling between the two modes of belief and doubt is what keeps the firm honest about what could fail, and ambitious about what might work.

“If you’re a paranoid optimist, by that I mean you’re not so optimistic that you actually get reckless, but you’re not so paranoid that you get depressed. And this constantly questioning whether it really will work, or how big it will be, or whether you should be doing something else, that doubt, doubtful optimism which is really a tortured state to be in, is perfect for evolution.”

Noubar Afeyan, in conversation with Karim Lakhani of Harvard Business School

Afeyan has come to see this mindset as something absorbed from his Armenian background. Armenian history contains a pattern of toggling between survival and thriving, of populations that have endured by simultaneously preparing for the worst and refusing to be defined by it. Lebanon, he notes, has its own version of stubborn optimism, which has sometimes carried the country through crises and sometimes prevented its people from confronting the harder truths.

It also helps to explain something more subtle about how Flagship operates internally. Afeyan describes his most distinctive working habit as a deliberate willingness to be wrong out loud.

“I think that if you’re doing something that’s never been done before, you have to be very humble to recognise that you don’t know what the right thing to do is, but also nobody else does. And in that milieu, you need to create a culture where people are willing to say things that may be wrong so that others can say other things. And over time, whatever is right can emerge.”

Noubar Afeyan, in conversation with Tyler Cowen

Embodied Platforms

As you might have gathered, Flagship is not a venture capital firm in the conventional sense, as it does not primarily fund other people’s companies. Neither is it really a research institute, since it commercialises everything it produces.

It is closest, perhaps, to an industrial company whose products are other companies.

“Flagship is itself an experiment in a new breed of company that conceives and creates and launches and develops companies. So a company that develops companies is not a traditional way one fathoms what a company does. Our products are effectively platforms that are embodied within companies.”

Noubar Afeyan, in conversation with Tyler Cowen

Flagship reorganises the usual venture capital structure so that the people doing the science are protected as well as the people with the capital, increasing their tolerance for risk and the ambition of the challenges they tackle. Within Flagship’s VentureLabs, any team working on a platform hypothesis that fails will just move on to the next, recognising failure as a valid outcome. Disproving a hypothesis is as important as proving one if your goal is process-led experimentation.

This enables something else that the usual venture model has trouble with, which is patience. Because Flagship is a principal investor in its own creations rather than a fund manager working to a defined liquidity window, it can afford to spend a decade developing a platform like Moderna without forcing the issue. Some ventures have taken nine, ten, eleven years from exploration to clinical traction. Others, such as Generate:Biomedicines, have gone faster but only because the underlying tools, in this case modern AI, suddenly caught up with the ambition.

Developing Polyintelligence

Generate:Biomedicines is the company that best illustrates where Afeyan thinks the model is heading. It emerged around six or seven years ago from a Flagship exploration that asked whether it was possible to computationally design proteins of arbitrary function. The conventional wisdom, even within the AI community, was that this required exhaustive knowledge of folding structures and underlying physics. The Flagship team believed that the necessary information was implicit in the DNA sequence itself, since nature managed without ever knowing what a protein was.

“DNA has no idea what a protein is. There’s no idea what folding is, and yet, boom, the function follows. So we said there must be an encoding of that knowledge in the DNA. There’s enough data in there somewhere. There’s patterns that are basically encoded which we don’t understand.”

Noubar Afeyan, in conversation with Sarah Guo of Conviction

Generate:Biomedicines now has more than fifteen computationally designed antibody programmes, several in or advancing to the clinic. It is also the first major biology partnership Nvidia made. Other Flagship ventures have applied similar techniques to cell biology, DNA and RNA molecules, lipid nanoparticle design and what Afeyan calls “autonomous scientific discovery,” in which AI agents generate hypotheses, specify experiments, run them and iterate without continuous human intervention.

Afeyan uses the term “polyintelligence” to describe this interaction of human intelligence, machine intelligence and what he refers to as nature’s intelligence, the logic implicit in biological systems themselves. The argument is that biology is itself a form of computation, and that AI is now sophisticated enough to read patterns inside it that humans never could.

The Hit Factory

Of course, not every Flagship company has succeeded. Biotech is a brutal business and a meaningful share of even well-funded platforms reach the clinic and stall, or never reach the clinic at all. Some of the firm’s more recent listings have suffered in the post-2021 biotech downturn.

However, Flagship’s aggregate record is hard to argue with. The firm has originated more than 120 companies, supported dozens of public listings, and is one of the few venture vehicles in history that can credibly claim to have generated a public-health outcome at the scale of a global pandemic response. In the process, Flagship has scaled its capital base into the multi-billion-dollar range.

The rate of exits from Flagship’s portfolio (around 40% if you exclude acquisitions without disclosed terms) is remarkable by venture industry standards. With 21.6% of their portfolio companies achieving an IPO, the firm has double the estimated rate from a pool of “blue chip” venture firms.

There is some survivorship bias at work here, as biotech companies go through a more strenuous process prior to incorporation, though we can also see that Flagship has significantly outperformed an average of their biotech peer group.

Flagship’s success contradicts several pieces of common wisdom in venture capital.

• Most venture firms pride themselves on finding the best companies; Flagship has shown that some of the most valuable companies are the ones you build yourself.

• Most innovation strategies favour focused execution on a single asset; Flagship has shown that there can also be strength in optionality.

• Most cultures of innovation treat the founder as protagonist; Flagship treats the process as the protagonist and the founders as a supporting cast.

However, Afeyan has cautioned against interpreting this as a recipe for success. The system, as it has been designed, depends as much on its mindset as on its methodology. A mindset has been formed by decades of learning to toggle between fear and possibility. It is also a system that suits a particular kind of work. Companies in market-led, execution-driven industries, where the customer need is obvious and the technology is established, may find emergent discovery to be overkill. Where it makes sense are the domains that the venture industry has decided are too speculative, too long-term, or too unreasonable to invest in.

“The reason we use the word ‘pioneering’ is that there’s a first-in-kind character to what we do. There are always new things to explore and new places to go. Opportunities will constantly be created for what can come next.”

Noubar Afeyan, in conversation with Gary Pisano

Effectively, Afeyan has turned the perspective of the outsider into a methodology and applied it at scale with the development of Flagship. His willingness to pose questions that credentialed insiders wouldn’t dare to ask has been a critical asset.

His contribution to the history of innovation may end up being less about any single drug or company than about that broader proposition, which is that the patiently paranoid and optimistic pursuit of unreasonable ideas can be organised, scaled and repeated.

“We take hard earned money, we deploy it to do things that are damn near impossible once in a while. We reduce them to practice so they become not only possible but valuable. And yet people treat it like, ‘Oh well, you know, it didn’t work. There’s twenty different things we tried. One of them worked.’”

Noubar Afeyan, in conversation with Sarah Guo of Conviction

Launch and run your VC firm from your phone, with Odin

Watch our latest episode of Going Solo, with of Earthling VC

In this episode, Dan sits down with Arian Ghashghai, who founded Earthling VC in 2023 to back weird, fringe, early-stage companies across future computing, robotics, AR/VR, AI, bio, and more. They discuss the challenges of raising in one of the hardest LP markets in memory and why a wonky strategy can be an asset, and debunk some of the myths and incentive structures distorting venture capital today.

Watch the full conversation on YouTube

“You know, I think track records are really overrated. Some of Yale’s best investments have been with people that don’t have a track record. We took a couple of people out of proprietary trading at Goldman Sachs 25 years ago. If they had had a track record, it wouldn’t really matter because Goldman Sachs has a very different form of organization and a different way of giving resources to the people that are making investment decisions. But we didn’t even have those numbers. And it was really just a decision that this was a woman and this was a man that we thought were going to produce great returns, and they’ve done a really good job for the university.”

David Swensen, Chief Investment Officer at Yale Endowment

A common perception of venture capital is that of passive limited partners paying sophisticated managers to invest on their behalf. For some parts of the market, this is accurate. Particularly in mature segments, where managers have a clear track record to illustrate their performance and a scaled service designed to meet LP needs.

Where that doesn’t match reality is the emerging manager category, covering the first three funds of a new GP. Here, LPs have to make a judgement on potential that is remarkably similar to that of an early-stage investor. This means understanding all of the same principles of risk, with no shortcuts to success and much greater expected volatility.

Is that extra burden of work worth it? Well, the parallel stretches into performance. As with venture capital, those that embrace risk often reap the greatest reward. Emerging managers not only serve a vital role in the innovation ecosystem, but they also offer strong performance. Indeed, many of the most impressive outcomes have been debut funds.

Opportunities Worth Funding

When evaluating a potential pre-seed investment, there’s little to go on except the combination of the people and the idea. This has been explored by researchers through the lens of entrepreneurs as “designers of problems worth solving”, to encompass both aspects. Extending this concept to emerging managers, addressing the same challenge, they may be viewed as “designers of opportunities worth funding”.

In a practical sense, this means identifying managers who have found a novel way to frame venture capital’s central challenge. Those who understand the mechanics of the industry well enough to have produced a strategy that is differentiated and clever. There must be some reason for LPs to believe it will beat the mediocre median.

The need for this level of qualitative analysis, in both cases, reflects the lack of metrics to examine. The founder offers a proposition with no precedent or evidence, while an early-stage GP offers a portfolio of such propositions. Both are, to some extent, outliers. It is their idiosyncrasy that gives them the ability to build or recognise what others have not or cannot. That is their edge.

The two relationships also share a similar horizon of uncertainty. Until investments turn into realised outcomes, neither the GP nor the LP can be sure that they are making good decisions.

In venture capital, this uncertainty produces predictably bad investments, as investors seek the false confidence of “obvious” signals or patterns. Founders that went to the prestigious university, worked at the right company, or have the desirable personal characteristics. Insecurity encourages decisions that are superficially defensible; a trade-off against the outlier potential that venture capital requires.

“Few data points and long feedback loops make for slow learning, while the extreme uncertainty makes it difficult to learn the right lessons. In this business, it is too easy to confuse luck with skill. Cognitive psychologists Daniel Kahneman and Amos Tversky have shown that humans are prone to thinking errors—cognitive biases—when assessing uncertain events or making decisions in the face of uncertainty.”

Applying Decision Analysis to Venture Investing, by Clint Korover

LPs suffer from the same problem, for the same reasons. The difficulty in selecting managers, combined with the long wait for any validation, results in the same overweighting of superficial traits. GPs who went to the right universities, span-out of the right firms, or have the sheen of a natural salesman. Generally, allocators prefer highly-networked deal-guys with a pedigreed background. Unfortunately the industry is already filled with this archetype, and the data isn’t looking good for them.

Indeed, research shows more diverse investment teams tend to outperform. For example, increasing the share of women on the team by 5 percentage points (e.g., from 10% to 15% female hires) is connected to a 4.7% increase in success rates. Which is not to say that women make better venture capitalists, but merely that the firms which less obviously fit the Patagonia-wearing archetype are more likely to be judged on actual merit. Often held to a higher standard than their peers.

If your job is to identify outliers, which is the case for both GPs and LPs in the emerging manager category, there are no shortcuts. Understanding emerging managers through a qualitative framework like “designers of opportunities worth funding” keeps the focus on useful data, avoiding the comfort of pattern matching.

Alpha Isn’t Easy

As with founder background, or GP profile, the trap of chasing the obvious signals for confidence extends to the thesis itself. Be wary of any manager pitching the easy thing, because generating outperformance is definitionally not easy.

In 1998 the easy thing was an internet firm. In 2026 it might be the physical AI thesis. Capital flows toward the obvious idea because the obvious idea feels easy to price, easy to find downstream investment, and easy to defend if it fails. Critically, it’s far harder to explain why you passed, if it succeeds.

As a result, it takes a remarkably stubborn manager with unusually strong conviction to avoid the gravity of consensus.

“An early rise in market heat often gives rise to herding behavior, both due to resource providers taking others’ actions as credible signals of attractive opportunities and psychological biases toward relying on social proof.”

Windows versus waves of opportunity: How reputation alters venture capital firms’ resource mobilization, by Alicia DeSantola, Pavel Zhelyazkov and Benjamin Hallen

Along the way, a non-consensus founders will see their peers raise money more easily. Non-consensus emerging managers will see peer firms generate more impressive metrics and expand their fee income. Non-consensus LPs will see other allocators celebrating exposure to hot companies and impressive growth in their paper wealth. None of it matters. When it comes to investment, history really only remembers cash returns (or major cash destruction).

This can be a difficult truth to hold onto, because the industry is increasingly built around selling the confidence of simple ideas; whatever it takes to keep the capital flowing, and the party going. Every conversation, podcast or post is either aimed at marketing to LPs or marketing to founders, rather than offering a truthful reflection of the market.

For example, the line that a venture capitalist’s job is simply to “invest in great people” is a wild oversimplification. Any manager who repeats it without caveat or clarification likely hasn’t developed any real conviction of their own. However, as a sales pitch it presents venture capital as a simple product, accelerating capital velocity and reducing scrutiny.

The Persistence Problem

Venture funds are long-dated instruments, and they have been getting longer. The classic ten-year structure has become a twelve to fourteen-year structure once the standard extensions are used, and the time to fully return capital can stretch up to twenty years. The consequence for an LP is that a fund’s interim marks tell you very little about its eventual returns, and aren’t usefully predictive until around year eight.

Even when data on a previous fund is available, it’s no guarantee of performance. Persistence in venture returns is statistically significant, but slight. This is particularly obvious when looking at more recent data, as the expansion of private markets and influx of managers has further weakened performance over time.

Research indicates that top-quartile VC funds repeat top-quartile status ~45% of the time based on fully realised performance data, but only 33% of the time based on data known at the time of investment. This matches with Pitchbook analysis suggesting ~35% of top-quartile funds will repeat their quartile status in a subsequent fund based on the data about one previous fund.

In summary, having a couple of funds that look good based on current marks is a very different proposition from a legacy firm with twenty years of audited outcomes, where track record truly becomes meaningful. Again, this has a close analogue on the other side of the table. The successful serial founder is somewhat more likely to succeed a second time, but not vastly more than first-time founders or those with a past failure.

Fundamentally, an LP evaluating emerging managers is making a forward-looking judgement based on qualitative information, without the comfort of any hard numbers. As with the early-stage VC, this takes conviction and sophistication.

Cheap Talk

“The only thing that matters in this business are actual, realized, and distributed returns to LPs — which is why Sequoia Capital, one of the most respected venture capital firms, specifically tells its LPs to ignore unrealized gains and focus only on the cost of the investments and the actual gains distributed to LPs.”

When Is a “Mark” Not a Mark? When It’s a Venture Capital Mark, by Scott Kupor

There’s a phenomenon in hot venture markets whereby investors are more easily seduced by “cheap talk”, where founders describe huge future growth potential to inflate their ability to raise capital. This sets them on the slippery slope of trying and failing to meet challenging expectations with increasing desperation.

The equivalent of this for an LP results in what the Kauffman Foundation described in their 2012 report as the “n-curve”. Typically, venture fund IRR would produce a j-curve (an initial dip, followed by subsequent growth), but GPs were found to inflate initial performance metrics to meet expectations and make raising subsequent funds easier. Inevitably, those inflated metrics would crash later in the fund’s life, thus the n-shape of the NAV profile.

This deception is a well-studied feature of private markets, covered in a number of studies that show GPs systematically inflating metrics at around the time of fundraising.

“Our results are consistent with GPs managing reported performance in a way that maximizes the probability of raising a follow-on fund.”

Interim Fund Performance and Fundraising in Private Equity, by Brad Barber and Davis Ayako Yasuda

Fortunately, LPs are often smart enough to detect inflated metrics and deceptive GPs are less likely to close a subsequent fund.

“Using a large dataset of buyout and venture funds, we test for the presence of reported return manipulation. We find evidence that some underperforming managers inflate reported returns during times when fundraising takes place. However, those managers are less likely to raise a next fund, suggesting that investors can see through the manipulation on average. In contrast, we find that top-performing funds likely understate their valuations.”

Do Private Equity Funds Manipulate Reported Returns?, by Gregory Brown, Oleg Gredil and Steven Kaplan

This aligns with related findings from 20VC, whose data suggested that managers who were honest and transparent with their reported performance would ultimately deliver the strongest returns.

“The best managers mark down their portfolios fastest. There is a direct correlation between the best performing DPI managers and the speed and accuracy with which they mark down their positions.”

20VC Market Breakdown, by Harry Stebbings

The lesson here for LPs is that strong growth in early marks is not a valuable signal to drive funding decisions without careful scrutiny, and it may even be a negative signal if it reflects managers chasing “hot” investments, rather than good ones.

Certainly, there should be no prejudice against managers whose companies get off to a slow start, as constrained early funding may actually be associated with stronger future potential.

Hyperfluency

There’s a large body of research about what makes a good founder, and identifying good venture investment opportunities, some of which applies to selecting emerging managers through the parallels discussed in the sections above.

However, emerging managers are still operating within a known industry, rather than inventing new categories. Consequently, the risk is lower, the scale of extreme outcomes is smaller, and there’s a little more “ball knowledge” that a prospective manager can be tested on.

This knowledge can be classified in six main areas:

• Their approach to origination; sourcing potential investments

• Their process for screening and selecting investments

• What they understand about monitoring and capital formation

• How they view capital requirements and constraints

• The fundamentals of idiosyncratic and systematic risk

• The end-customers of VC (M&A and IPO dynamics)

The trickiest part of evaluating managers on these criteria is that there isn’t necessarily a right answer; idiosyncrasy can’t be scored with a checklist. Instead, the ideal answer is thoughtful, differentiated, and not obviously wrong. Just like startups, there are no patterns that predict success, but there are patterns that predict failure.

To put this in context with an example:

Imagine a hypothetical emerging manager. Someone who recently worked at one of the major AI labs and has now decided they want to pivot their career into venture capital.

They decide to raise a $10M debut fund, with a generalist thesis. They would have an easier time selling LPs on their “access” edge in AI, but they recognise while it’s the frontier today it may not be tomorrow. At $10M, they’re making smaller investments, which many peers warn them just isn’t enough to compete in this market, but they know it’s enough for the startups they’re interested in.

When asked about their fund horizon, they can clearly articulate the bifurcation of companies that will exit earlier versus positions that may be sold to larger firms, allowing a return on capital within the 10-12 years of the fund’s life at a compelling multiple.

When asked about competing against other firms making bigger investments, they explain that more capital doesn’t produce a higher rate of success, and actually it’s often toxic to the development of a startup. A capital constrained business that develops greater efficiency has a cleaner shot at an earlier exit.

When asked about their generalist thesis they can clearly articulate the lessons from past cycles, and the dangers of undiversifiable systematic risk versus the opportunity for alpha in idiosyncratic risk.

When asked about how they source opportunities, they have a clear and unique strategy for origination which includes university networks, online communities and data-driven sourcing tools.

For a new manager, this reflects a thorough understanding of the role of venture capitalist; to find and fund experiments that are relatively cheap and have a huge payoff if they work out. It shows they’re not greedy, and are motivated by returns rather than fees. Finally, it demonstrates they understand market dynamics, exits and the basics of entrepreneurial finance. This level of competence and clarity is what will give them the conviction to take unpopular but highly profitable positions.

For an LP, recognising all of this requires a similar familiarity. Just like screening pitches, it’s not something that should be farmed out to a junior allocator or automated through simple heuristics. However, the more you understand these principles, the more quickly and efficiently you can use them to aid selection.

Human Capital

Beyond the qualitative signals, there is some value in understanding the background of an emerging manager and what it might imply about their potential.

The most common bias amongst LPs is to believe that former founders have a particular edge, having experienced entrepreneurship first-hand. This is what could be called the “player-coach fallacy”, where people assume that the best coaches were former professional players. It turns out, this isn’t true, as they are quite different skill-sets.

There’s a specific study on this question, and how it relates to venture capital, based on analysis of 13,000 firms. Overall, former founders produced a lower rate of success than professional investors. While successful founders outperformed, it’s worth recognising that that group is likely to outperform at anything they attempt.

Another area of interest is the professional background and qualifications of an investor. Here, research from a couple of sources indicates that skills like engineering or law may be helpful in the mix of a small firm, when reviewing certain opportunities. On the other hand, a generic MBA may actually be a negative signal due to encouraging risk aversion. Again, there are no easy boxes to check, but you can consider the background of a manager as a part of their story, to see if it contributes to the coherence of the strategy they describe.

In a similar manner to evaluating founders, the strength of an emerging manager is likely what makes them unique rather than the attributes which fit a pattern. Outlier returns come from outlier people, and it’s the “spiky” qualities which may offer the greatest signal.

Network Capital

It’s often stated that venture capital is a relationship game, and (as we’ve covered recently) these relationships increase the surface area for biases that warp investment decisions.

This spreads up to the LP layer, where an LP is 1.78x more likely to select an emerging manager if there is an existing personal relationship. For contrast, they’re otherwise only 1.21x more likely to invest in a GP with a personal relationship, so this increase reflects the relationship compensating for additional uncertainty.

The paper also finds that an LP is 4.73x more likely to invest in a manager if there has been a prior LP-GP relationship, which reflects the significance of having some historical perspective, even if it’s just a professional relationship rather than performance data.

According to a survey by Commonplace, LPs prioritise referrals from within their network when looking for new managers. This is another parallel with the early-stage investor, where there’s a typical preference for referrals from other investors.

Again, there are downsides to relationship-driven investing. Just like the equivalent founders, strong but poorly networked managers (often those from non-traditional backgrounds, underrepresented groups, or outside major hubs) get filtered out early.

So, LPs that focus on leads sourced through their peer network are likely seeing a skewed and incomplete set of emerging managers. The very best opportunities may be invisible unless they actively expand their network or develop other channels.

Finally, while the strength of a manager’s network with other GPs is beneficial, there’s an important nuance to how that network is structured and thus how it influences performance.

Based on the premise that larger, broader networks are beneficial, and deeper relationships are not, relationships begin to look less like a sustainable moat for GPs. Effectively, any manager that builds a large network of other managers with aligned interests (relatively easy if you maintain a good reputation) has theoretically greater potential and less fragility than one leaning on privileged relationships with a handful of “tier 1” firms.

Founder Friendly

Another lens on the quality of managers, also often misunderstood, is founder NPS score. Essentially, how positively do founders rate their investors, and how does this relate to performance? Asymptotically, this might be negatively correlated.

In recent years, “founder friendly” has increasingly meant “moves quickly and doesn’t go hard with due diligence”, as larger firms seek to hoover up more competitive deals and give founders leeway to pursue high-risk strategies for growth.

“To compensate founders for their risk exposure, VCs offer an implicit bargain in which the founders agree to pursue high-risk strategies and in exchange the VCs provide them private benefits. VCs can promise to give founders early liquidity when their startup grows, job security when it struggles, and a soft landing if it fails.”

Risk-Seeking Governance, by Brian Broughman and Matthew Wansley

This direction clashes with past research that connects VC monitoring with a greater rate of innovation and successful startup exits. On the other hand, the recent laissez-faire attitude enables faster growth of paper value, from which founders and investors can enrich themselves through secondary transactions despite weaker ultimate outcomes.

This might explain why data from 20VC indicated a negative correlation between NPS and DPI, as investors who care more about exits might be more hands-on with founders than those just seeking to exploit market failure.

“There is an inverse correlation between the NPs of the manager, and the DPI provided. Most often, the higher the NPs, the lower the DPI, and vice-versa. There are core misalignments between founder and GPs that often aren’t spoken about, but most notably around liquidity timing and availability.”

Harry Stebbings, Founder of 20VC

The lesson for LPs is that, like early markups, NPS (and a general focus on founder friendliness) is not necessarily a positive signal. Just like DPI beats TVPI, successful companies beat friendly founder relationships. Which, of course, is not to say that anyone should be careless with how they treat founders. Wasting a founders time or interfering unnecessarily in their business remain cardinal sins in venture capital.

In summary, managers should understand that their role as a fiduciary means monitoring investments appropriately, and doing what yields the best possible exit, rather than the shortcut of bleeding inflated valuations with secondaries.

Mercenaries and Missionaries

To conclude, another popular framework for understanding founder quality and determination, applied to emerging managers; the mercenaries and the missionaries.

A mercenary will always choose the path that allows them to most easily access capital, whether they are a founder raising from VCs, or a GP raising from LPs. Their major constraint is the friction in that relationship, and they will do everything they can to reduce it, which might mean fitting a certain pattern and offering a certain narrative. Their career risk concerns will lead them to take less idiosyncratic risk (more likely to chase consensus), which will erode the quality of outcomes. They’ll compensate for weaker exits by exploiting inflated NAV for greater fee income and secondary liquidity. Essentially, they are rent-seeking chameleons.

On the other hand, a missionary has by definition chosen to embrace a hard problem, and will find a way to make it work regardless of difficulty. Their major constraint is the time and energy required, and they will do everything they can to maximise output and bring their vision to life. This is sheer determination at work, and it is how new ideas and new technologies are forced into the world. For an emerging manager, this might mean a ramen diet for the first few years while their portfolio slowly demonstrates value that may encourage LP interest. These are the people that venture capital needs more of, to surface the most important opportunities and increase the volume and velocity of profitable exits.

Over the last fifteen years, the market has been shaped by mercenaries to meet the demands of short-term opportunism. LPs would benefit from considering how they can better identify and reward the missionaries, particularly when looking at emerging managers.

It’s not an easy path, but then there’s no such thing if you’re seeking alpha.

Launch and run your VC firm from your phone, with Odin

“We found that when it came to success, cold deal flow and warm deal flow deals performed relatively similar in terms of likelihood for success or failure, with warm deal flow companies outperforming in this statistic by a little less than 2%.

However, there was an interesting finding when it came to the size of success. The successful cold deal flow companies outperformed in terms of overall exit size.

Cold deal flow companies saw, on average, a 16.2% higher ROI and required almost 18% less capital.”

Success and ROI of cold deal flow companies versus warm deal flow companies, by Anthony W. Richardson

If you ask a group of VCs to rate the primary sources of dealflow (warm intro from other investors, warm intro from founders, cold outbound, cold inbound), it’s safe to assume that cold inbound would come dead last.

This is reflected in their observed practices; a 2020 survey of VCs in the US found that only around 10% of deals originated from cold inbound, versus 20% referred by other investors and 30% from professional networks and warm intros via founders. Analysis from VentuRank found that cold inbound represented 64% of the opportunities that firms saw, but only 6% of their ultimate investments.

Superficially, it is easy to imagine why. Cold inbound comes with no network signal or peer validation. It is a sign that the founder is not coming from a position of strength, without existing relationships and in-group credibility to lean on. For an industry so heavily influenced by qualitative signals, this feels sub-optimal. The inbox is also just cumbersome to manage, beyond a certain point.

This is precisely why cold inbound is the most potent source of alpha.

Put another way, generating alpha is implicitly difficult. The easier it is to make an investment decision, the less alpha it is likely to generate. Cold inbound has the most friction, takes the most effort, and requires the greatest conviction, thus it yields the strongest outlier returns when it works.

The Unprepared Mind

“In the beginner’s mind there are many possibilities, but in the expert’s mind there are few.”

Shunryu Suzuki, Sōtō Zen monk

It’s certainly effective to build a hypothesis about a particular opportunity and make targeted investments, but most VCs will readily admit that the future is messy. That uncertainty is why the venture strategy exists to begin with.

So, while a “prepared mind” will contribute to great investment decisions, there are limitations. An early-stage investor must be fundamentally open-minded, willing to be surprised, and have a method of origination which exposes them to the totally novel ideas and founders. An unprepared mind, free from bias and full of curiosity, is the optimal lens with which to view these opportunities.

The critical biases in venture capital, mostly related to hindsight and attribution, influence what types of founder and company an investor is likely to look for, as well as what will be referred to them through their network. Cold inbound addresses this with a pipeline of opportunities that are not the product of a selective process, while referred opportunities (including via scout programmes, or junior roles at larger firms) are the product of a black box of third-party biases.

There are two specific vectors for bias in referred opportunities:

The Keynesian Beauty Contest

“We have reached the third degree where we devote our intelligences to anticipating what average opinion expects the average opinion to be. And there are some, I believe, who practise the fourth, fifth and higher degrees.”

The General Theory of Employment, Interest and Money, by John Maynard Keynes

Any investor whose dealflow comes primarily via referrals will face a problem that economist John Maynard Keynes explained with the analogy of a beauty contest.

In his example, the most beautiful contestant was defined as the one who collected the most votes from judges. As a result, each judge is left considering not which is the most beautiful contestant, and not even which contestant other judges will find most beautiful, but which contestant the other judges are likely to think will get the most votes. These layers of abstraction pull focus further away from the central question of beauty, degrading the quality of the judgement.

Through the act of a referral, the view on quality is similarly distorted by the referrer trying to predict how the referral will be received. The desire to be seen to offer high-quality investments leads referrers to prioritise superficial (and often misleading) signals, thus undermining their judgement on quality.

This issue plagues venture investments, generally, as investors become anxious about the enthusiasm of downstream investors. However, the issue is explicit when looking at referred opportunities that introduce another party into selection.

“This suggests that the very act of trying to use the information contained in the decisions made by others makes each person’s decision less responsive to her own information and hence less informative to others.”

A Simple Model of Herd Behavior, by Abhijit V. Banerjee

Affinity Bias

Everyone has heard venture capital described as a “relationship business”, but relationships are dangerous to any discipline that relies on objective judgement.

This problem is commonly referred to as affinity bias, which reflects our inclination to overweight the opinions of people we feel close to, regardless of their merit. Prioritising the opinions of poorly informed friends over expert strangers.

“The credibility and influence of information will be higher as the relationship between the sender and receiver is close [...] People with whom one has strong ties are regarded as more credible and reliable than people with whom one has weak ties [...] Information derived from strong ties is perceived as more useful and information derived from weak ties is considered less valuable, even questionable.”

Impact of tie strength and experience on the effectiveness of online service recommendations, by Dong-Mo Koo

Affinity bias is reflected in a body of research studying the influence of relationships in venture capital. To summarise the findings, it appears while the breadth of a VC’s network is a positive influence on performance, the depth of those relationships actually degrades performance over time.

“Once we include VC-pair fixed effects into the analysis, we find that higher levels of past coinvestment activity leads to fewer, not more new coinvestments. Moreover, higher levels of past coinvestment lead to lower exit performance, a result that continues to hold after controlling for endogeneity.”

Getting Tired of Your Friends: The Dynamics of Venture Capital Relationships, by Qianqian Du and Thomas F. Hellmann

Essentially, the most beneficial relationships are those that are held at arm’s length, in order to remain objective about any signal that is transferred. Obviously, this is easier said than done given the incestuous nature of venture capital.

Process Alpha

Theoretically, the highest potential venture capitalist is a solo generalist, with no bounds on strategy, stage or category. Like Manfred Macx of Accelerando, they would consume vast amounts of data, distil those insights for opportunity, and identify potential in the most unique and unlikely places.

Unfortunately, that’s not yet possible. Instead, venture investors must build a process that curates their pool of opportunities, rather than trying to boil the ocean. It’s a necessary trade-off, sacrificing an unknown number of outliers in pursuit of a broadly better rate of success.

This gets at the core of the cold inbound challenge. How can investors properly manage the firehose of founders looking for investment?

Logic Gates

“Our job is to find a few intelligent things to do, not to keep up with every damn thing in the world.”

Charlie Munger, former Vice Chairman of Berkshire Hathaway

The first step of this process is for an investor to decide the scope of their inbound.

At the top of the funnel, there’s a question of brand. A firm can position themselves to maximise visibility, becoming a default destination for founders raising capital, or they can target exposure to particular communities and categories. A maximally visible brand will receive a significantly larger volume of applications, most of them poorly fitted, and will need much more active filtration to remain useful. A more targeted brand will see much less, but it will tend to be more relevant.

Then there is the question of explicit filters on stage, sector, geography and ticket size. These are useful, and most firms apply at least two of them, but it is worth being honest about what they cost. Every filter narrows the pool, and an unspecified portion of the most interesting opportunities fall outside of those bounds.

A firm that describes itself as investing in pre-seed rounds for vertical AI will have an easy narrative to pitch LPs, but it will miss many great opportunities that may only be a short step outside that thesis.

On the whole, filters allow a firm to more reliably make good decisions by preserving bandwidth, but they also dampen the upside. The right approach balances these two forces around the competencies of the partners, rather than catering to industry norms or LP expectations.

Judging Books by Their Covers

“The single most powerful pattern I have noticed is that successful people find value in unexpected places, and they do this by thinking about business from first principles instead of formulas.”

Zero to One, by Peter Thiel

The next question is how investors approach screening the opportunities that make it through the filter. The most common shortcut in early-stage investing is pattern-matching on founder credentials (e.g., which schools they attended, which companies they worked at, or past entrepreneurial experience). These data points feel like they should be meaningful, but more often become a vector for dangerous biases rooted in comfortable familiarity or common wisdom.

Venture capital is an outlier business. The returns are produced by a small number of companies that, by definition, cannot be systematically identified. There is no checklist that reliably surfaces a higher-potential founder, and the more an investor leans on that approach, the more they are optimising for the mediocre middle of the distribution.

Markers like a Stanford degree, FAANG pedigree, or simple charisma, are all appealing because they address the insecurity associated with staged capital and the question of downstream investor interest. They help build confidence, pushing up prices without actually being predictive of future success. A Keynesian beauty contest emerges, with the seductive promise of easier markups.

A more literal version of this superficiality is seen in investor’s preference for founders with facial features that match their own. “Mirrored matching” is responsible for a 7% drop in the rate of successful exits, according to one study.

Productive Constraints

“James Montier pointed out in a recent piece that when athletes were asked what went through their minds just before competing in the Beijing Olympics, the response again and again was that the competitor was focused on the process, not on the outcome. The way to maximize outcome is to concentrate on process.”

Seth Klarman, President & CEO of The Baupost Group

The most instructive precedent is the way that Y Combinator presents batch companies to investors, through their platform. This single-slide format allows investors to efficiently browse through around 250 companies per batch.

Alongside the obvious efficiency gain, this format strips out a great deal of the surface area for biases. The first point of contact isn’t a resume or a discovery call. Investors are focused on the critical elements of what is being built, and are therefore less vulnerable to over-weighting irrelevant similarities. As a result, Y Combinator’s approach has been determined to reduce credentialism and the associated drag on investment performance.

Essentially, the applications that stand out at this stage are those that effectively demonstrate that the founders are designers of a problem worth solving.

This is an example of thoughtful process design, and a great example for any firm that takes the research on biases in venture capital seriously. The objective is to maximise the probability of catching outliers by compressing a pitch down to the critical idiosyncrasies, thus magnifying an investor’s analytical ability.

Staging Information

The practical application of this theory is a staged process that prioritises the most valuable (high signal, low risk of bias) information at each step.

The goal is to present the partner with the most meaningful inputs, ordered by importance, to enable the highest ROI use of time. A maximally efficient and high-fidelity interface between founder and investor.

For an early-stage venture capital firm, the implementation could be as simple as a form, laying out the questions below:

1. A reasonable first stage asks for the basic information and a hook for ambition.

What is the problem you are working on? This screens for ideas that are obviously bad or in conflict with what the firm has already backed. It also surfaces, in the choice of words, how the founder thinks about scope.

Why do you believe it matters? This is partly a check against the founder’s personal conviction, and partly a chance for the investor to discover something fundamental about the problem itself.

2. A second stage, for those that pass the first, presses on the qualitative signal in the founder’s relationship to the problem.

What do you believe about it that nobody else does? This is the test for novel insight, and for the kind of problem design described by Bianchi and Verganti. A founder without a compelling position is more likely to be an opportunist, chasing capital.

What is your relationship to the problem? This surfaces the founder’s personal stake in the concept. This is not a hard requirement, but it is one of the better predictors of the persistence, and a more explicit lens on domain expertise.

3. A third stage, finally, looks at trajectory.

What have you achieved so far? Traction and milestones, articulated honestly. Founders who oversell at this stage are offering useful information of a different kind.

What is the next milestone? Velocity matters as much as the current position, and a founder who can describe concrete near-term goals is providing useful signal on how they think about execution.

The total length of all six answers should be modest, with each subject to a character limit. The goal is to deliver something a partner can read in two or three minutes, with enough signal to know whether to keep going but nothing that introduces material bias. As an added bonus, this structure discriminates against the overly-long AI slop which plagues cold inbound today.

Finally, a firm may choose to take the lesson from Y Combinator a step further, and ask founders to record a short video talking about their company. This would be the final step in the process before a meeting, to get a sense of whether the founder can credibly and fluently articulate their ideas. It also adds a degree of friction to the process which will filter out a large volume of spam submissions.

Undoubtedly, the process will need refinement over time.

Some firms will find they remain swamped, and choose to add more friction.

Others may adjust character limits to make answers more easily digestible, or simply recalibrate their personal hurdle for accepting a meeting.

This ability to learn and iterate is a core strength of a well-designed process, which offers a measurable input and output with dials to adjust.

The Atomic Question

The objective of all of this is to reduce a noisy inbox to a single binary question that a partner can hold in their head while rapidly reviewing potential investments.

Has this team managed to design a problem worth solving?

It’s a simple question, though it encompasses a holistic view of the opportunity.

Submission should be compressed into a form that allows that question to be most easily answered, whereafter the partner may easily move forward with a meeting or swipe it onto the discard pile. Each subsequent stage is a deeper investment of time which should be made on the basis that it has been earned by the previous stages.

This is what process alpha looks like in practice. It’s not about being smarter than the market in some absolute sense, but gaining an edge through exposure to a wider field of potential, stripping out the inputs that distort judgement, and focusing attention on the single most important question.

Cold inbound is no poor man’s deal flow. It is a structurally underpriced source of alpha which rewards the few investors who are willing to take it seriously.

“VC funds must further experiment on the product of VC, building new, more open processes that leverage our industry’s considerable resources to unlock the potential of outsiders and gain access to new markets and new ways of thinking. Funds can start this journey by eliminating warm introduction requirements.”

Ban Warm Introductions, by Del Johnson

Launch and run your VC firm from your phone, with Odin

That is the textbook version of venture capital, but it doesn’t really exist today.

In April 2026, Robert Bartlett and Paolo Ramella of Stanford Law School, released a paper examining the influence of extended liquidity horizons in venture capital. “The Disappearance of the Ten-Year Fund” draws on linked quarterly cash flow, NAV, and portfolio company data from PitchBook covering funds formed between 1995 and 2014.

Their finding is that the ten-year horizon (which supposedly anchors fund accounting, performance reporting, fundraising cycles, and LP expectations) no longer corresponds to anything in the underlying economics of venture market.

“Unrealized NAV late in fund life has risen sharply across vintages, especially in venture capital, and many funds continue distributing capital for more than twenty years.”

The Disappearance of the Ten-Year Fund, by Robert Bartlett and Paolo Ramella

For the 2010 to 2014 vintage, the median venture capital fund still reports year-ten NAV that exceeds total committed capital. Such a significant share of fund value sits unrealised when, in theory, the vehicle should be winding down.

Bartlett and Ramella observe that fund lengthening is not because modern funds are converting NAV into cash more slowly than their predecessors; the speed of distributions after a liquidity event hasn’t changed significantly. Portfolio companies are simply staying private longer, and getting much larger.

“Elevated late-life NAV primarily reflects greater value creation: these funds hold portfolio companies that reach substantially larger valuations by year ten and exhibit more extreme right-tail outcomes.”

The influence of extended liquidity horizons on venture capital performance is obviously negative, from a time-value perspective. If a sizeable amount of value remains unrealised in year ten, interim IRR must blend actual distributions with estimated valuations. Those metrics drift downward as the liquidity horizon stretches further, unless the unrealised portion appreciates exceptionally quickly.

This downward drift is systematic, larger in venture capital than private equity, and increasingly pronounced in the most recent vintages where late-life NAV is the most inflated.

“Investors evaluating managers, especially in VC, should expect greater IRR-unwinding when a large share of value remains in NAV, even when interim IRRs appear strong […] These findings challenge the use of ten-year fund structures and interim performance metrics as reliable guides to the timing, risk, and performance of private fund investments.”

So, if the ten-year structure is functionally dead, why does the industry keep using it?

Parkinson’s Law

Industry practitioners have been quietly aware of the lengthening for years. In some ways it is a parallel to Parkinson’s Law:

“Work expands so as to fill the time available for its completion.”

In a venture capital context, that might be reframed as:

“Funds expand so as to absorb the capital available for management.”

This is a reflection of the goal shifting from the fiduciary relationship of delivering the best possible performance, to a service relationship of managing capital at scale for a larger community of LPs who needed a place to park it that was labelled “venture capital”.

As a result, SVB’s State of the Markets H1 2026 report describes a venture market that has split into two largely separate industries, operating out of one allocation bucket. At one end are the massive growth rounds dominated by mega-funds; at the other, a shrinking cohort of disciplined early-stage investors. In 2025, 33% of all US venture dollars flowed to the top 1% of companies by valuation, up from 12% in 2022. The mega-fund segment, with deals over $500 million, captured a far larger share of total investment than at the 2021 peak.

Inside this distorted market, fund extensions are just a fact of life for LPs. Top-quartile funds routinely take sixteen to twenty years to fully return capital. A growing share of 2010 to 2015 vintage funds remain active, with material NAV still on their books.

The traditional harvest phase has been replaced by an extended growth phase, creating space for larger pools of downstream capital from larger funds. Companies in modern portfolios continue to scale, often dramatically, well after the ten-year mark.

So the argument is not that capital is stuck and entirely unproductive, but that it has enabled investing across timelines that aren’t exactly what LPs signed up for, and that there is a large question mark about the resulting performance.

Paper Tigers

If the ten-year structure has lost its connection to the reality of venture capital, why has it not been replaced? The answer is that it serves several functions that have nothing to do with actual fund outcomes.

• LPs compare funds by vintage year and use these reference points to make portfolio construction decisions. Cash-flow models for pensions, endowments, and sovereign wealth funds are built around predictable distribution schedules. So, even when the underlying funds routinely blow through those schedules, the schedules provide an illusion of standardisation that enables comparison.

• A ten-year fund is intelligible to a board, to a trustee, to anyone who must explain the asset class to people who aren’t familiar with it. An indefinite-horizon vehicle feels much less certain and therefore is harder to defend, even if it is closer to the truth. The industry has continued to write ten-year LPAs partly because change is uncomfortable, and invites too many other difficult questions.

• Successor funds are typically raised three to four years after a predecessor closes, on the implicit understanding that the predecessor will be approaching early distributions by the time the new vehicle is investing. However, when predecessor funds carry large unrealised NAV positions for years past their stated end date, there’s an awkward overlap. The GP is asking LPs to evaluate them based on interim performance metrics that, as Bartlett and Ramella demonstrate, are mechanically inflated by exactly these unrealised positions.

Normalisation of Deviance

“Normalization of deviance is a term first coined by sociologist Diane Vaughan when reviewing the Challenger disaster. Vaughan noted that the root cause of the Challenger disaster was related to the repeated choice of NASA officials to fly the space shuttle despite a dangerous design flaw with the O-rings. Vaughan describes this phenomenon as occurring when people within an organization become so insensitive to deviant practice that it no longer feels wrong. Insensitivity occurs insidiously and sometimes over years because disaster does not happen until other critical factors line up.”

When Doing Wrong Feels So Right: Normalization of Deviance, by Mary R Price and Teresa C Williams

For all that maintaining this comfortable status quo suits the managers who profit most from it, there are costs involved.

For a start, it probably doesn’t set a good precedent that the GP:LP relationship begin with a lie. If neither side expects a ten-year fund, yet both sign a contract explicitly acknowledging that structure, it raises questions about whether other contractual obligations are also perceived as rough guidelines.

The large unrealised positions that persist alongside successor fundraising put strain on the LP model. The GP is simultaneously stewarding old assets and marketing new vehicles, drawing on overlapping resources, reporting on overlapping metrics and benefitting from overlapping fee streams.

Another concern is what could be referred to as the re-risking trap. Mega-funds with billions of dollars in dry powder push companies into ever-larger rounds at ever-higher valuations, betting on power-law outcomes that justify their scale. This pattern keeps risk elevated far longer than the traditional venture curve. A company that might once have reached an early profitable exit at $200 million is now urged toward a Series E round implying a multibillion-dollar outcome, because anything smaller does not move the needle in a $5 billion fund. Smaller managers, who could profitably exit at the lower threshold, get caught in the same long-horizon game even though their fund economics, ownership levels, and investor expectations are entirely different.

Permanent Capital

“Ironically, innovations in venture capital haven’t kept pace with the companies we serve. Our industry is still beholden to a rigid 10-year fund cycle pioneered in the 1970s. As chips shrank and software flew to the cloud, venture capital kept operating on the business equivalent of floppy disks. Once upon a time the 10-year fund cycle made sense. But the assumptions it’s based on no longer hold true, curtailing meaningful relationships prematurely and misaligning companies and their investment partners.”

Roelof Botha, Sequoia Capital

Sequoia’s 2021 restructuring was the most visible acknowledgement that the ten-year fund had become a liability for a firm operating at the top of the market. By moving to an open-ended structure, Sequoia could hold post-IPO positions indefinitely, fund new investments through internal recycling rather than periodic LP commitments, and offer LPs semi-annual redemption rights instead of fixed distributions. The model, which resembles a hedge fund grafted onto a venture franchise, was made possible because Sequoia’s track record allowed the firm to do something novel without spooking LPs.

Other large firms have moved in similar directions. Andreessen Horowitz and General Catalyst registered as investment advisers, expanding their flexibility to hold public securities and pursue non-traditional asset classes. The architecture of the mega-firm has, in practical terms, abandoned the ten-year horizon. What remains in the LPAs is a vestigial structure, preserved out of convention and regulatory familiarity.

The largest investors that entered the venture asset class in size during the 2015 to 2022 boom were sovereign wealth funds, pension allocations, family office platforms, and crossover funds that operate on indefinite or very long horizons. That capital has flowed into mega-funds, which deploy it into very large rounds at very high valuations. Companies that receive that capital have less reason to go public and more capacity to scale privately. The ten-year framework, designed for an era of smaller funds and faster exits, was left supporting a market that had outgrown it.

The logical response for firms operating at this scale is Sequoia’s; permanent capital, open-ended structures, and indefinite horizons match the underlying economics of holding the top of the power law for as long as it continues compounding.

Nimble Capital

The same logic does not apply for small funds, where the analysis breaks in the opposite direction and the disciplined ten-year fund may be favourable.

This is partly a story about the law of large numbers and partly a story about ownership. A small fund concentrating its capital in early rounds at modest valuations can convert a single outsized exit into a fund-returning event where a large fund cannot. To move the needle on a $1 billion fund, a manager needs $3 billion or more in realised value, which requires either many large outcomes or one truly enormous one. The mathematics push large funds toward the re-risking trap and small funds toward the discipline that has historically produced the best venture returns.

A 2025 paper by Guanrou Deng and colleagues, published in The European Journal of Finance, helps to illustrate this point. Using a sequential investment allocation model calibrated to PitchBook data on US later-stage venture rounds, the authors derive the relationship between portfolio payoff and investment period length. The result is an S-shaped curve.

“The optimal payoff is achieved when the VC fund is terminated around the 10-year period […] Beyond year 10, the curve flattens, and payoffs may even decline, suggesting diminishing returns from holding investments for too long […] Our conclusion that a 10-year investment horizon is optimal for general VC funds remains valid.”

Optimizing investment period length and strategies for later stage venture capital staged financing portfolio, by Guanrou Deng et al.

In their model, years one to four are flat, with portfolios in their early build phase. Years four to ten produce the bulk of payoff growth, as companies mature and successful exits compound. After year ten, the curve flattens and may decline. Holding for longer does not reliably produce additional return, but does mechanically reduce IRR through the time-value drag that Bartlett and Ramella also document.

The peak falls between years eight and ten, which matches the average exit duration for successful venture-backed companies in the PitchBook sample and pre-dotcom IPO trends. For a small fund operating with disciplined portfolio construction and a clear view of when individual positions should be sold, the ten-year horizon appears to be an optimal target.

The Handoff

If small funds are to deliver liquidity within the ten-year window, then the question becomes how to convert paper gains into cash within that time.

One answer to this, in the longer term, is for smaller funds to proactively align companies toward IPO more quickly. This means trading top-line growth of portfolio companies (and spurious markups) for stronger economics, producing companies that are capital efficient and rationally valued, ready for public market scrutiny.

In the shorter term, the answer is the secondary market. Darian Ibrahim’s 2012 paper, in the Vanderbilt Law Review, made the foundational case more than a decade before the current liquidity crunch.

“Secondary markets offer initial investors a new path to liquidity, offer buyers access to a previously untapped class of assets, and produce governance benefits for traded firms. The realization of these benefits in venture capital should lead to a net increase in the total amount of entrepreneurial activity. Given the surplus that entrepreneurial activity produces for society, VC secondary markets should be studied by academics and encouraged by policymakers.”

The New Exit in Venture Capital, by Darian M. Ibrahim

Ibrahim’s observation, drawn from interviews with secondary market participants, was that the direct market for stock in private start-ups was already functioning as a release valve well before the current cycle. Late-stage VCs frequently bought preferred shares from early-stage VCs as part of larger financing rounds. One of his interviewees estimated that 60–70% of late-stage VC financing rounds have a secondary component to them. Early investors could obtain partial liquidity, the company could remain private, and late-stage capital could enter without being constrained by the early-stage fund’s expiry clock.

The elegance of this relationship is that it aligns each participant’s natural horizon with the segment of the company’s life cycle they are best positioned to underwrite. Early-stage funds bring an appetite for idiosyncrasy, qualitative expertise and pricing discipline at the seed and Series A stages. Late-stage funds bring scale capital, quantitative expertise, and the patience to hold through the long pre-IPO window. The secondary transaction at Series C or D is the point where these two parties naturally meet, and both would appear to benefit.

For a small fund operating on a ten-year horizon, the implications are straightforward. Concentrate on early-stage where the pricing edge is real, consider whether or not to reserve capital for companies through Series B, and treat Series C or D as a default decision point for partial or full secondary sales to multi-stage and crossover investors.

Secondaries are no longer the peripheral market that Ibrahim was describing in 2012. Wellington estimates the venture secondary market reached approximately $160 billion in 2025 and is on track to become a mainstream liquidity tool. The conditions Ibrahim anticipated have arrived, although important questions remain about transparency and efficiency.

Collective Conservatism

“We conjecture that venture capitalists and their investors often fall prey to what is known as ‘collective conservatism.’ We investigate this conjecture by analyzing boilerplate provisions in limited partnership agreements. When investors accept suboptimal boilerplate provisions it is not because they believe that the standardized terms and conditions sufficiently align the interests of investors and fund managers, but merely because they think their peers, including their competitors, prefer to include them in the limited partnership agreement.”

Conservatism and Innovation in Venture Capital Contracting, by Joseph A. McCahery and Erik P. M. Vermeulen

The strangest feature of the current venture market is how obviously it has bifurcated while continuing to offer a single standard fund product. Capital has stratified, strategies have stratified, exits have stratified, and the underlying companies have stratified. Fund structures, through sheer inertia, have remained the same on paper despite clearly divergent market realities.

Mostly, this can be explained by the reluctance amongst venture capital firms to present their LPs with novel concepts and risk rejection. Indeed, the way in which venture capital models structures and strategies around LP expectations has clear parallels to “catering” in the GP:founder relationship, likely with the same negative consequences for performance. However, as LPs become increasingly uncomfortable with the market reality, there’s an obvious opportunity for change.

The case made by Bartlett and Ramella is thorough, drawing on data that allows them to trace fund-level outcomes back to individual portfolio company exits. However, their conclusion doesn’t call for the ten-year structure to be abandoned. Instead, they highlight it’s no longer a reliable description of industry practices, and that performance evaluation, fund design, and LP expectations need to be reconsidered accordingly.

Embracing Divergence

What the industry now needs is a divergence in fund products that matches the divergent market reality and LP requirements.

Mega-funds and platform managers with the scale, brand, and track record to demand permanent capital should pursue the path that Sequoia has outlined. The companies they back stay private longer, grow larger, and justify the holding period. Here, the ten-year structure has become a fiction, and the honest response is to move on to more rational arrangements.

Small funds should head in the opposite direction. They should use the ten-year horizon to enforce discipline as a competitive feature that will be appreciated by their community of LPs. A fund that explicitly targets full or majority realisation by year ten, focuses on early stages where it has structural ownership advantages, and uses Series C or D secondary transactions as the default hand-off to larger investors.

The rationale for the small fund’s ten-year revival is the same that Bartlett and Ramella identify for the structural breakdown; modern private markets create more value, more slowly, in fewer companies. That logic supports very long horizons at the top, where the power-law winners are held. It also supports disciplined short horizons in early-stage, where capital efficiency and risk management are most critical.

LPs also face their own pressure to choose a deliberate strategy. Endowments and pensions whose annual payout obligations are growing have begun to question whether the indefinite horizon embedded in mega-fund holdings serves their cash flow needs. A small-fund product designed around genuine liquidity within a decade is an increasingly attractive proposition.

In summary, the ten-year fund disappeared in a period where the venture market was struggling with capital indigestion. Now that is resolved, through the divergence of strategies with clearly differing objectives, there’s an opportunity to bring it back.

Essentially, the opportunity for small and emerging managers may be to finally make good on a promise that venture capital has neglected for well over a decade.

SPVs & Fund Administration purpose-built for global venture capital

Rock’s conviction, which he repeated for decades, was that backing people was the core of venture capital. A great management team, he liked to say, can find a good opportunity even if they have to jump from the market they currently occupy.

His contemporaries saw things differently. Tom Perkins at Kleiner Perkins trained his eye on technology, asking whether it was proprietary and meaningfully better than alternatives. Don Valentine, who had run marketing at Fairchild before founding Sequoia, became obsessed with the market. When Sequoia considered an early investment in Cisco in the mid-1980s, most of his peers passed; the founding team was considered weak. Valentine invested anyway, reasoning that the networking market was so large that even a mediocre team would sell a great deal of equipment.

The three men gave rise to three distinct philosophies of American venture capital; but Rock’s has won the cultural war. Not only does “venture capital is a people business” make an excellent slogan, it also puts the founder at the center of the story. And if you’re in the business of selling capital to founders, that’s exactly what they want to hear.

But is it really that simple? And what does a “people business” actually look like?

Normative Conformity

Today, virtually every venture capital firm presents itself as founder-first.

In 2016, four economists (Paul Gompers, William Gornall, Steven Kaplan, and Ilya Strebulaev) surveyed 885 venture capitalists at 681 firms to understand how they made decisions. Their study is the most thorough analysis of the industry’s decision-making, and appears to close the book on the Perkins and Valentine philosophies.

Around 53% of early-stage respondents ranked founders as the single most important factor in deal selection. Business model and product, Perkins’s traditional territory, were selected by roughly 10%. Market and industry, Valentine’s preoccupation, were selected by about 6%. The remainder scattered across valuation, fit with the fund, and the investors’ own ability to add value.

“96% (92%) of VC firms identified team as an important factor and 56% (55%) identified team as the most important factor for success (failure). Team was the most important for all subsamples, but particularly important for early-stage and IT VCs.”

How Do Venture Capitalists Make Decisions?, by Gompers, Gornall, Kaplan, and Strebulaev

Looking at other responses from the survey, 9% of investors admitted to using no financial metrics, which rose to 17% among early-stage investors. An industry so reliant on qualitative judgement might be expected to have thought about the judgement criteria and how the outcomes may be tracked.

Unfortunately, that answer remains a vague promise of investing in “the best founders” without the ability to articulate what that means, or why.

“The findings suggest that VCs are not good at introspecting about their own decision process. Even within the confines of a controlled experiment, which greatly reduces the amount of information considered, VCs lacked a strong understanding of how they made decisions.”

A Lack of Insight: Do Venture Capitalists Really Understand Their Own Decision Process?, by Andrew Zacharakis and G. Dale Meyer

As a result, the founder-first approach to venture capital has produced an epidemic of lazy thinking, infiltrated by biases and credentialism. This, in turn, is reflected in slipping performance and frequent horror stories of fraud and negligence.

A Billion-Dollar Blind Spot

In 2022, a Chicago Booth economist named Diag Davenport put a dollar figure on what this overly simplistic attitude has cost the industry.

Davenport built a machine-learning model on a dataset of more than 16,000 startups, representing over $9 billion of committed capital. He trained the model only on information that had been available to investors at the time of the decision, and asked how many of the investments that venture capitalists actually made could have been identified, ex ante, as worse than putting the same money into a standard public-market alternative? The answer was approximately half.

By dropping the bottom half of investments and redirecting the capital into a public-market option, Davenport found that venture returns would have been between 7 and 41 percentage points higher across the sample. In the data he worked with, that amounted to over $900 million of avoidable loss. The cost of the poor investments, expressed as a spread over the outside option, was around 1,000 basis points.

Davenport trained two parallel algorithms, one predicting which startups would become the best investments and another predicting which would become the worst. When he compared the signals that each model relied on, there was a curious pattern. The algorithm built on good outcomes leaned on product characteristics, while the algorithm built on bad outcomes leaned heavily on the founder ‘s background. When investors were making good decisions, they were looking more carefully at the idea. When they were making bad decisions, they appeared to be looking more carefully at the team.

To test the overweighting, Davenport built a separate model that used nothing but founder-education data, and he asked whether two companies that looked equally promising under the full model received different investment outcomes depending on how they looked under the education-only model. The model indicated that investors systematically overweighted education, and they did so most heavily for the startups that would go on to perform the worst.

“Investors seem convinced that the founder-first model of the world is the correct one. This likely facilitates investors neglecting to notice features that are predictive and a feedback loop of never noticing or learning persists, consistent with the model and evidence presented in Hanna et al. (2014).”

Predictably Bad Investments: Evidence from Venture Capitalists, by Diag Davenport

Davenport’s paper is part of a growing body of research that has reached similar conclusions, indicating that investors overweight shallow founder attributes in ways that produce predictably bad investments (errors of comission) and predictably good missed opportunities (errors of omission).

There’s a structural explanation for this; “success” in venture capital is more easily measured by incremental fundraising than by distant exits, and funding friction falls if investment decisions become a simple box-checking exercise.

Somewhere along the way the industry convinced itself that the ability to raise capital was itself a desirable founder trait and this logic became recursive. Investors began pattern-matching on the founder archetype most likely to raise the next round, making that archetype easier to fund, reinforcing the pattern. As a result, the quality of returns has largely declined while capital velocity (and fee income) has accelerated.

This loop is explained by the economist Daniel Kahneman, who describes how even sophisticated professionals can be seduced by simple, coherent ideas if they are aligned with the right incentives. Even when they produce obviously bad results.

“The statistical evidence of our failure should have shaken our confidence in our judgments of particular candidates, but it did not. It should also have caused us to moderate our predictions, but it did not. We knew as a general fact that our predictions were little better than random guesses, but we continued to feel and act as if each particular prediction was valid.”

Don’t Blink! The Hazards of Confidence, by Daniel Kahneman

Make it easy to back people building great ideas in great markets - run your venture firm from your phone with Odin.

The Paradox of the Good Investor

This creates an interesting puzzle. The data shows that overweighting founder attributes leads to worse investment decisions, particularly among the worst-performing deals. Yet some of the most successful firms in the industry are also the most aggressively founder-first.

• Founders Fund has spent two decades backing unusual people before anyone else would. Peter Thiel also created the Thiel Fellowship, for young entrepreneurs who didn’t have a college degree, which has produced incredible success stories.

• Y Combinator has run for twenty years on the premise of identifying great founders. In fact, the programme has been shown to reduce credentialism in venture capital by providing an alternative source of signal for investors.

If founder-first thinking were simply a systematic pathology, the firms most committed to it should be among the worst performers. Instead, they are among the best.

The answer is actually fairly straightforward. When great investors say “founders first,” they mean something significantly more sophisticated than the skin-deep interpretation of the industry at-large.

The Great Man Fallacy

The desire to reduce founder success to a checklist of predictable attributes is a modern manifestation of the Great Man Theory; the belief that history is shaped by exceptional individuals with innate greatness, ignoring how success itself forges those qualities.

“A successful company with a strong record of performance? The leader appears to be visionary, charismatic, with strong communication skills. A company that has suffered a downturn? The same leader appears to be hesitant, misguided, or even arrogant.”

The Halo Effect, by Phil Rosenzweig

For example, industrialists like Elon Musk have shaped investor expectations about hard tech founders, through the many stories about his multi-domain hyperfluency, discipline and assuredness. So, this is what they look for in first-time founders, not realising that Musk developed these attributes over time and they are denying others the opportunity to do the same.

Consider also Thiel’s investment in Mark Zuckerberg, a Harvard dropout. Today, it’s often cited as an example of Thiel’s ability to identify great founders early. Yet contemporary accounts reveal that Thiel was taken with Facebook itself, the early traction, and the specific way that Zuckerberg had chosen to frame the problem of online identity.

Would Thiel have recognised anything in Zuckerberg had he been building a flower-delivery startup? It is hard to imagine. The idea about how a university’s social network ought to function, and the particular form that Zuckerberg had already given it, was the magic that Thiel was looking for.

Indeed, at Andrew Ross Sorkin’s DealBook Conference, Peter Thiel was asked how he evaluated founders, and his answer matches the Facebook example.

“I don’t separate the ideas and the business strategy and the technology that much from the people. It’s all some sort of a complicated package deal.”

Peter Thiel, Co-founder of Founders Fund

He cannot, he says, assess the quality of a founder without assessing the quality of the idea that the founder is working on. He cannot assess the idea without understanding the ways in which the founder has shaped it. The two are inseperable.

Problems Worth Solving

A complementary argument has been developed by academia. In a 2022 paper published in the Journal of Business Venturing Design, Mattia Bianchi and Roberto Verganti at the Stockholm School of Economics and the Politecnico di Milano argue that entrepreneurship has been systematically misunderstood as an exercise in problem solving, when in fact it is primarily an exercise in problem finding.

The founder’s most important creative act, in their framing, is the identification and framing of a problem worth solving. Everything else, the pitch deck, the go-to-market plan, the product road map, follows from the quality of that initial framing.

“Seeing problem finding as an act of design and not just discovery expands the potential impact of design practice from the creative generation of solutions to that of problems. Speculatively reframing problems is an additional lever for breakthrough innovation, since an unconventional problem formulation can open up unexpected solution paths.”

Entrepreneurs as Designers of Problems Worth Solving, by Bianchi and Verganti

If this framing is correct, the dichotomy at the heart of the jockey-versus-horse debate is false, and a founder should be assessed through the problem they have chosen to attack and the particular frame they have used to understand it. The idea cannot be assessed in isolation either, as it reflects the material expression of a founder’s beliefs about what the world will look like in ten years. Each illuminates the other, and any investor who claims to assess them separately is certainly doing neither well.

“By Their Fruits You Shall Know Them”

This combination approach is well framed by Nabeel Hyatt of Spark Capital. Asked how he distinguishes founders who are genuine operators from those who just check a lot of superficial boxes, his answer is surprisingly straightforward.

“The way we separate hucksters who do a good pitch from people who are real executors is to look at the thing that comes out of their hands. I’ve never evaluated a company by looking at the product or using the website and being like ‘this person should have a $15M check’. You look at the product, and you try to learn about the humans behind the product by evaluating the product.”

Nabeel Hyatt, GP at Spark Capital

The product is the manifestation of the founder’s ambition, and a deep reflection of their judgement, their priorities, and of the problem they have chosen to solve.

An investor who says “I invest in people”, and has not looked carefully at the product, is either investing in shallow patterns or in charm and charisma. There are precisely the habits which reliably produce predictably bad investments.

Sam Altman, sharing his application-screening heuristics at a 2016 Khosla Ventures summit with Keith Rabois, put the same point in slightly different language:

“The hardest trait we’re looking for is determination. There are a few other topics in the middle we look at: Clarity of vision, communication skills, and the non-obvious brilliance of the idea we look at very hard. These are ones that you can’t always be right on, but you can usually get quite a lot of data, and unlike determination they’re not quite as difficult to figure out.”

Sam Altman, former President of Y Combinator

He does not say the brilliance of the founder. He says the brilliance of the idea, qualified as “non-obvious”, which suggests the founder has picked a novel problem. And clarity of vision, which suggests looking at how they perceive and articulate that problem. And, of course, the determination they direct at that process.

He is talking, in Bianchi and Verganti’s language, about the founder as a designer of problems worth solving.

The Entire Ocean, In a Drop

There are two things an investor can mean when they say that they invest in people.

The first is the belief that attributes such as pedigree, biography, charisma, and past fundraising success carry more signal than what a founder has chosen to dedicate their time to. Essentially, that founders are a fungible commodity that can be stack-ranked. This is the version that Davenport’s data most directly contradicts.

The second, much rarer version is the belief that the subject being evaluated is a unique alchemical mix of people and ideas. It is the investor’s job to assemble a complete picture; the choice of problem, the form of the solution, and the character of the team. Only then, can they fully perceive the opportunity in front of them.

The two are easily confused because they use the same vocabulary. Both are expressed in the language of backing people and celebrating human potential. The first is lazy and well-rewarded by the norms of the industry. The second is hard, and often misunderstood, but it is clearly the path to better quality investments.

The argument is not that investors should abandon qualitative team analysis and return to Perkins and Valentine. The conclusion is simply that the team cannot be usefully assessed without the context of what they are working on, and the attempt to do so is where investors fall into problematic pattern matching.

This is why the atomic unit of entrepreneurship is not the founder nor the idea, but the unity of both. The venture capitalist must stand far enough back to see both at the same time, and to assess them as a single entity.

Rather than the tired question of jockey or horse, the investor’s job is to recognise the centaur.

N.B. A paper from 2009 offers empirical justification for focusing more on ideas when evaluating companies, by analysing how many had changed either leadership team or core product by the point of IPO. However, this covers a period when VCs regularly brought in new executives prior to listing, and no longer seems relevant.

Run your venture firm from your phone with Odin.

In this episode, he joins Paddy Ryan to discuss chip investing, photonics, and why the next decade may be built in the physical world.

Watch the full conversation on YouTube

“I’ve always had an affinity for the fringe. I go with what I feel is the most important problem being solved by the most impressive people who are being neglected for one reason or another. I try to find that blank space.”

Adam Draper, Managing Partner and Co-Founder, Boost VC

Sometime around 2011, Adam Draper and Brayton Williams found themselves at a P.F. Chang’s in the Stanford Shopping Center, becoming friends over a topic of conversation that has always animated Silicon Valley, “What if we tried something new?”

The two initially worked together on a startup attempting to solve private company liquidity through a secondary exchange. While that idea never quite took off, they formed a strong partnership which, over the next nine years, evolved into one of the most impactful, contrarian, and ultimately successful pre-seed firms of this generation.

Boost VC’s Fund I, a $38 million vehicle raised in 2016, has already returned 4.61x to their LPs. In an era where most funds struggle to return more than 2x on horizons that stretch out to 16 years, that’s a remarkable achievement.

The firms success is rooted in a particular combination of people, philosophy, and sheer audacity. As many of the largest institutions in venture capital were straying down the path of herd behavior, Draper and Williams saw an open season on outliers.

An Insider and an Outsider

Central to the Boost VC lore is Adam Draper’s relationship with the industry. His great-grandfather, William Henry Draper Jr., co-founded Draper, Gaither and Anderson in 1959, one of the first venture capital firms on the West Coast and the originator of the limited partnership structure that the industry uses to this day. His grandfather institutionalised venture capital through Sutter Hill Ventures before helping launch the first international fund in India. His father, Tim Draper, globalised it through Draper Fisher Jurvetson, seeding nearly 700 funds across the world.

Adam Draper dropped out of UCLA and co-founded a company that he eventually sold for parts. He spent years in the wilderness of Silicon Valley trying to figure out whether his real gift was building or backing. By his own account, the answer only became clear when he started making angel investments in companies nobody else would touch. These early investments included Coinbase, when most people had barely heard of Bitcoin; Benchling, which became a $6 billion life science platform; and Amplitude, the analytics tool that went public in 2021.

“While most investors start out by making terrible investments because we don’t know anything, five of Adam’s first twenty-five pre-seed bets are unicorns already.”

Elizabeth Yin, GP at Hustle Fund

But Draper’s pedigree is only half the story. Brayton Williams arrived at their partnership with almost the opposite background. A relative outsider, a Cal Poly graduate in finance and econ, Williams had no real notion of how venture capital operated, or the industry’s history. This turned out to be a superpower, as he dodged decades of accumulated bias and dogma that many career investors naturally absorb.

The partnership between Draper and Williams works because of this contrast. Draper brings the fourth-generation instinct for greatness, and a risk appetite to match. Williams brings a blank slate; a vision for how things should be done, rather than how they have always been done. Their take on venture capital is a revival of the industry’s historical contrarianism combined with a more contemporary portfolio strategy, and they’re exactly the people to execute on it.

An Appetite for Risk

Boost VC’s investment model runs contrary to much of the common wisdom about ownership targets. Where most firms concentrate capital into 20–30 companies, Boost VC disperses their investments across a large portfolio of around 120.

The thesis is that at the earliest stages, the portfolio maths of venture capital demands breadth. If 2–3% of pre-seed investments eventually become unicorns, a fund of twenty companies might miss completely, while a fund of 125 companies could hit two to four. The strategy recognises that most venture capitalists are undone by their focus on “picking”, through overconfidence in their ability or by letting insecurity push them toward consensus. Diversification solves this problem.

This is a perspective supported by research. Larger portfolios enable a greater appetite for idiosyncratic risk, as investors are more willing to make investments that may fail completely. This enlarged risk appetite captures a wider range of outcomes and generally leads to outperformance.

Essentially, by making more investments, Boost VC is able to lean further into moonshots, rather than safer feeling iterative ideas.

“Because fund diversification also affects fund risk, risk-averse fund managers are tempted to invest in companies with higher idiosyncratic (i.e., company-specific) risk when pursuing a strong diversification strategy. Doing so eventually leads to higher realized fund performance on average, since investments in riskier ventures only make sense if they come with better prospects of high returns.”

Diversification, Risk, and Returns in Venture Capital, by Axel Buchner, Abdulkadir Mohamed and Armin Schwienbacher

This approach also requires that Boost VC is able to move quickly on investments. Williams has spoken candidly about the firm’s investment approach, which allows them to reach conviction with a minimal amount of time wasted.

“You’re still typically taking three calls per founder, but we always generally believe more diligence does not get you to a better decision. You’re basically just judging, can this person do what they say they’re going to do, and are they committed for the next 20 years to do that thing?”

Brayton Williams, Co-Founder, Boost VC

They recognise that the signal in a pre-seed investment is human. Nine out of ten of Boost VC’s deals come via referrals from trusted sources, and by the time a founder sits across from Draper or Williams, the firm has already run a parallel diligence process through that network. The goal of the meeting is to ascertain character, conviction and the magnitude of the idea.

Their diversified strategy also helps the firm manage systematic risk. As Boost VC spans sectors including cryptocurrency, virtual reality, artificial intelligence, space, biotech, and nuclear energy, a collapse in any one corner of the market is unlikely to harm the overall fund. A downturn in crypto doesn’t touch a space company. A regulatory crisis in biotech has no bearing on a VR startup.

Another benefit of their strategy is a price advantage. Being the first money into extremely idiosyncratic ideas means they are typically investing at valuations between $3 million and $7 million, well below the median for pre-seed. While other investors complain about prices creeping upward due to increasing concentration and consensus, Boost VC’s average ownership for $500,000 has gradually increased.

Early Crypto Returns

The companies in Boost VC’s first fund cover a spectrum of technological imagination from the mid-2010s, some of which has become reality and some of which remains aspirationally strange. It certainly isn’t the more typical dull cocktail of enterprise SaaS solutions of that era.

Their most consequential relationship, with crypto, began with Adam Draper’s investment into Coinbase in 2012 and was followed by the Fund I investment in 2016. When Coinbase went public via direct listing in 2021, the company reached a market cap of over $85 billion. That track record helped establish a reputation for truly understanding crypto before any other institutional investor. In fact, Boost VC was among the first funds to hold Bitcoin as a portfolio investment.

“Boost VC was one of the first funds, if not the first, to buy bitcoins as portfolio investments. That enabled them to return 1.5x the fund during Venture Capital’s worst year.”

Augustin Sayer, GP at OVNI Capital

Within Fund I’s direct portfolio, the returns have been driven by a cluster of winners that demonstrate the significance of their early crypto conviction:

• Polychain Capital, the crypto hedge fund founded by Coinbase’s first employee Olaf Carlson-Wee, returned more than 90x to Boost VC’s limited partners despite the fact that many investors had passed because its structure was not a conventional Delaware C-corp.

• Etherscan, the Ethereum blockchain explorer that became the default transparency layer for the entire ecosystem, turned into one of the most important blockchain infrastructure plays in the portfolio.

• Decentraland, the blockchain-based virtual world that anticipated the metaverse wave years before the term entered the mainstream, and another example of a category that the rest of the market would later pile into.

• Polkadot, the blockchain interoperability protocol developed by Ethereum co-founder Gavin Wood, another major infrastructure project in the crypto ecosystem.

• 21Shares, the Swiss crypto exchange-traded product issuer, was acquired in a deal that pushed it past a $1 billion valuation in 2025.

Then there is the recent phenomenon of Moltbook. The story of Matt Schlicht, the founder Boost VC backed through his earlier company Octane AI and who later pivoted to build an AI agent social network.

Schlicht grew up at a Waldorf school where he never touched a computer. When he eventually got a laptop as a teenager he became so obsessed with building products that his school expelled him for it. He hacked the school’s systems that summer, helped them fix the vulnerabilities, and they let him back to graduate. He skipped college, moved to Silicon Valley at nineteen, and by 2016 was building crypto social products and chatbot platforms.

In March 2026, his company Moltbook was acquired by Meta, with Schlicht and his team joining Meta’s Superintelligence Labs. This decade-long story of overnight success pushed Fund I’s DPI over 4.61x.

“I love the Boost VC team. The support they have given me, no matter the situation, is so strong and rare. They were the first check in my companies. I will work with them until I die.”

Matt Schlicht, Founder, Moltbook and Octane AI

Schlicht’s arc is not unusual for Boost VC’s portfolio; Williams’s instinct has always been to look for people who are constitutionally incapable of doing anything other than building, regardless of whether the world is ready for what they are building.

The portfolio also carries a substantial set of positions that have not yet reached their exit, but where the trajectory suggests the fund’s total value to paid-in capital of 7.04x still has plenty of room to run.

• Ledger, the French hardware cryptocurrency wallet company, has reached valuations of $1.5 billion with IPO discussions mooted at significantly higher figures.

• Unstoppable Domains, the blockchain domain name provider, has grown into one of the most recognisable brands in the Web3 identity space.

• Volley, the voice game platform, has built a substantial audience across smart speaker devices and represents an early bet on ambient computing as a distinct category.

• FitXR, the virtual reality fitness company, has raised over $18 million in follow-on capital and continues to grow as VR hardware becomes more mainstream.

• Ripio, the Latin American crypto platform, has expanded across Argentina, Brazil, and beyond, serving a region where the use case for digital currencies has always been more urgent than the Silicon Valley imagination of it.

• VennBio and Shake Labs represent the fund’s deeper investments in life science and fintech infrastructure respectively, both still in the compounding phase that Boost, by its own design, is built to be patient about.

Frontier Science

The companies that attract the most attention are not always the ones that generated the largest financial returns. Over the last decade, Boost VC has backed companies working on problems that more conservative venture firms would not put in a slide deck. De-extinction. Telekinesis-adjacent brain-computer interface research. Personal flight. Space observation networks designed not to watch Earth from orbit but to watch the cosmos from a fixed platform. VR platforms, hardware wallets, and AI voice tools when all three categories were considered too early, too strange, or too difficult to commercialise.

This is a consequence of the strategy and incentives that Draper and Williams have designed. Because they make small investments at very low valuations, they are free from the tyranny of consensus. A $500,000 investment in a company working on something that might take fifteen years to mature does not break a fund. If the technology proves out, the returns will be transformative. If it doesn’t, the fund absorbs the loss and the other 120 companies carry on. Their strategy allows them to back the ideas that other funds cannot justify to their limited partners.

Colossal Biosciences, the de-extinction company whose recent news includes the revival of the dire wolf and a $10 billion valuation, is perhaps the most vivid illustration of what happens when that philosophy catches a tailwind. When Boost VC first backed the company, the idea of bringing an extinct species back from the dead using CRISPR and ancient DNA was the kind of thing that appeared in science fiction. The company’s scientific claims have attracted significant scrutiny. But the point is not whether every bet pays off. The point is that the firm was in the room, with conviction, before anyone else was willing to be.

“Boost VC makes bold bets on the frontier. Make science fiction real.”

Brian Armstrong, CEO and Co-Founder, Coinbase

If you include investments made prior to Fund I, Boost VC’s core thesis has now been tested across thirteen years, four funds, and more than 500 companies.

The thesis holds that transformative technologies are systematically underpriced at inception because most investors cannot evaluate what they do not yet understand. Bitcoin in 2013. Small modular nuclear reactors in 2020, when every investor wanted SaaS and remote-first businesses. Voice AI in 2015, when speech recognition was considered a solved problem that belonged to the large technology companies. The founders who front-run the market in these categories might look like eccentrics to the average venture capitalist. To an investor like Boost VC, that eccentricity is a strong positive signal.

Run your investment firm from your phone with Odin.

Entrepreneurial Idealism

Boost VC’s success is built on a specific and unusual relationship with a certain kind of founder.

Most venture capital is designed, whether its practitioners acknowledge it or not, to select for founders who already speak the lingua franca of the industry. The pitch deck structure, the ritual dance for setting terms, the ability to turn an irrational ambition into a straight-laced investment memo. These conventions tend to advantage founders with existing access to networks and institutions, who have seen how the game is played and know the rules.

Draper and Williams are less interested in founders that cater to the venture capital industry, and more interested in how they can cater to founders.

Matt Gould, the CEO of Unstoppable Domains, a Boost VC portfolio company, described the firm in terms that capture something real about its character:

“Boost VC is one of those rare VCs with as strong a conviction in technology’s ability to shape the future as the founders of the companies in their portfolio. Brayton and Adam are both dedicated to changing the world and that passion comes through in their support and advice. Founding a company is hard. Having great investors makes it easier. You can’t get any better than Boost.”

Matt Gould, CEO of Unstoppable Domains

The dynamic at Boost VC is unusual because they arrive already disposed to believe that hard, strange, long-timeline ideas are worth pursuing. The founder’s job is not to overcome scepticism but to demonstrate that they are the right person to pursue this specific version of the impossible. This is a stark contrast to the typical interaction with venture capitalists, and it turns out that matters enormously to the most ambitious founders.

The ability to identify people with the combination of technical capability, commercial instinct, and the psychological resilience required to build something new cannot be taught with a checklist of attributes or credentials. It comes from volume, attention, and a genuine openness to being surprised. Recognising the absolute idiosyncrasy of greatness.

Draper describes his investment process in terms of looking for evidence that someone is “confident in themselves enough to build a business,” able to be uncomfortable every day, and capable of learning the skills they do not yet have.

Drivers of Performance

Boost VC’s Fund I will mainly be measured by its performance, which (so far) is 4.61x DPI and 7.04x TVPI, on the original $38 million. That result places the fund among the best of its vintage. For context, the top decile performance for venture capital funds in the 2016 vintage stood at 1.66x DPI and 5.4x TVPI, according to 2025 Cambridge Associates benchmarks.

In 2025 alone, the firm returned $42 million to limited partners, predominantly from Fund I, while four portfolio companies across all funds reached billion-dollar valuations, including 21Shares (acquired), K2Space, Radiant Nuclear, and one other undisclosed company.

The Radiant Nuclear offers another great example of Boost VC’s strategy at work. When a SpaceX engineer named Doug Bernauer came to them in 2020 with a plan to build small modular nuclear reactors, Boost VC made the investment in a single twenty-minute meeting. Nuclear energy was, at that moment, considered a dead category by most of the technology investment community. Capital was flowing into software, remote work tools, and anything that did not require physical infrastructure, long regulatory timelines, and the specific stigma that has surrounded nuclear since the accidents at Three Mile Island, Chernobyl, and Fukushima. Radiant is now valued at over $1.8 billion, having recently closed a $300 million Series D with Boost VC co-leading the round.

Adam Draper reflected on that in a note that captures the spirit of the whole enterprise:

“If we believe something that the rest of the investment world doesn’t, that is a scarce resource and we must act.”

Adam Draper, Managing Partner, Boost VC

Boost VC’s performance is driven by their willingness to sit in a category alone, potentially for many years, waiting for the rest of the market to catch up with them.

The Next Generation

Boost VC’s Fund IV, a $87 million vehicle that they began investing out of in November 2024, is an extension of everything described above. The thesis has not changed, and neither has their investment strategy. What has changed is the breadth and ambition of the technologies being built as dynamism grips the market. Suddenly, “hard problems” are back in vogue and Draper and Williams are in a great position to capitalise on that.

The fund is only partially deployed, and most of its companies are operating in stealth or at a stage of development where they have little public profile. The handful that have begun to emerge say a lot about how years of frontier investing have sharpened the firm’s instincts.

Icarus is perhaps the most immediately legible of the Fund IV investments. The company, which emerged from Y Combinator’s 2025 cohort, is building solar-powered stratospheric drones designed to operate at roughly 60,000 feet for weeks or months at a time. The pitch is that these platforms could serve as persistent, high-altitude infrastructure for communications, imagery, and data relay: effectively cell towers floating at the edge of the atmosphere.

FeltSense occupies a different register of ambition entirely. The San Francisco company, founded in January 2025, is building AI agents designed to function as autonomous founders; systems that can ideate, build products, launch them, and scale companies without human direction. It claims to be capable of running the equivalent of a Y Combinator batch entirely within AI, at speed and at volume. The company raised a $5.1 million seed round in February 2026, led by Draper Associates, with Boost co-investing alongside Precursor Ventures and Liquid 2 Ventures.

Then there is American Housing Corporation. Based in Austin, Texas, the company is building factory-produced modular homes shipped in standard containers and assembled on-site in days, with the explicit goal of industrialising the solution to the United States housing shortage. The company built and publicly showcased its first home within just over a year of founding. When venture capitalists had largely abandoned the physical world for software, a company trying to re-engineer how homes are manufactured and delivered is precisely the kind of heterodox idea that Boost is drawn to.

Serova Bio sits at the opposite extreme of legibility. The company is developing personalised cancer vaccines powered by machine learning, designed not merely to target the mutations currently visible in a tumour but to anticipate what those mutations will become as the cancer evolves. It describes itself as “building the future quietly,” and has almost no public footprint beyond its website.

Finally, there is Capable, who are working on what might be described as Ozempic for sleep, a therapeutic targeting sleep disorders through mechanisms related to the GLP-1 pathway that underpins the current generation of weight-loss drugs.

The Fund IV portfolio helps to illustrate a deep truth that might be lost in the survivor bias of looking back on Fund I outcomes; these companies are not obviously going to work. Several are operating in categories where the scientific or commercial case is not yet clear. Some of them will fail. The ones that succeed may well create entirely new categories and markets. This is exactly what should be expected in an early-stage venture capital portfolio at the outset, and if an investor feels like they need more certainty then they may well be in the wrong business.

“This is a game of endurance and optimism.”

Adam Draper, Managing Partner, Boost VC

Manifest Destiny

The success of Boost VC is evidence that the technologies which will shape the future rarely look obvious at their inception. The most transformative ideas are sitting in a conference room somewhere, being pitched by someone who is being told no by almost everyone they meet, until they find their first believer.

Boost VC built a fund to be that first believer, at scale, across sectors, over time. Their strategy was designed to fill a gap that institutionalised venture capital had abandoned. It turns out the gap was enormous, and that the founders waiting on the other side of it were building some of the most important companies of their time.

The world will produce more founders like Matt Schlicht, who builds obsessively without permission, and more companies like Deepgram and Radiant Nuclear, which seemed too early and too strange until suddenly they were not.

Fund I demonstrated that being the first believer in those founders, unconditionally and at volume, might be the most rational investment strategy in the venture business.

Run your investment firm from your phone with Odin.

“The fact is that the amount of money start-ups raise in their seed and Series A rounds is inversely correlated with success. Yes, I mean that. Less money raised leads to more success. That is the data I stare at all the time.”

Fred Wilson, of Union Square Ventures

As venture capital moves through cycles of hot and cold markets, investors often find themselves wondering whether they should keep pace with the market. This is particularly true in categories where rounds are the most competitive and prices are at their highest.

Essentially, is “paying up” for access to hot companies worthwhile? Can participating in outsized rounds be justified with the “kingmaking” logic, that a exceptionally well capitalised startup has a better shot at success?

On one hand, venture capital is a contrarian sport, where returns are won by making investments in categories that are overlooked and underfunded. On the other hand, there’s the temptation to optimise for momentum and downstream investment, by choosing categories with greater interest and higher velocity.

Data from four sources, covering various parts of the startup life-cycle, suggest that the disciplined investor probably comes out ahead; high prices are often a red flag. This overlaps with a variety of principles about the psychology of resources management, optionality, and risk management.

The Principle of Least Commitment

One of the major features of young startups is their flexibility and adaptability. Founders must experiment in order to find true product:market fit, which often requires a major pivot of the initial vision. In fact, there’s even evidence that executing a pivot may be a positive indicator for future success, reflecting a founder’s ability to exploit this flexibility.

This is where greed at the earliest stages can become problematic. If startups are able to raise an irrational amount of capital due to hot market effects, they often end up overinvesting in development or hiring in a way which makes them brittle.

“The more you raise, the more you spend, and spending a lot of money can be disastrous for an early stage startup. Spending a lot makes it harder to become profitable, and perhaps even worse, it makes you more rigid, because the main way to spend money is people, and the more people you have, the harder it is to change directions.”

How to Raise Money, by Paul Graham

This effect was confirmed in a comprehensive study by Startup Genome, which spun out of a wider project analysing the major causes of failure in 3,200 early stage companies. The phenomenon of raising too much capital and growing certain dimensions more quickly than others (e.g. overhiring without the appropriate revenue growth), referred to as “Premature Scaling”, was present in 70% of the cases they examined.

This finding was replicated in another report out of Zayed University, looking at the relationship between the amount of capital raised and the ultimate valuation of the company.

“The findings suggest that there is a U-shaped relationship between the amount of capital raised and post-money valuation, indicating that while capital funding has an overall positive effect on market valuation, raising too much capital has a negative impact.”

The Impact of Funding on Market Valuation in Technology StartUp Firms, by Amril Nazir and Dina Tbaishat

The conclusion here is simply that startups shouldn’t raise too little, as they must be able to run experiments and build what they need to hit the next milestone. Equally, they shouldn’t raise too much, as they need to remain disciplined and flexible.

It’s tempting to believe that you could keep that money in the bank, de-risking the company, but the truth is that you’ll be under pressure to spend on growth, whether you are ready or not.

Indeed, risk increases with capital raised when that capital is treated as a debt against future expectations. It is far more beneficial to raise minimally, and retain maximum optionality at the earliest stages. In AI research circles, this concept is referred to as the principle of least commitment.

“The least commitment principle is one of several strategies people use when they make plans, such as preparing for a day’s outing. The principle gained currency in the 1980s in the early days of artificial intelligence research. It simply means to prioritise tasks in such a way that you keep certain decisions about resources and timings as open as you can for as long as possible.

In computational terms the principle means the sequence of tasks you want to plan for has a large number of variables with interacting and conflicting constraints. Keep as many variables open, vague and with as wide a range of values as possible. The principle assumes that the creation of a plan is a search process. At some stage in the search you are bound to strike certain constraints that decide parts of the task sequence. The less constrained tasks simply fall into line.”

Least Commitment Principle, by Richard Coyne

Raise and deploy righteous capital with Odin; fund administration for builders on the frontier.

The Principle of Indifference

“There will be certain points of time when everything collides together and reaches critical mass around a new concept or a new thing that ends up being hugely relevant to a high percentage of people or businesses. But it’s really really hard to predict those. I don’t believe anyone can.”

Marc Andreessen, Co-Founder of Andreessen Horowitz

It’s tempting to believe that the best outcomes are known, long before they make it to an exit. In every cycle, category winners are called based on fundraising signals and rapid markups. However, these companies may end up just being early pioneers, like the internet search engines of the 90s, that will be passed by “fast followers” further down the line.

There’s also a problem of selection bias, where investors pile into high-performing companies and drive up the price, leading many to believe that a high price is indicative of performance. For this to make sense, round size growth would have to correlate with probability of future success. Otherwise, investors are just bidding-up the price and eroding their future returns.

So, does early round size have a meaningful correlation with success? According to analysis by Santé, looking at data from 1,980 exits, it would appear not.

None of this should be surprising to anyone with a good grasp on the risk that is being managed. If outcomes were so predictable, where paying more would produce a meaningfully greater likelihood of success, venture capital would have a radically different distribution of outcomes. In fact, the venture industry as we know it would be largely unnecessary.

Simplified, this is the principle of indifference. If you cannot honestly judge the likelihood of different outcomes then you must apply equal probabilities to all. In a venture capital context, that would imply all seed investments should be priced similarly, because price should follow probability. Unfortunately, investors often behave as if probability follows price.

“The principle of indifference, also known as the principle of insufficient reason, is attributed to Jacob Bernoulli, and sometimes to Laplace. Simply stated, it suggests that if there are n possible outcomes and there is no reason to view one as more likely than another, then each should be assigned a probability of 1/n. Quite appropriate for games of chance, in which dice are rolled or cards shuffled, the principle has also been referred to as the “classical” approach to probability assignments.”

The Principle of Indifference, by Cambridge University Press

The Paradox of Plenty

The problem of overcapitalisation is consistent all the way through the life of a company, not just at early stages. Indeed, research out of the University of Beijing’s School of Finance indicates, there’s a similar convexity in pre-IPO growth.

“Our results suggest that pre-IPO growth only improves the post-IPO long-run performance to a saturation point, above which pre-IPO growth negatively influences the post-IPO long-run performance. In other words, firms with lower or higher pre-IPO growth do not achieve better post-IPO long-run performance. Instead, it is the middle-level growth prior to IPO that is significantly related to the best post-IPO long-run performance.”

Pre-IPO growth, venture capital, and the long-run performance of IPOs, by Jiangjing Que and Xueyong Zhang

Essentially, there is a minimum scale at which IPOs perform well and provide a solid foundation for future growth. However, there’s also a point at which an IPO might be oversized, which is reflected in a slower growth and a less appealing opportunity for public market investors.

This appears to be connected to the amount of venture capital raised by large firms, leading to a similar problem of brittle overinvestment to that studied in early-stage companies.

“We find that the VC backed IPOs underperform non-VC backed IPOs, especially when firms have high excess cash, and propose that the higher excess cash retained in VC backed firms at IPO date may result in wasteful investment. Post-IPO real investment also confirms this notion.”

Do venture capitalists improve the operating performance of IPOs?, by Hung-Kun Chen and Woan-lih Liang

Again, this stems from a problem of selection bias, where high-performing firms often have more revenue, but more revenue does not necessarily indicate high performance. Private market investors misunderstand this, behaving as if scale supersedes other critical factors like defensibility, growth rate and profitability, and so overoptimise around scale (which has conveniently enabled larger funds).

Indeed, there’s relatively little correlation between implied ARR and the resulting ARR multiple at IPO, which would reflect a “scale premium”, and what little there is can be explained by selection bias.

So when is the right time to go public? The answer is a trade-off between scale and growth. You need to be large enough that the costs are manageable, but small enough that your growth rate remains appealing. Additionally, you need to at least indicate the potential for improving economics over time, which is often where cash incinerating venture-backed companies struggle.

“Be mindful that the public markets want fast (and durable) growth rates, improving margins, and a path to profitability (not just free cash flow positive but GAAP operating profit eventually) i.e. ‘Can you make money?’

While companies don’t need to be at $1B of ARR or revenue, you should have at least $250M of ARR or $200M in trailing revenue. Are you growing revenue 30-40%+ year-over-year with improving free cash flow margins? Is your growth rate durable? Are you a Rule of 40 business?”

2024: When to IPO in the Age of Uncertainty, by Alex Clayton and Cathy Choi

In short, larger growth rounds will produce inflated exits, but they happen much later in a companies life and appear to lead to negative consequences post-IPO, including slower innovation, higher rates of failure, more fraud and weaker performance.

This is emphasised by the fact that aggressive post-lockup selloffs are associated with overvaluation, and this appears to be a particular issue for venture-backed IPOs. More than any other category of company, insiders in venture-backed companies are keen to exit their position as soon as they possibly can.

“The authors show that the IPO lockup period price reduction is strongly related to overvaluation. Zero-investment portfolios long in the lowest overvaluation quintile and short in the highest overvaluation quintile of IPO firms have positive significant returns.”

IPO overvaluation and returns prior to lockup expiration, by K. Stephen Haggard and Yaoyi Xi

The performance problems connected to overfunding relate to a broad body of research on the paradox of plenty, which looks at how resource availability changes outcomes at all levels of application, from the individual to the state.

To repeat the theme of this article, this research demonstrates a convex relationship where it is helpful to be sufficiently resourced that you can handle risk, but not so well-resourced that you become complacent.

“This article suggests that there is an inverse U-shaped relationship between slack and innovation in organizations; both too much and too little slack may be detrimental to innovation. Two related mechanisms governing this relationship are proposed: Slack fosters greater experimentation but also diminishing discipline over innovative projects, resulting in the hypothesized curvilinear relationship. Comprehensive worldwide data on 264 functional departments of two multinational corporations support the prediction.”

Is Slack Good or Bad for Innovation?, by Nitin Nohria and Ranjay Gulati

Cui bono?

Given all of the above, demonstrating that “more” is categorically not “better”, who actually benefits from kingmaking?

In practice, it’s a strategy that consumes vast sums of capital, accelerates price inflation of private companies, and enables oversized IPOs.

All of this would imply that the beneficiaries of kingmaking are the scaled venture capital firms, whose central challenge is rapidly deploying large pools of capital and generating attractive performance metrics. They’re also often the firms who promote the legitimacy of this strategy in relation to their favoured portfolio companies.

At best, kingmaking is a lazy narrative device designed to scare off competition. At worst, it’s a shallow justification for larger funds, fabricated performance and less disciplined investing.

Importantly, it serves these purposes at a cost to the long-term success of founders and their companies, which is unforgivable.

The best path, for founders who are building a generational company, is to work with investors who understand the existential importance of capital efficiency.

“I think frugality drives innovation, just like other constraints do. One of the only ways to get out of a tight box is to invent your way out.”

Jeff Bezos, Founder of Amazon

Raise and deploy righteous capital with Odin; fund administration for builders on the frontier.

Two exciting product announcements I thought you might find interesting:

From today, you can set up a multi-currency, institutional Delaware Limited Partnership structure on Odin, launch the same day and invest in any asset, anywhere in the world

• More information here.

• Customers in the UK and Europe now have a fast to launch (deals live in <2 hours), off the shelf limited partnership structure familiar to investors globally. It is fully tax transparent, works in multiple currencies (USD, EUR, GBP), with banking, accounting, K1, FATCA / CRS and tax reporting built in. For UK/EU customers it’s an AIF and fully compatible with AIFMD (with some caveats on compliance).

• Deal by deal investors, founders and emerging managers in the USA can also now benefit from a US structure their investors will be more familiar with (similar to structures offered by AngelList, Carta, etc), via a platform that understands the nuances of global deals & LPs. Plus they get an epic events space in New York (see below) to engage their investor community, AND (in my unbiased opinion) a much better looking team supporting them.

• Our customers everywhere can, of course, continue to use the UK Bare Trust structure they’re familiar with for deal by deal SPVs. It’s still generally the best setup for smaller syndicates unless they’re US-based. Depending on your location & circumstances, you may be able to use our Delaware structures too - we can discuss and advise based on your circumstances.

• Cost-wise, our Delaware SPVs will start from $6k per deal for simple US deals led by US sponsors. Proceeds depends on deal size and complexity.

• If you’re interested in setting up a Delaware LP structure, drop me an email or click the link below to speak to someone from Odin!

Speak to our sales team

Any fund manager or syndicate lead who is an Odin customer can now host investor events & dinners - free of charge- in the beautiful spaces we have in New York and London. Non-customers in our extended community can also get in touch.

If it’s a mission-aligned event or dinner, we’re open-minded.

Host an event with Odin

When we talk to emerging managers in our network, there are two things, above all else, that they care about:

1. Access to [the right] capital

2. Investing in great deals that deliver strong returns

An often-overlooked way to solve these two problems is to simply get the right people in a room together.

Venture is, at its core, a people business. There is extremely scant data to support your decisions, so instead you run somewhat on intuition. You invest in and with people you trust.

Being able to get to know someone in a relaxed setting, and getting a clear read on what you think of them - are they smart, do you have shared values, are they driven, etc. - is fundamental to making good investment decisions as an LP, and to building the trust that leads to transactions as a GP.

The serendipity of community & networks is also vital to innovation itself. SF has become the beating heart of technological innovation precisely because it has a very high density of talented, creative people who meet in-person. This creates increased surface area for great ideas & individuals to bounce off each other - and for investors to fund what they build. Sparks fly, fires blaze.

This is what our events announcement is all about. Some details:

1. My cofounder Mary is going to be spending April, May & June in New York, joined by Emily - our head of venture operations - from May.
   We’ve got a beautiful apartment in Manhattan, and we’ll be co-hosting dinners, poker nights and other events with investors and founders building cool stuff.
   Obviously, we are new to the US of A, so we are keen to speak to people plugged into the NY ecosystem. Don’t be shy.

2. I’ll be holding the fort in London.
   I believe the youth are calling it Londonmaxxing?
   We’ve recently moved into a fantastic new office a stone’s throw from Covent Garden & Tottenham Court Road. We’ve got a dedicated space for events. We’ll support in hosting investor meetups, hackathons, dinners and more.

So if you’re looking to get your community of investors or builders in a room with other interesting people, give us a shout or fill out this form.

Thank you for reading!

We will return to our regularly scheduled programming Sunday with another banger from Dan. Have a great week.

Cheers

Paddy

Also, we rebranded! Hopefully you like it, feedback welcome.

Speak to our sales team

Host an event with Odin

“Storytelling is literally fundamental to every business. There is not a business that exists that there isn’t storytelling at the foundation and the core of it. Whether it is the moment in time in which a founder has an idea and needs to explain that idea to other people, or they need to go raise capital, or they have a product and need to find their go-to-market. Storytelling is at the core of business building.”

Jeffrey Katzenberg, Founding Partner at WndrCo

In the spring of 2016, Diego Berdakin, the founder of CloudKitchens, sat down for a poker game with Jeffrey Katzenberg. The legendary Hollywood executive had just sold DreamWorks Animation to Comcast for $3.8 billion and was restless for a new venture. He wanted to start a holding company, inspired by what Barry Diller had done with IAC. Berdakin knew someone who had been thinking along similar lines; a former Dropbox executive named Sujay Jaswa, who had recently left the file-sharing company after helping scale it to a multi-billion-dollar enterprise.

“Diego calls me and says, listen, I just played poker with Jeffrey Katzenberg. He just sold DreamWorks to Comcast. He wants to start a holding company. And I did. I wanted to talk to Jeffrey Katzenberg”

Sujay Jaswa, Founding Partner at WndrCo

He and Katzenberg had lunch the following Saturday. By August, they had agreed to build something together, although neither fully understood why it should work. As Jaswa put it: “I’m like a Silicon Valley software guy. You’re a Hollywood guy.”

Almost a decade later, WndrCo manages approximately $2.8 billion in assets and has backed 201 companies across three integrated strategies. It has produced at least eleven unicorns, a hit rate of 5.4% which places it in a similar performance category to Y Combinator.

The WndrCo portfolio includes Cursor, the AI coding tool whose annual recurring revenue surpassed $2 billion in March 2026; Figma, which completed a blockbuster IPO in July 2025; Databricks, valued at $134 billion after a February 2026 raise; and Abridge, the medical AI company that doubled its valuation in four months.

Yet WndrCo remains, by its own partners’ admission, one of the best venture firms in the Valley that few people talk about.

Understanding why requires understanding how two people from entirely different worlds came to share a view of what makes companies succeed, and how that perspective has quietly produced extraordinary results during what might be the most consequential technology wave since the internet.

Quotes from Katzenberg, Jaswa and Wang are primarily sourced from excellent conversations they have recorded with esteemed technology yappers, Molly O’Shea and Logan Bartlett.

Narrative Devices

Katzenberg’s professional life has been defined by the pursuit of new technology for the purpose of telling better stories. At Paramount, working under Barry Diller, he rose to head of the studio by the age of thirty. At Disney, he saw a short film from a small company called Pixar, a piece called Luxo Jr. featuring an animated desk lamp, and became obsessed with what computer-generated imagery could do for animation. He tried to hire the film’s director, John Lasseter. When that failed, he tried to buy Pixar from Steve Jobs for $50 million. Jobs refused. Katzenberg settled for a three-picture deal. The first film was Toy Story.

“So there’s a connection here in this, through every chapter of my storytelling career, you can actually go and see that one of the foundational elements of certainly what I think was my success is how aggressive I was in finding state-of-the-art tech for my storytellers.

Wherever I was, whatever studio is at whatever era I was in, I always look to what are the best tools to be able to realize the most exciting and interesting stories. [...] So I was constantly having these pilgrimages to Northern California, to the tech world where I partnered with Hewlett Packard and Andy Grove and Intel and Steve Jobs on Pixar.”

Jeffrey Katzenberg, Founding Partner at WndrCo

When he sold DreamWorks at the age of sixty-five, Katzenberg found himself at a crossroads. He remembers asking himself what he would do if he were twenty-three again.

“I’d go to Silicon Valley. That’s where fortune and fame is today.”

But, unlike some of Silicon Valley’s most notorious figures, Katzenberg was keenly aware of the limits of his competence. “I know what I know,” he has said repeatedly. “I actually know what I don’t know.”

So, for the journey he was about to embark on, he’d have to find the right partner.

The Outsiders

Jaswa, meanwhile, was having his own crisis of ambition. After leaving Dropbox, where he had overseen the company’s growth from a cold start to $600 million in annual recurring revenue in roughly five years, he assumed he would return to venture capital. He had come to Dropbox from NEA, one of the oldest and most respected venture firms in the country. But something had changed. “I just realised my heart wasn’t into it,” he said. “I like the building.”

A friend introduced him to a book called The Outsiders, by William Thorndike, which profiled eight unconventional CEOs who dramatically outperformed their peers through shrewd capital allocation. Jaswa read about John Malone, about the long-term compounding strategies of patient holding companies, and saw a path that combined his love of building with the kind of durable value creation that pure venture capital sometimes struggled to deliver.

“The Outsiders, by William Thorndike, Jr., is an outstanding book about CEOs who excelled at capital allocation. It has an insightful chapter on our director, Tom Murphy, overall the best business manager I’ve ever met.”

Berkshire Hathaway’s Annual Letter 2012, by Warren Buffet

“If you look at the history of tech, many of the great businesses didn’t start the way we all storytell them,” Jaswa observed. “IBM started as a private equity roll-up.”

When Jaswa and Katzenberg found each other through Berdakin’s introduction, they both saw an obviously complementary relationship. Katzenberg brought a lifetime of storytelling instinct, a network spanning Hollywood, Washington, and the Fortune 500, and a work ethic that borders on the superhuman (he sleeps five hours and fifteen minutes a night, a genetic trait he has had since adolescence). Jaswa brought deep product sensibility, operational experience at scale, and a first-principles approach to talent assessment inherited from his mentors at NEA, Dick Kramlich and Scott Sandell.

“Without Jeffrey, we’re just a bunch of people with reasonable resumes,” Jaswa told Logan Bartlett. “There’s like 500,000 people like us. There’s one person like him.”

Katzenberg was equally generous in his assessment of Jaswa. “If you take the ten ideas that have created all of the value of WndrCo in our nine years,” he said, gesturing toward his partner, “they’re all his.”

Are you an outsider looking to raise your own fund or SPV?

Deploy capital seamlessly with Odin

Humble Beginnings

WndrCo launched as a holding company, rather than a venture fund. The initial strategy was to acquire controlling stakes in underappreciated technology companies and operate them directly, applying the kind of hands-on product improvement and go-to-market refinement that both founders knew how to deliver.

Between 2016 and 2018, the firm incubated its first “Build” companies, rolling up sleeves on product roadmaps and operations in a manner more reminiscent of private equity than of Sand Hill Road.

The shift toward a venture capital operating model happened gradually. The firm’s first major outside investment was 1Password, the cybersecurity company, which WndrCo initially considered as a potential acquisition target before deciding to invest instead. That experience opened a door. The partners realised that many of the skills they used on the Build side, the product intuition, the storytelling coaching, the willingness to make hundreds of sales calls alongside founders, were equally valuable when deployed as support for independent companies they backed with minority stakes.

In 2024, WndrCo made it official, closing over $460 million across its first dedicated venture and seed funds. Today the firm operates across three distinct strategies.

• The Build arm acquires and operates companies outright, with eight companies launched since 2018 that collectively generate substantial revenue and free cash flow, providing downside protection and dry powder for new investments.

• The Venture arm leads or co-leads post-product-market-fit rounds, typically writing cheques in the range of $5 million to $15 million.

• A $50 million Seed fund makes smaller, non-lead bets of around $500,000 each, placing roughly fifteen bets per year on the next generation of entrepreneurs.

The firm’s flagship Build company, Aura, a cybersecurity platform, is a good example of the model. Partnered with WndrCo since 2018, Aura announced in February 2026 that it would acquire Qoria, a youth online safety company, in a deal that would create a combined business valued at approximately $2.1 billion in pre-money equity, with combined annual recurring revenue exceeding $300 million. WndrCo remains Aura’s largest shareholder.

The holding company model gives WndrCo an unusual advantage in a world where most venture firms are dependent on management fees and carried interest. The cash generated by the Build portfolio offers the firm a degree of financial independence that most of its peers lack, and allows the partners to be patient with their venture investments in ways that the typical fund lifecycle does not permit.

“The holding company next year will probably be $400 million of free cash flow, extremely profitable and growing really, really nicely, which is kind of an amazing thing.”

Jeffrey Katzenberg, Founding Partner at WndrCo

Mastering the Story

What distinguishes WndrCo from other firms with operator backgrounds, and there are many, is the specificity of the value it offers founders. Every venture firm in Silicon Valley claims to be more than capital, but few can articulate exactly what that means.

Katzenberg’s contribution is perhaps the most unusual. He functions, in effect, as a storytelling consultant and an enterprise sales weapon for the companies he backs. He will sit down with a founder, go through their sales deck, and help them frame their narrative with the precision of someone who has spent forty-five years constructing stories for mass audiences.

Katzenberg will make a hundred sales calls alongside a founder if he believes in the product. He described himself, with evident relish, as having “a bottomless need to sell.” When WndrCo backed Harvey, the legal AI company now valued at approximately $11 billion, its founder Winston Weinberg told Katzenberg in their first meeting that he needed help with his storytelling. “I know my product crushes,” he said. “How do I get that out there into the world?” The same was true of Shiv Rao at Abridge, where the story needed to reach both the health systems buying the software and the doctors using it daily.

Katzenberg’s favourite maxim, borrowed from Walt Disney, is “My movies are only as good as their villains.” He can rattle off the list without pausing. “Ursula. Gaston. Scar. Farquaad. Tai Lung.” The lesson applies equally to business, where without a clearly articulated problem, without a compelling antagonist, the protagonist has nothing to overcome. It is a framework he applies rigorously to the way portfolio companies tell their stories to customers, investors, and recruits.

On the product side, the firm’s depth comes from ChenLi Wang, who has been with WndrCo since its inception and functions as something like a chief product officer across the portfolio. Wang, a Stanford computer science graduate who worked at Goldman Sachs, NEA, and Workday before joining Dropbox, where he ran the core product and engineering organisation, is described by Jaswa in terms that border on awe. When Wang applied for a job at Peter Thiel’s hedge fund Clarium Capital, Jaswa recounted, he scored the highest of anyone who had ever taken their aptitude test.

Together, this combination of storytelling and product craft gives WndrCo a distinctive pitch to founders. Their money, as Wang likes to say, is the same colour as everyone else’s. The question they try to answer is what else they can do.

“Listen, we all look for the cheat code. Our green is no different from anybody else’s. It’s not better, it’s not worse. It’s capital. But when you want to win deals, when you want to be with the best talent out there, they ask: what else can you do for me?

ChenLi Wang, General Partner at WndrCo

Fund Performance

The AI wave of 2025 and 2026 has been a vindication of WndrCo’s thesis, though the partners would be the first to caution against triumphalism.

“New technology just takes time to get moulded and shaped into products and into experiences that actually make an impact in people’s lives. This is really the year where we saw, oh yeah, this is different. AI is going to transform how we do business. And it’s not just people playing around with, hey, look how this is like a show-your-friend-over-their-shoulders type of moment.”

ChenLi Wang, General Partner at WndrCo

The numbers from WndrCo’s portfolio bear this out.

Cursor, the AI-powered coding tool, has been the firm’s most dramatic recent success. In November 2025, the company raised $2.3 billion at a $29.3 billion valuation. By March 2026, its annual recurring revenue had surpassed $2 billion, having doubled in three months, and the company was in early discussions for a new round at a reported $50 billion valuation. Wang described visiting startup offices and counting how many screens showed Cursor’s distinctive tab on the right side of the display. “In a year, if I just did a walk around in an office in twelve months, it just went from zero to 100 practically.”

Abridge, the medical documentation AI, has followed a similarly steep trajectory. After raising $250 million at a $2.75 billion valuation in February 2025, the company raised a further $300 million just four months later at $5.3 billion, led by Andreessen Horowitz and Khosla Ventures. The tool is now deployed in more than 150 health systems, transforming how physicians document patient interactions.

Figma’s IPO on 31 July 2025 was another landmark. The collaborative design platform raised $1.2 billion at $33 per share, giving it an initial valuation of approximately $19.3 billion. Shares surged more than 250% on their debut, briefly giving the company a market capitalisation exceeding $67 billion. Databricks, in which WndrCo invested in 2021, raised approximately $5 billion in equity plus $2 billion in debt in February 2026 at a $134 billion valuation, with a revenue run rate exceeding $5.4 billion. Deel, the global HR and payroll platform, reached a $1 billion-plus revenue run rate and is preparing for an IPO as early as 2026.

Yet Katzenberg is measured about what lies ahead. Asked whether the AI bubble will burst in 2026, he chose his words carefully.

“I don’t know whether it’s AI. Rather than look at it from the extreme notion of what it means for a bubble to burst, I think there’ll be a reckoning in which those that actually are producing real results and are being deployed in really effective and efficient ways, where you can actually see an outcome quickly and it’s real, will win. Not everybody is going to win at this.”

Jeffrey Katzenberg, Founding Partner at WndrCo

He pointed to the staggering sums being invested by the hyperscalers: $700 billion in capital investment over twenty-four months, by his estimate. “On the other side of that, it needs to produce a result,” he said. “It needs an ROI.”

Wang’s view is more positive, but complementary. He believes that even if model development were to stop at current levels, there would remain years of work to be done in productising AI for delightful user experiences.

“Even if you just froze model development at this moment in time, there’s still so much more that we need to figure out and do and productise to actually make delightful products and experiences for consumers, for businesses.”

ChenLi Wang, General Partner at WndrCo

This conviction that the application layer matters as much as or more than the model layer has guided WndrCo’s investment decisions, steering it toward companies like Granola (the AI note-taking tool that recently raised $125 million at a $1.5 billion valuation) and away from infrastructure bets that depend on winner-take-all outcomes at the foundation model level.

The Last Mile

There is an interesting tension in how WndrCo evaluates companies, one that speaks to a broader debate in the AI industry about what constitutes real quality.

Wang tells a story about Granola that reveals something about his sensibility. He was using the product in a meeting where a founder was rattling off the names of competitors in an esoteric domain. Later, when he reviewed the AI-generated notes, every company name had been spelled correctly, even the obscure ones. The product had clearly gone and performed web searches to verify the spellings, polishing the notes with a level of care that most competing products missed.

“Years ago, I read that Steve Jobs obsessed about the back of the circuit boards that went into the computers. He wanted them to look pretty, even though nobody really opens the computers to look at that. And that was sort of his craft and dedication and almost a message he sent to the team.”

ChenLi Wang, General Partner at WndrCo

That attention to craft, the thing that is almost impossible to measure but instantly recognisable when you encounter it, is central to how WndrCo thinks about what makes a great product. Wang has an explicit scepticism of benchmarks as the primary way to evaluate AI companies, a position that places him somewhat at odds with the dominant discourse in the industry.

“The irony is, as many people now talk about AI replacing humans, we have never assessed our best humans based on benchmarks. How many years have parents complained about standardised testing dumbing down their kids? And yet we’re going down the same route with AI. People do very well in benchmarks and they fail in the real world.”

ChenLi Wang, General Partner at WndrCo

His alternative framework centres on what he calls “the last mile”: the ingenuity, context, and workflow design required to turn a capable model into a product that transforms how a specific category of worker does their job. He believes that perhaps 60 to 70% of AI use cases do not actually require frontier models at all, and that the real competitive advantage lies in product craft, not raw model capability.

Katzenberg is emphatic that partners should try the product themselves before taking a meeting. “The first thing you should be doing is just try the product,” he said, with evident exasperation at the number of investors who do not. “And by the way, I can tell you when you’ve actually tried the product.”

Choosing Good Partners

“Partnership to me is honestly, I look at it foundationally as maybe the single most important thing I’ve had in every chapter of my career,” Katzenberg said, listing the mentors who shaped him: Barry Diller, Michael Eisner, David Geffen, Steven Spielberg. When asked which had the greatest impact, he was unequivocal: “Barry Diller. He just invested in me and in my career. I think he believed and saw things in me I didn’t actually see in myself yet.”

WndrCo’s partnership itself, now a decade old, shows no signs of entropy. “It’s more fun than ever,” the partners have said, and the evidence suggests this is not mere performance. They recruited a team of partners whose complementary skills, ChenLi Wang on product, Anthony Saleh on seed-stage pattern recognition, Jeffrey Nykun on analytical rigour, Justin Wexler on enterprise AI deployment, form something closer to a company management team than a typical venture partnership.

“The common thread across everything is that we’re really looking for founders that we think have a chance of cracking an important problem,” Jaswa told TechCrunch when the firm’s venture funds were announced.

In a world awash with capital and increasingly sophisticated AI tools, WndrCo’s bet is that the scarcest resources remain human. The founder with genuine craft, the investor willing to make a hundred sales calls, or the partner who notices when an AI product spells an obscure company name correctly. It reflects their conviction in taste, in partnership, in the belief that storytelling and product obsession are the hardest skills of all.

And their track record indicates that’s leading them to look in all the right places.

Are you an outsider looking to raise your own fund? Deploy capital seamlessly with Odin.

“Venture investments tend to be focused on a few areas of technology that are perceived to have great potential. Increases in venture fund-raising which are driven by factors such as shifts in capital gains tax rates appear more likely to lead to more intense competition for transactions within an existing set of technologies than to greater diversity in the types of companies funded.

Policymakers may wish to respond to these industry conditions by (1) focusing on technologies which are not currently popular among venture investors and (2) providing follow-on capital to firms already funded by venture capitalists during periods when venture inflows are falling.”

Short-Term America Revisited? Boom and Bust in the Venture Capital Industry and the Impact on Innovation, by Josh Lerner and Paul Gompers

In January 2026, Andreessen Horowitz announced that it had raised over $15 billion across five new funds, its largest haul to date, bringing total assets under management above $90 billion. A few months earlier, General Catalyst closed approximately $8 billion for its twelfth fund. These are staggering sums that represent an extraordinary concentration of capital, at a time when broader venture fundraising has been notably weak. In 2025, US firms secured just $66.1 billion for new funds, down from $101.3 billion in 2024 and well below the 2022 peak of nearly $223 billion.

And yet, for all this firepower, many industries where capital is most needed remain chronically underfunded. Building a small modular nuclear reactor costs billions and takes more than a decade. Bringing a new cancer drug through clinical trials demands an average of $1 to $2 billion over ten to fifteen years, with a roughly 90% probability of failure. The fusion energy companies that could supply limitless clean power face development cycles stretching decades into the future. These are the technologies on which civilisational progress most clearly depends. They are also the technologies that venture capital, as currently structured, is least equipped to fund.

The question of why this is so, and whether anything can be done about it, has occupied researchers and policymakers for years. Today, as governments across the West embrace reindustrialisation, and the limitations of financialised economies become harder to ignore, it is increasingly urgent.

A growing body of academic work describes a solution (or a combination of three solutions) which would operate in harmony with private investors to strengthen the entire venture and growth ecosystem without repeating mistakes of the past:

1. The government as a limited partner for emerging managers. Addressing the network-driven constraints in venture capital that favour incumbents, reducing competition via limiting the threat of new entrants.

2. The government as a venture philanthrophist for sovereign technology. Addressing the difficulty that highly novel and capex intensive ideas may have in raising capital from the general pool of venture capital firms.

3. The government as a megafund guarantor of growth equity and debt. Addressing the hurdle into public market by providing later stage mezzanine financing which brings down the cost of capital for strategically valuable companies.

Each pillar of this strategy exploits the government’s ability to direct investment on longer time horizons and appreciate benefits beyond simple cash returns.

Impatient Allocation

Venture capital has been, by almost any measure, one of the most potent engines of innovation in economic history. Research illustrates that a dollar of venture capital is three to four times more effective at stimulating patenting than a dollar of traditional corporate R&D. The model is powerful, for a particular kind of company.

“The estimates therefore suggest that venture capital, even though it averaged less than 3% of corporate R&D from 1983 to 1992, is responsible for a much greater share—perhaps 10%—of U.S. industrial innovations in this decade.”

The Venture Capital Revolution, by Josh Lerner and Paul Gompers

The trouble is that the model’s structural incentives push capital towards certain categories of investment and away from others. A traditional venture fund has a ten-year life. Its general partners are rewarded on carried interest, a share of profits above a hurdle rate that must be realised within that fund’s lifetime. This creates a preference for investments that can reach a liquidity event, whether through IPO or acquisition, within that time period. Software companies with low capital requirements, rapid iteration cycles, and high gross margins are a natural fit. Companies building heavy physical infrastructure over fifteen-year timelines are not.

“The decision to continue funding certain compounds or technologies is often based on the likelihood of exiting the investment at a later date. If capital market conditions are unfavorable, or if the economy is not robust, pharma companies and venture capitalists may postpone or withdraw funding until the market environment improves, irrespective of the scientific merits of the project.”

Frequently Asked Questions About Megafunds, MIT Laboratory for Financial Engineering

This creates what researchers call the “valley of death,” a funding gap that opens after initial proof of concept but before a technology is commercially viable. For capital-intensive, long-horizon industries, this gap can be fatal. An estimated 80% of energy startups fail not because their technology does not work, but because they cannot sustain funding through the scaling phase.

Moreover, venture capital is intensely cyclical. During booms, capital floods into whichever category of technology appears hottest, to the point of oversupply. During busts, funding for even promising ventures evaporates. Government programmes designed to spur venture activity have frequently compounded this problem by targeting sectors already awash with private capital, rather than addressing gaps in the market. As the researchers note, many policy steps have had the consequence of exacerbating the cyclical nature of venture funding, rather than smoothing it.

Far too often, the short-term appearance of a successful programme is more important than actual success in driving innovation. This is a direct reflection of venture capital’s wider problem with the incentives of generating markups for fundraising purposes.

“While there were other sectors that held enormous technological promise and where entrepreneurs were struggling to raise money (alternative energy technologies, for one), these officials were drawn, like moths to a flame, to the sectors that were already overfunded by angel investors and venture funds alike. This decision may have been rational for the ATP bureaucrats eager to ensure their program’s survival. But it was profoundly at odds with the program’s mission to identify and rectify failures in the market for funding early-stage technologies.”

Boulevard of Broken Dreams: Why Public Efforts to Boost Entrepreneurship and Venture Capital Have Failed—and What to Do About It, by Josh Lerner

Scaled Funds, Limited Appetite

You may look at the emergence of scaled venture funds and see the beginnings of a solution. Firms like Andreessen Horowitz and General Catalyst now command capital at a scale that was unthinkable a decade ago. Their limited partners include sovereign wealth funds, major pension systems, and other large institutions that are able to invest on longer time horizons. In theory, this should allow for more patient investment in capital-intensive technologies.

In practice, it does not, or at least not sufficiently. The structural incentives of the traditional limited-partnership model remain in place regardless of fund size. LPs hope for returns of 20–30%. The pressure to invest in sectors where such returns are achievable has intensified. As Ben Horowitz wrote in announcing the raise, the firm’s mission is to ensure that the United States wins “the key architectures of the future,” meaning AI and crypto.

General Catalyst, similarly, has framed its latest fund around “applied AI” and “global resilience,” with a strong emphasis on building technology stacks that can be rapidly deployed within existing industries. These are sophisticated strategies, but they remain oriented towards categories where there is already a significant concentration of private capital to drive activity.

Essentially, their strategy to achieve competitive returns at scale is not to exploit larger LPs patience to invest in longer-term projects, but instead to hold “hot” companies for longer, harvesting more of their growth in private markets before seeking an exit, which has a number of negative consequences.

The result is a venture industry with more capital available than at any point in history, yet one that remains structurally unable to fund the technologies most needed for long-term economic security and industrial prosperity. The biggest funds chase the biggest near-term opportunities, and the hardest problems remain ignored.

Sovereign Megafunds

This is where the work of Andrew Lo and his collaborators at MIT’s Laboratory for Financial Engineering offers something genuinely different. Lo’s “megafund” thesis, first developed in a 2012 paper in Nature Biotechnology and subsequently extended to fusion energy, proposes a financial structure specifically engineered for what Josh Brown, CEO of Ritholtz Wealth, has termed “HALO“ companies; Heavy Assets, Low Obsolescence.

Rather than asking venture-style equity investors to bear the entire risk of long-horizon, high-failure-rate R&D, the megafund pools a large number of projects, typically fifty to three hundred, into a single special-purpose vehicle. It then finances this portfolio not through traditional LP commitments alone, but through securitisation; issuing tranched financial instruments that appeal to different categories of investor.

The senior debt tranches offer low-risk, fixed returns, similar to corporate bonds, and are designed for conservative institutional investors such as pension funds and insurers. Mezzanine debt tranches offer moderate risk and return. Equity tranches absorb the highest risk but also capture the greatest upside. The structure is analogous to mortgage-backed securities, though applied to a portfolio of research and development projects rather than home loans.

“In 2018, $254B was invested globally through venture capital financing, while the size of the global bond market stood at $102.8T, more than 400 times larger.”

Financing Fusion Energy, by Alhamdan, Halem, Hernandez, Lo, Singh, and Whyte

The pool of capital available through debt markets dwarfs the venture capital industry by orders of magnitude. By creating instruments that allow conservative institutional investors to participate in breakthrough R&D at a risk level they find acceptable, the megafund taps into a reservoir of financing that has been largely inaccessible to innovation. The ability to finance through debt also carries a lower cost of capital than equity, thanks to the tax deductibility of interest expenses, further improving the economics of the portfolio.

Lo and his collaborators have run extensive simulations demonstrating that this approach produces viable returns. Using historical oncology data from 1990 to 2011, they showed that a $5–$15 billion megafund could generate average annual returns of 8.9–11.4% for equity holders and 5–8% for debt holders, even accounting for the brutal attrition rates of drug development. Applied to fusion energy, simulations of a $10–$30 billion fund yielded projected returns of 7–12%, assuming moderate correlations between projects and occasional breakthroughs.

Crucially, the megafund also incorporates mechanisms for de-risking that go beyond simple diversification. Governments and philanthropic partners can provide “first loss capital” guarantees, effectively backstopping the most junior tranches to give private investors confidence. Lo and his collaborators have proposed that this philanthropic first-loss component would require initial funding of at least $500 million to guarantee five or more years of payments to senior tranche investors. This public-private hybrid design makes the megafund a particularly attractive instrument for government funding initiatives and large institutional LPs such as sovereign wealth funds and public pension funds, which are not only patient and well-capitalised but ideologically aligned with the kinds of positive externalities such as clean energy, public health breakthroughs, and national strategic resilience.

“We further de-risk the megafund by leveraging first loss capital guarantees from philanthropic sources and governments to fund coupon payments to senior and mezzanine bondholders in its early years.”

Financing Fusion Energy, by Alhamdan, Halem, Hernandez, Lo, Singh, and Whyte

This public-private hybrid potential makes the megafund extremely relevant to the current policy conversation. Governments seeking to accelerate energy transitions or secure domestic supply chains do not need to shoulder the entire financial burden themselves; by guaranteeing junior tranches, they can catalyse multiples of private investment. The taxpayer’s exposure is limited to the first-loss layer, while the vast majority of capital comes from private institutional sources attracted by the risk-adjusted return profile of the senior debt. The Israeli Life Sciences Fund already serves as a proof of concept for how government loan guarantees can stimulate private investment.

“The government provided some $50 million as a limited partner. Its funding serves as the first-loss capital through a preferred return scale, which allows for positive returns to other limited partners even if the fund suffers as much as a 10 percent loss. This structure has proved an attractive proposition for the fund managers (OrbiMed Israel Partners LP), who were able to raise the $172 million on top of the government's $50 million.”

Financing High-Risk Medical Research, by Melissa Stevens

Launch and run a private investment firm online, and work with over 10,000 VCs, angels and founders globally.

Set up an SPV or fund vehicle in less than a day, with Odin.

The Discovery Layer

One important qualification is that Lo’s megafund is not designed for inception-stage companies. It operates at the point in the cycle where the technology has been validated at some initial level but still requires massive capital to scale.

This is where small VC firms and emerging managers play a critical role, forming what can be considered the “discovery layer”, the initial interface between capital and innovation. Investors who find, fund, and develop the raw ideas that will eventually become candidates for megafund investment. A seed-stage investor working with a university spin-out developing novel fusion confinement technology, or a pre-seed fund backing a team designing advanced reactor components, are doing the essential work of making scientific breakthroughs “legible to capital”.

Research consistently supports this view. A 2025 study of government venture capital investment across Europe found that redirecting allocated capital towards early-stage investments produced the same productivity uplift as doubling the total capital allocation. The authors documented that government-linked VC financing had its strongest and most persistent effects when focused on young firms, despite the fact that public VC portfolios were systematically skewed towards later-stage companies. Separate research across other EU member states demonstrated that where governments fund emerging fund managers, they may produce above-market success rates and stronger job creation.

Indeed, a 2025 report by the Institute for European Policymaking and the ifo Institute reinforced this finding, showing that the EU’s roughly 100 billion euros of Horizon innovation funding had disproportionately flowed to large corporations with modest innovation records, while single-recipient grants to small, independent firms yielded the best long-term results.

“Any new Framework Program should thus focus on funding ideas, not companies. The key is not more money, but rather leaving space for disruptive innovation by encouraging bottom-up initiatives, especially by small independent companies.”

Funding Ideas, Not Companies, by Fuest, Gros, Mengel, Presidente, and Rujan

A healthy innovation ecosystem, then, would feature small funds scouring early-stage companies for breakthrough potential; developing those ideas through initial prototypes and regulatory milestones; and then passing the most promising candidates upward into a megafund structure capable of sustaining them through the capital-intensive scaling phase.

This complementary relationship could, if properly structured, strengthen both ends of the ecosystem. Emerging managers would have a clearer path for their portfolio companies beyond Series A, and potentially avenues for earlier liquidity, making their own funds more attractive to LPs. Megafunds would have a steady pipeline of de-risked, validated opportunities to absorb. The valley of death narrows considerably when there is a bridge waiting on the other side.

Venture Philanthropy

The megafund model and the discovery layer of emerging VC managers address the two critical stages of large-scale capital deployment and early-stage deal sourcing. There remains another category of technology for which even these structures may not move quickly or deliberately enough. When a government determines that a particular capability is strategically essential, whether in advanced materials, quantum computing, or biodefence, it may not wish to wait for a venture ecosystem to emerge around that technology.

Kyle Briggs, a physicist at the University of Ottawa who works on innovation policy and technology transfer, has developed a framework for exactly this problem. In a 2026 study drawing on PitchBook data covering 9,199 university-affiliated startups across North America and Europe, Briggs simulated the long-term performance of several strategies for investing in the earliest stages of research commercialisation.

His findings reflect a growing consensus about early stage investment in venture capital, where the most important variable in determining success is not the picking ability of the fund managers, but the portfolio size. Broad investment of small amounts of capital in a large number of startups statistically outperforms concentrated investment in fewer companies, by a wide margin. The returns from emerging technology commercialisation follow a highly asymmetric distribution in which a tiny minority of investments generate the vast majority of value. Under-sampling from such a distribution, as risk-averse investors often will, dramatically reduces the probability of capturing those outliers.

“Risk-intolerance creates a negative feedback loop in which under-sampling driven by risk aversion leads to subpar returns, which in turn reinforces the risk aversion that is in fact the root cause of the problem.”

Bootstrapping an Innovation Ecosystem, by Kyle Briggs

If it is effectively impossible to predict which early-stage research commercialisation attempts will succeed, then the correct strategy is not to try to pick winners but to fund widely, tolerate high failure rates, and trust that the statistical properties of the portfolio will deliver returns over time. A high rate of failure is a positive signal, indicating that the fund is investing broadly enough to be confident that it is not missing the valuable minority.

Briggs proposes a vehicle he calls “venture philanthropy”, in which a not-for-profit entity, seeded with public funds, makes small equity investments, typically $50,000 to $250,000, into a large number of university spin-outs and early-stage deep-tech companies at founding or shortly thereafter. Returns from successful exits are recycled into the fund to support the next generation of technologies. Because there is no profit-sharing obligation and no fixed fund lifetime, there is no time pressure to realise returns, making these investments genuinely patient capital. Because the fund is structured as a charitable entity, it can also solicit private donations, sharing the risk between public and private sectors while offering tax benefits to contributors.

The UK’s Innovation and Science Seed Fund provides a real-world example. It is credited with the existence of 78% of the companies in which it invested, and claims £29 of private follow-on capital for each £1 it has deployed.

Briggs’s simulations, using Monte Carlo methods across thousands of portfolio scenarios, demonstrate that this model can be self-sustaining. In his central scenario, a fund investing $4 million per year across eighty companies annually, taking small equity positions via convertible instruments, reaches breakeven at a median of approximately twenty years. The median annualised internal rate of return rises to 11.6% by year thirty and 13.5% by year forty.

More importantly, once breakeven is achieved, the fund enters an exponential growth phase as returns from successful exits are reinvested into an ever-larger number of new companies, and the ecosystem itself begins to compound. The vast majority of individual investments, over 96%, fail. But the realised value of the successful outliers more than covers the cost of those failures, and the broader economic benefits, in job creation, talent retention, and intellectual property kept within the sovereign jurisdiction, compound well beyond the fund’s direct financial returns.

“If seeded with public funds that are then augmented with private donations, venture philanthropy shares the risk of emerging technology investment between the public and private sectors, accepting a high rate of failure, long timelines, and sub-market profits that neither the public nor private sectors would tolerate alone. Because the goal is not profit beyond the need to be self-sustaining and operationally solvent, venture philanthropy measures success in terms of talent and IP retention, training first-time founders, company and job creation, and advancement toward societal goals.”

Bootstrapping an Innovation Ecosystem, by Kyle Briggs

The decades-long timeframe, initial target returns and the sheer number of investments required to reduce variance are mostly incompatible with a private-sector-led initiative. This is precisely why the private sector has not addressed this funding gap, and many emerging technologies have no reliable pathway from the lab to the market. A competitively-minded government, willing to accept modest financial returns in exchange for the broader economic and strategic benefits of a thriving domestic innovation ecosystem, is the obvious enabler.

The American SBIR programme, which until recently funded a comparable stage of research commercialisation, is estimated to create $22–$33 of economic value for every dollar invested. If a government can achieve even a fraction of that return while simultaneously retaining intellectual property and entrepreneurial talent within its borders, the case is easily made. The challenge is they must also be willing to tolerate high visible failure rates, resist the temptation to pick winners, and commit to investment horizons that extend well beyond electoral cycles, which are not natural instincts for public institutions.

The alternative, watching scientific breakthroughs and human capital leak to other jurisdictions, is a far more expensive form of failure.

A Long Time Coming

The problem of short-termism in venture capital and in the broader economy is not new. As far back as the 1980s, researchers were warning about “capital market myopia,” the tendency of financial markets to undervalue long-horizon investments.

“The purpose in writing this article is to focus attention on a phenomenon we call capital market myopia, a situation in which participants in the capital markets ignore the logical implications of their individual investment decisions. Viewed in isolation, each decision seems to make sense. When taken together, however, they are a prescription for disaster. Capital market myopia leads to overfunding of industries and unsustainable levels of valuation in the stock market.”

Capital market myopia, by William A. Sahlman and Howard H. Stevenson

The cyclical nature of venture funding, documented extensively by Gompers and Lerner, has been a persistent feature of the industry since its inception.

What has changed is the broader economic context, and increasing financialisation. Carlota Perez, the queen of techno-economics, has argued that the prolonged period since the dot-com crash represents a historically unusual failure to transition from an “installation period” of creative destruction to a “deployment period” in which new technologies are broadly diffused through the economy. In previous technological revolutions, from the steam age to the era of mass production, a financial crash midway through the cycle prompted institutional reform that redirected capital towards productive investment. This time, that correction has not occurred.

“The most acute impediment to a golden age deployment is the continued decoupling of finance from production. The usual casino of the frenzy periods has remained in force since the NASDAQ bubble, with major variations in form, but not in essence.”

A Long Delayed Golden Age: Or why has the ICT ‘installation period’ lasted so long?, by Carlota Perez

The reason, in Perez’s analysis, is that government responses to successive crises, from the dot-com bust to the 2008 financial crisis to the pandemic, have consistently prioritised the financial system over the productive economy. Quantitative easing pumped liquidity into banks without conditions requiring productive lending. Interest rates remained at historic lows for over a decade, making speculative financial activity far more attractive than patient investment in industrial capacity. As Perez observes, since the favoured banks were not required to use the money to fund the real economy, production investment and productivity increases remained at all-time lows, while the financial world played with derivatives and synthetic instruments across the globe.

This has clear alignment with developments in venture capital over the last few decades. Cheap capital encouraged larger fund sizes and higher valuations, but did not necessarily encourage better allocation. When money is abundant and cheap, the incentive to undertake disciplined, long-horizon investment in difficult technologies actually diminishes. It is easier, and more immediately rewarding, to fund software companies with two-year paths to revenue than to fund nuclear energy companies with fifteen-year development cycles.

"[VCs] furthered the financialization of firms, allocating capital to far from ideal uses and obtaining a share of the gains incommensurate with the risk or labor invested. In sum, VCs deepened asset-driven inequalities by devising a particular handling of abundance, which materialized in building a financial infrastructure for their needs."

Solving the problem of abundance: venture capital and the making of asset-driven inequalities, by Nils Peters

Today, however, the mood is shifting. The reindustrialisation movement, accelerated by supply chain disruptions, geopolitical competition with China, and growing awareness of the vulnerabilities embedded in deindustrialised economies, has created a new political and institutional appetite for productive investment. Governments across the West are deploying industrial policy at a scale not seen in decades.

In this environment, the government’s role as an LP and a VC, alongside the megafund model, offer a framework that aligns private incentives with public goals, directing large pools of institutional capital towards the most consequential challenges.

Putting It Together

Consider how this three-layer stack would operate in practice for a government pursuing an advanced nuclear strategy.

At the LP layer, the government commits capital to a cohort of emerging fund managers who invest across a range of technologies including energy and advanced manufacturing. These managers make hundreds of seed investments across the sector, including in companies developing novel reactor components, advanced materials, control systems, and grid integration technologies.

Simultaneously, at the venture philanthropy layer, the government seeds a dedicated fund targeting nuclear technology spin-outs from national laboratories and universities. This fund invests $50,000 to $250,000 in every credible spin-out attempting to commercialise relevant research, regardless of perceived probability of success.

Five to ten years later, a handful of companies from both layers have demonstrated genuine technical progress. One has developed a novel fuel fabrication process. Another has built a prototype coolant system. A third has achieved regulatory pre-approval for a small modular reactor design. These companies now need capital at a scale that neither small VC funds nor venture philanthropy can provide.

A $20 billion energy megafund, backed by securitised bonds rated for institutional investors and an equity tranche for endowments and sovereign wealth funds, absorbs these companies into a diversified portfolio alongside fusion ventures, grid-scale storage projects, and other energy technologies. The government guarantees the fund’s junior tranche, limiting taxpayer exposure to the first-loss layer while catalysing multiples of private institutional capital. The companies receive the patient, large-scale financing they need. The pension funds receive steady coupon payments from the senior debt. The equity holders gain exposure to transformational upside. And the early-stage investors, both the emerging managers backed by government LP capital and the venture philanthropy fund, have a clear exit path for their positions, improving their own returns and making them more attractive to future investors.

The beauty of this structure lies in the fact that each layer’s weakness is addressed by another layer’s strength. Small VC funds are excellent at discovery but cannot deploy capital at scale or over long timelines. The megafund can deploy capital at scale over decades but cannot source unproven, early-stage opportunities. Venture philanthropy can fund broadly and patiently at the inception stage but produces returns too slowly and too modestly for private investors. Government LP activity can support an ecosystem of emerging managers but cannot direct capital towards specific strategic priorities.

The total public expenditure required is, by the standards of government budgets, remarkably modest. An LP commitment of perhaps $200 to $500 million spread across a generation of emerging fund managers. A venture philanthropy seed of $50 to $100 million in each of several strategically targeted technology domains. A first-loss guarantee of $500 million to $1 billion to catalyse a megafund that raises $10 to $20 billion from private institutional sources. Against these outlays, the government receives a growing domestic innovation ecosystem, strategic control over critical technology supply chains, high-skill job creation, and, over the long term, a financial return on its investments that compares favourably with most other categories of public expenditure.

What this framework demands of governments is long-term thinking, institutional patience, and a tolerance for visible failure. Most companies funded by the venture philanthropy layer will fail. Many of the emerging managers supported at the LP layer will produce unremarkable returns. Some projects within the megafund will collapse entirely. These outcomes are the necessary cost of a strategy built on portfolio logic rather than the illusory pursuit of certainty. The alternative, which most governments currently favour, is to fund a small number of carefully selected projects and companies, demand accountability at every stage, and declare success or failure on the basis of individual outcomes. It appears more responsible on the surface, but is, in fact, far more wasteful on the balance of ultimate outcomes (never mind the risks associated with adverse selection and choking innovative companies with red tape).

Right now, innovation is unable to channel public investment and large institutions to the points in the pipeline where they will have the most impact. The three-layer model described here (government as LP, government as venture philanthropist, and government as megafund guarantor) offers a blueprint for each stage of the pipeline, from the first hint of an idea in a university laboratory to the deployment of a functioning power plant or manufacturing facility. And it does so in harmony with the private venture capital ecosystem, which stands to benefit at the growth stage from government LP/VC activity catalysing opportunity, and at the early stage from megafunds expanding the aperture of what is fundable by providing downstream capital and exit liquidity.

As the world asks how to rebuild its industrial foundations, we must look at how to scaffold this movement with new financial structures.

You do the investing we’ll do the admin. Start and run your VC firm online. Everything from structuring, capital calls, fund admin, SPV’s and more at your fingertips.

When: Friday March 27th
Where: London (Covent Garden)
What: a screening of ‘The AI Doc’

Request to Join

The AI Doc: Or How I Became an Apocaloptimist purports to be the most important documentary of the year. It features Sam Altman, Dario Amodei, Demis Hassabis, Eliezer Yudkowsky, Tristan Harris, and 40 or so others wrestling with what AI means for the future of humanity.

​We thought it was worth watching with a room full of people who have something interesting to say about it. Far from ridiculing it, we intend to deal with it seriously.

​Whether or not you share its conclusions, what happens when the most powerful technology ever built outpaces our ability to govern it is an interesting question.

​Come watch (it’s free). Then let’s argue about it.

We’ve got 4 spots left.

Now, over to Mr Gray…

“Many of the largest companies in the U.S. by market capitalization, including Apple, Google, Microsoft, and Cisco, were backed by venture firms in their early days. In addition, more than 60% of the IPOs that have occurred since 1999 have been venture-backed. Given that venture firms provide capital to highly innovative companies with the potential to create large social surpluses, distortions in their capital allocation decisions could have important welfare implications.”

Propagation of Financial Shocks: The Case of Venture Capital, by Richard R. Townsend

Every year, Venture Capital Journal surveys the institutional investors who provide the financial backbone of the venture capital industry. These limited partners, the pension funds, endowments, foundations, and family offices that allocate capital to venture funds, collectively shape the direction of innovation finance through their commitments, or their reluctance to commit.

Of the LPs surveyed, 32% said they plan to invest more in venture capital this year, while 58% intend to invest the same amount. Only 10% plan to invest less, the lowest figure recorded in at least six years. On the question of new commitments, 14% said they would make significantly more and 43% said slightly more, both figures up from the prior two years. The headline numbers seem promising.

If you look a little closer, the survey hints at allocation decisions that risk compounding many of todays problems. LPs are retreating from the smallest, earliest, and newest funds at a pace that should concern anyone who cares about the long-term health of venture capital’s engine of innovation.

The Emerging Manager Exodus

The most striking finding in the 2026 survey is that 57% of LPs said they would not consider backing an emerging manager in the next twelve months. That figure was 33% just one year earlier. In a single year, the proportion of LPs unwilling to invest in new fund managers nearly doubled, despite consistent evidence of strong performance in that group.

This was similarly reflected in stage preferences, where only 7% of LPs said they planned to invest more in seed-stage funds, 32% said they would not invest in seed at all, and 14% said they would invest less, leaving 46% investing about the same. Compare this with early-stage and growth funds, where 10% of LPs planned to invest more and 66% planned to hold steady. Capital is retreating to the “safety” of established names and later stages, away from the frontier.

Of course, this is something we have all sensed over the last couple of years, but it is fascinating to see the extent of the issue borne out in the data.

The top three reasons LPs give for their frustrations with venture capital provide some context:

• 31% said returns have not lived up to expectations

• 21% said they were unable to get allocation to the best funds

• 14% said it was too hard to pick winners

In reality, these are three different ways to say the same thing: performance has been disappointing due to an extended period of weak liquidity.

Others cited an inability to invest as much as they would like, inflexible terms, the burden of due diligence relative to small check sizes, and a lack of transparency by managers. These are real concerns, some of which reflect a need for greater professionalization and stronger standards in venture capital.

Problems of Venture Capital’s Own Making

“As a group becomes ten times larger, expenses increase much less than tenfold. As a result, management fees themselves become a profit center for the firm. These steady profits may create incentives of their own which may not be very appealing to investors. For instance, there will be an incentive to raise a larger fund at the expense of lower returns, which in turn may be tied to the greater concentration of capital held by a few investors; an incentive to put funds to work quickly and with a subpar amount of vetting so that a new fund can be raised sooner; and an incentive to focus on excessively safe investments that will not have as much upside but will pose less risk of a franchise-damaging visible failure.”

Venture Capital’s Role in Financing Innovation: What We Know and How Much We Still Need to Learn, by Josh Lerner and Ramana Nanda

The period from roughly 2016 to early 2022 was one of the most overheated in venture capital’s history. LPs flooded the market and the mega-funds emerged. Valuations climbed to levels disconnected from any reasonable expectation of near-term returns.

The consequences of that era are now well understood. Overcapitalization slowed exits by creating an overhang of overvalued private companies that couldn’t go public at prices their investors needed. The IPO market, which was already in decline before 2022, effectively shut for many venture-backed companies. The M&A landscape offered limited alternatives. Distributions to LPs dropped sharply. And in the most damaging cases, the loosening of governance standards enabled fraud and mismanagement at companies that had raised enormous sums with weak oversight.

Lerner and Nanda found that among publicly traded firms worldwide, seven of the top eight by market capitalization in 2020 had been backed by venture capital, and that VC-backed firms accounted for nearly 89% of recorded R&D expenditure among the cohort of post-1995 IPOs still public at the end of 2019. Venture capital, at its best, is an extraordinarily effective mechanism for financing innovation. Their concern was that structural changes in the industry were undermining that effectiveness. The concentration of capital in a small number of large funds, the narrowing of investment focus toward software and away from harder technologies, and the erosion of active governance were all trends they flagged as potentially detrimental to the broader innovation ecosystem.

Crucially, these problems are almost entirely correlated with the rise of mega-funds. The overcapitalization, the degraded governance, the herd behavior, the pursuit of capital velocity to justify ever-larger fund sizes, these are all features of the largest end of the market. The small and emerging fund managers that LPs are now abandoning were, for the most part, bystanders. Where these small firms did participate in the excesses of the boom, it was typically by chasing the same hot themes that mega-funds had already bid up, a reflection of LP demand signals.

Want to jump into the small fund gap? Do it with Odin, your full-stack partner for SPVs and fund administration

Collateral Damage

“Despite the fact that venture firms are structured to be more robust than most intermediaries, their portfolio companies can still face increased difficulty accessing capital when others that share the same investor suffer a negative shock. This can ultimately lead companies with high innovative potential to fail.”

Propagation of Financial Shocks: The Case of Venture Capital, by Richard R. Townsend

The pattern of innocent bystanders absorbing the damage of a correction they did not cause is not unique to the current cycle. There are clear examples from history, documented in a study by Richard Townsend of Dartmouth’s Tuck School of Business, examining how the collapse of the dotcom bubble in 2000 propagated through venture capital portfolios to companies with no connection to the internet sector.

Townsend’s research focused on non-information-technology companies, firms operating in sectors like biotechnology, medical devices, and energy, that happened to share venture investors with internet companies. When the bubble burst, these non-IT portfolio companies experienced a 26% larger decline in their probability of raising follow-on financing if their investors had high exposure to the internet sector, compared with similar companies whose investors did not. Indeed, after five years, the estimated probability of a company not having raised another round increased by over 8% for the affected group, a difference that, in the venture context, typically translates into company failure.

Critically, these companies were not worse investments. Townsend examined their patenting productivity, both the number of patents filed and the citations those patents received, and found no evidence that non-IT companies backed by internet-focused venture firms were of lower quality than their peers. They were just unfortunate collateral damage. Their investors’ portfolios had been contaminated by losses elsewhere, and the contamination spread to them through a mechanism that had nothing to do with the quality of their own technology or business prospects.

By exploiting the fact that many startups have multiple venture investors, Townsend was able to examine the same company across investors with varying levels of internet exposure. For the same portfolio company receiving capital from multiple venture firms, those investors with greater internet exposure were significantly more likely to drop out of follow-on rounds after the bubble collapsed. A one standard deviation increase in internet exposure was associated with a 38.5% increase in the probability of an investor dropping out, relative to an unconditional dropout rate of just 10.8% in the pre-bubble period. This within-company variation makes it far less likely that the results were driven by unobservable differences in company quality.

The Performance Paradox

The survey data on performance perceptions highlights the extent of LP disappointment. Only 13% of respondents to the Venture Capital Journal survey said venture capital had exceeded their benchmarks for 2026, up slightly from what appears to have been a low of around 6% in 2024, but far below the roughly 46% that was typical in previous years. 53% said venture fell below benchmarks, an improvement from 63% in 2025 but still more than half of all respondents.

The anomalous year in the dataset is 2022, when 68% of LPs said venture had exceeded benchmarks over the last 12 months, the only year where the figure topped 50%. This was driven by the surge of distributions in 2021, the last gasp of the ZIRP liquidity cycle. What followed was a sharp and sustained deterioration in sentiment, with fewer than 10% of LPs satisfied with performance by 2024.

Curiously, in 2020, LP sentiment was also surprisingly low. Looking back on 2019 and facing the uncertainty of the pandemic, LPs could not see the wave of exits on the horizon. That pessimism came just before one of the best years for venture distributions in recent memory. Here, we can conclude that LP sentiment is a lagging indicator, not a predictive one. The survey measures how the market felt at the time, not how it is likely to perform in subsequent years.

If LPs are making allocation decisions based on recent disappointment, they are doing so at a moment when the drivers of that disappointment are already several years in the rearview mirror. The companies in current seed-stage portfolios were not part of the 2021 vintage frenzy, and have not yet had time to mature. Venture fund performance takes approximately eight years to become a meaningful reflection of a manager’s ability. Judging emerging managers by the returns of a period defined by mega-fund excesses is a major attribution error.

The No-Invest Equilibrium

“Like in a bank run, if current investors believe that future investors will withdraw financing from such a project, they should also withdraw their investment, even though all investors would be better off in the equilibrium in which everyone invests. This is not an irrational decision and furthermore, does not depend on information asymmetries. There are simply two equilibria – one in which everyone invests in a sector and one in which no one does.”

Financing Risk and Bubbles of Innovation, by Ramana Nanda and Matthew Rhodes-Kropf

A theoretical framework developed by Ramana Nanda and Matthew Rhodes-Kropf helps explain why the current LP retreat is so destructive. Their model of financing risk, published in a paper that has become influential in venture capital scholarship, shows that the availability of future funding is itself a determinant of present value. When investors believe that future investors will not fund a given sector or stage, they rationally withdraw, because the absence of expected future capital genuinely reduces the value of current investments.

The mechanism works through bargaining power. A startup whose investors cannot credibly signal future funding has a weaker negotiating position with potential acquirers. The acquirer faces a lower competitive threat from the startup, because the startup is less likely to survive and grow into a formidable competitor. This reduces the price the acquirer is willing to pay, which lowers the net present value of the investment, which makes the decision to withdraw capital rational.

The result is a self-fulfilling equilibrium, as the belief that funding will not be available becomes the reason that funding is not available.

Nanda and Rhodes-Kropf demonstrate that this dynamic hits innovative projects (where coordination costs are highest) the hardest. Investors can try to mitigate financing risk by committing more capital upfront, giving startups a larger cushion to survive a funding drought. But doing so destroys the real option value of staged investment, the ability to invest a little, learn, and decide whether to continue. This option value is most significant for the most uncertain, most innovative projects; precisely the kind that small seed-stage funds specialize in.

The paper is explicit that it is the most innovative sectors and firms that need the abundant funding equilibrium to invent and commercialize radically new technologies.

Townsend’s work on the dot-com collapse is a good illustration of this theory. His model shows that in principle, the direction of contagion within a venture portfolio is ambiguous. When some investments in a fund perform poorly, the remaining investments become relatively more attractive, since the venture firm has a fixed pool of capital to deploy. This reverse contagion effect, analogous to the bright side of internal capital markets, would predict that non-IT companies in internet-heavy portfolios should have benefited from the collapse, receiving more capital as their investors redirected resources away from failing internet bets.

The fact that Townsend’s evidence decisively rejected this prediction, showing that contagion was ordinary rather than reverse, tells us something important about the power of the fundraising channel. The theoretical possibility of capital reallocation within a portfolio was overwhelmed by the venture firm’s diminished capacity to raise new funds from limited partners.

Indeed, venture firms in the top quartile of internet exposure experienced a 47.5% larger decline in their fundraising capacity after the bubble burst compared with those in the bottom quartile; a fundraising shock that would cascade to their portfolio companies. The effect was strongest for young venture firms and for firms that had not raised a new fund recently, exactly the profile one would expect if the transmission channel was LP capital commitments. For companies backed by a venture firm that had just raised a fresh fund, investor internet exposure had no statistically significant effect on continuation financing.

Essentially, the damage was concentrated among companies whose investors were dependent on near-term fundraising, the firms most vulnerable to a shift in LP sentiment.

This parallel will be felt by emerging managers who raised Fund I in 2021 or 2022 are now attempting to raise successor funds in an environment where 57% of LPs have declared them off-limits. Townsend’s evidence shows that this kind of fundraising shock will propagate to portfolio companies, reducing their access to capital regardless of their underlying quality, and it does so most acutely for the youngest and most capital-constrained investors.

In summary, when 57% of LPs say they will not consider emerging managers, they are sending a signal that pushes that part of the market toward what Nanda and Rhodes-Kropf call the “No-Invest equilibrium”. Founders become less willing to take capital from emerging managers because they worry about follow-on risk, and co-investors grow cautious. The emerging managers’ portfolios lose value because the ecosystem around them is less optimistic. Returns decline, and LPs cite the declining returns as evidence that they were right to withdraw, completing the circular logic.

The Concentration Trap

“While venture funding is very efficacious in stimulating a certain kind of innovative business, the scope is increasingly limited. This concentration may be privately optimal from the perspective of the venture funds and those who provide them with capital. It is natural to worry, however, about the social implications of these shifts. For instance, promising startups developing renewable energy technologies and advanced materials, which might have broad societal benefits, may languish unfunded.”

Venture Capital’s Role in Financing Innovation: What We Know and How Much We Still Need to Learn, by Josh Lerner and Ramana Nanda

Lerner and Nanda’s research documents just how concentrated the venture capital industry has become. As of their 2020 analysis, the top 50 investors, representing roughly 5% of all venture firms, had raised half of all capital deployed between 2014 and 2018. The partners at these firms are overwhelmingly male graduates of a small number of elite universities, based in the San Francisco Bay Area. This concentration amongst investors also ends up limiting the diversity of ideas that get funded, via the well documented problems of homophily, herd behavior and heat seeking.

The authors note that while this concentration may be privately optimal for individual funds (expanding fee income, easier fundraising from LPs), it raises concerns about the cost to progress when categories of promising technology go unfunded. Additionally, such concentration dramatically raises the risk from a market correction, similar to the post-dotcom retreat documented by Townsend.

PitchBook data confirms this concentration problem has deepened in the years since. The belief that AI is the defining opportunity of this era, and that AI companies require enormous amounts of capital which small firms are ill-suited to provide, has pushed LPs further toward large funds.

There are two problems with this reasoning. The first is that the history of venture capital consistently shows that the hot theme of any given vintage is rarely where the largest outcomes emerge. Venture returns are driven by outliers, and outliers, by definition, are rarely in consensus categories. The second is that consistent allocation across stages and fund sizes is essential to a healthy venture ecosystem and good long-term performance.

“We find that consensus entrants are less viable, while non-consensus entrants are more likely to prosper. Non-consensus entrepreneurs who buck the trends are most likely to stay in the market, receive funding, and ultimately go public.”

The Non-consensus Entrepreneur: Organizational Responses to Vital Events, by Elizabeth G. Pontikes and William P. Barnett

Early-stage managers, including emerging managers who bring fresh perspectives and networks, are the entry point of the pipeline. Without adequate funding at the base, the later stages that LPs favor will eventually run short of high-quality companies to back.

LP’s Accountability for Allocation

Consider the eight categories of challenges that LPs cited in the survey. Returns not meeting expectations, inability to access the best funds, difficulty picking winners, constraints on capital deployment, inflexible terms, due diligence burdens, lack of transparency: many of these can be traced to structural features of the mega-fund era or to shortcomings in LP portfolio construction and monitoring.

If returns have disappointed, the relevant question is what funds those LPs selected and at what point in the cycle. If they were chasing oversubscribed brand-name funds during the 2020 to 2021 boom, paying for access to an overheated market, the underperformance is primarily a portfolio construction failure.

Not being able to access the “best” funds may relate to an inability to recognise promising managers early to secure allocation, or suffering from hindsight bias in relation to fund performance. Finding it too hard to pick winners may reflect a similar failure to develop the expertise required to evaluate firms and make objective judgments.

If terms are inflexible, it is worth noting that LPs have collectively failed to reform venture fund economics for decades, despite well-documented misalignments.

Indeed, Lerner and Nanda note that management fees, originally designed to cover costs for small funds, have become a significant profit center as funds have grown.

Additional research by Metrick and Yasuda shows that over $23 of every $100 invested ends up with the venture firm, the majority coming from fixed fees rather than performance-linked carry. Their study also questions the fundamental logic of larger venture capital funds.

“The VC skills that are critical in helping firms in their developmental infancy are not applicable to more mature firms that are ten times larger and already in possession of core management skills. So when successful VC firms increase the size of their fund, they cannot just scale up the size of each firm they invest in without dissipating their source of rent.”

The Economics of Private Equity Funds, by Andrew Metrick and Ayako Yasuda

In aggregate, many of the challenges expressed by LPs are what a less diplomatic observer might call a skill issue.

The problems LPs face are substantially of their own making, or of the making of the large-fund ecosystem they continue to favor. The small and emerging fund segment, which is bearing the brunt of the retrenchment, is the least connected to the dynamics that have driven the downturn.

Maintaining the Techno-Capital Interface

Nanda and Rhodes-Kropf include a finding that deserves particular attention from LPs looking to structure a venture capital portfolio. They suggest that the mix of investors should shift toward smaller participants during periods of low financing risk, because smaller and more frequent investments are well suited to periods when future capital is expected to be available. For example, a skew toward small funds in the 2014-2018 period would have produced a larger and more diverse set of opportunities to absorb capital at the peak in 2020-2022, reducing fragility and the extent of the correction which followed.

In contrast, the model predicts that smaller investors will virtually disappear from the market during downturns, as the coordination costs to continue funding non-consensus projects are simply too high. This is not a desirable outcome, just the unfortunate nature of how the market suppresses the most innovative projects during the periods when contrarian investment would be most valuable.

Indeed, Nanda and Rhodes-Kropf suggest that some genuinely valuable technologies may require robust financial markets to get through the initial period of development (i.e. Carlota Perez’ “Installation Periods”), because otherwise the financing risk is too extreme. If this is correct, then the withdrawal of LP capital from early-stage and emerging funds is a clear impediment to future innovation. The most radically new ideas, the ones most likely to produce the outsized returns that LPs claim to seek, may be unable to survive the current financing environment precisely because the investors who specialise in finding them have been cut off from capital.

A sensible LP should understand that you invest in small and emerging managers because you trust them to go out onto the frontier and find the next opportunity. You invest in mega-funds (or mid-sized firms who service the megafirm consensus) because you believe there’s a present opportunity that requires more capital. These are entirely different categories, with different risk profiles, and treating one as a substitute for the other reflects a fundamental misunderstanding of how venture returns are generated.

Importantly, any “present opportunity” was, at some point in the past, the “next opportunity”, and relied on frontier investors that provide the interface between capital and technology. If that interface is underallocated, technological progress will quickly stagnate as the stream of new opportunities dries up.

The Cost of Consolidation

The venture capital market went through a period of extreme overcapitalization driven by mega-fund competition and the entry of non-traditional investors. That period produced inflated valuations, weakened governance, and an exit overhang that continues to depress returns and distributions. LPs, surveying the damage, appear to be consolidating their commitments into the very segment of the market most responsible for these outcomes while withdrawing from the segment least connected to them.

The emerging managers who raised Fund I in 2021 or 2022, now struggling to raise a successor, are generally not victims of a meritocracy; their portfolios have not had time to produce meaningful performance data. They are victims of a market correction they did not cause, operating under a standard of judgment shaped by the excesses of a different part of the industry.

The performance of a general partner’s previous private investment funds can predict future fund performance to a degree, but that’s only with at least eight years of hindsight, according to PitchBook’s latest Allocator Solutions report. In reality, LPs typically make decisions about where to commit fresh capital about three and a half years, on average, after a fund’s predecessor has closed—making it too soon to know exactly how the GP’s latest investments will pay out to existing LPs.

Fund performance data unhelpful for LPs when it counts, by Emily Burleson

Nobody knows what the venture market will look like in five years. The LPs who were so pessimistic in 2020, convinced the market was broken, were about to get one of the best years for liquidity in a generation. The LPs who piled in during 2021, certain they understood the opportunity, funded many of the worst outcomes.

The lesson, repeated across every cycle, is that the most valuable thing an institutional investor can do in venture capital is maintain consistent allocation across stages and fund types, resist the temptation to time the market, and support the parts of the ecosystem that the consensus has abandoned. Right now, that means emerging managers and seed-stage funds.

Unfortunately, the survey suggests that most LPs are doing the opposite.

But therein lies an opportunity…

Get started with Odin, your full-stack, digital-first partner for SPVs and fund administration.