The Efficient Venture Markets Hypothesis (summary)

Last updated: 17 October 2026

Summary of an essay by Abraham Thomas, published in Pivotal on 17 October 2026. Read the full essay on Pivotal →

Abraham Thomas argues that the strange behaviour of today’s venture market is a rational response to venture becoming more efficient, in a way unlike public markets: “In public markets, efficiency manifests as faster convergence to the right price. … In venture, efficiency manifests as faster convergence to consensus.” He calls this the “efficient venture markets hypothesis”. AI needs far more capital than software did, that capital has to come from outside the traditional venture industry, and consensus is the tool scaled VCs use to make tech “legible to capital” and channel it in. The essay is the third in a trilogy on modern venture capital, after Minsky Moments in Venture Capital (2022) and Making Markets in Time (2025).

What does the hypothesis explain?

Thomas lists ten phenomena that seem to make no sense in venture today:

  1. Startups bifurcate into hot and not; “excellent businesses” that aren’t hot struggle to raise.
  2. VC firms bifurcate into multi-stage behemoths and niche strategies (solo capitalists, seed, frontier, SPV co-investment); mid-size, mid-stage firms are dying out.
  3. Traction is either ignored or “no number is high enough”, alongside some of the fastest revenue ramps in history.
  4. King-making and capital-as-a-moat, with a sense of pre-anointed winners and losers.
  5. Companies stay private longer than ever.
  6. Uniform cap tables, as VC firms co-invest in mega rounds alongside their rivals.
  7. Investors ignore conflicts of interest with existing portfolio companies.
  8. Pro rata rights are valued highly but routinely re-traded.
  9. Isolation and solipsism: inside jokes, closed networks, industry-owned media.
  10. Self-destructive PR from market leaders, especially the frontier labs, about destroying jobs and dooming humanity.

He argues all ten are rational once you see venture as a market that is efficient at reaching consensus.

How is venture efficiency different from public-market efficiency?

Public markets are usually called efficient in two senses: strong-form (“the price is right”: prices reflect all known information) and weak-form (“no persistent alpha”: you can’t consistently beat the market without providing a service or having a structural edge).

Neither fits venture. Companies are unique and opaque, terminal value estimates range from zero to trillions, financings are discontinuous, secondary markets are thin, there’s no good way to short, and there are no regulators, information standards or clearinghouses. So prices can’t be pushed toward fair value: “valuation is not value”. Persistent alpha does exist: the same firms stay at the top for decades, because top investors attract the best founders, who drive the returns that define the top investors.

Instead, venture’s efficiency is about time. Venture has always run on consensus: staged rounds align investors across a startup’s life and turn extrapolation into a tractable bridging problem (the argument of Making Markets in Time). Reaching consensus normally takes time, to hit milestones and reduce uncertainty. Efficiency in venture means compressing that time, and firms that can do so beat those that can’t.

Why is consensus endogenous?

“Consensus is endogenous: it exists because we decide it exists.” If consensus can be manufactured, investors can skip the intermediate steps of staged fundraising, product-market-fit discovery and company-building: “Why run the whole race when you can jump straight to the finish line?” The endgame is venture today, “the most consensus market in years”: LPs herd into a few mega VC firms, those firms herd into a few startups, and those startups herd into a few business models, fast.

The cost is steep: “Consensus decisions lead to mediocre outcomes.” If every VC chases the same deals, they all get the same returns. Why professional investors accept this, Thomas says, comes down to AI.

How does AI change the math?

Software dominated tech returns for two decades and is a great business: asset-light, recurring revenue, high gross margins, negative working capital, strong free cash flow. But it doesn’t need much capital, which capped the size of the venture industry and made relative performance between VCs matter.

AI “absolutely incinerates capital”: capex for chips, data centres and power; high inference costs; fast-depreciating training runs; lower gross margins; expensive data and talent; and no apparent upper limit, because of the bitter lesson.

What is the inbound trade?

That capital has to come from outside the traditional venture allocation (“the measly 5% of every diversified portfolio”), from net-new investors, and VCs stand ready to intermediate it. “If you’re a scaled VC, nothing is more important than inserting yourself into this flow. And consensus is the tool you use.” Consensus removes friction: if every GP has essentially the same portfolio, fees and performance, no LP has to make hard choices or take career risk. “Consensus lubricates the flow of non-tech capital into the tech industry.”

As an example of scale, Thomas cites Anthropic’s $65 billion raise in May 2026, as much as the entire traditional US VC industry raised in 2025, led by venture, growth and crossover firms. He has described the “venture majors” as market makers rather than investors, and says they have now found “the perfect flow to intermediate”.

Why would LPs accept consensus portfolios?

Because of the global opportunity set. The AI trade has been so lucrative that “median portfolios within the AI bucket utterly dominate even the best portfolios outside of it.” For a global investor, what matters is exposure to private AI, and a consensus portfolio that indexes the space is “a feature, not a bug”: “AI beta is global alpha.”

The full chain of the argument: AI needs unprecedented capital; it must come from outside tech; VCs want to intermediate it; consensus is their mechanism; consensus trades performance for efficiency; investors accept that because they get to index the tech sector; and the industry has organized itself around consensus.

How does the hypothesis explain each phenomenon?

Phenomenon Thomas’s explanation
Hot-or-not startups Hot means part of the consensus; outside it, you’re not in the flow and won’t get funded. Heat is self-fulfilling.
Hunters and gatherers Only two roles matter: hunters discover consensus targets (risk capital), gatherers deploy consensus dollars (growth capital). “You can’t be ‘kind of consensus’”, so the middle disappears.
Inconsistent attitude to traction Traction is neither necessary nor sufficient for consensus; sometimes consensus comes first and the metrics follow.
King-making If consensus is manufactured anyway, big funds can buy it, scaring off rivals and their funders.
Staying private Going public risks falling out of consensus (regulators, short sellers, analysts); a controlled cap table plus secondaries is safer.
Uniform cap tables There are fewer consensus companies than VC firms, and the biggest career risk is “missing consensus names”. This also explains vibe investing and thin diligence.
Ignored conflicts Everyone, founders included, knows mega firms have to own the big names, so the stigma of backing a competitor has gone.
Re-traded pro rata Venture runs on honour, not contracts, and late-stage cheques “bestow the seal of consensus”, which is worth more than the dollars. Dual-priced rounds show the same thing.
Isolation and solipsism Intermediaries of new capital justify their gatekeeping by stressing the industry’s otherness: “The SF bubble is an intentional construct!”
Alarming lab PR Net-new inflows pay off only if tech’s share of the economy grows dramatically. Admitting it can’t would imply less capital moving into tech, “and that kills the AI funding trade.”

What are the implications?