For forty years, venture capital operated on a closed feedback loop. Venture funds raised capital from limited partners, deployed it into startups at prices set by other venture funds, and returned it through exits priced by public markets. The discipline of the system, from ownership targets to return hurdles to time to liquidity, rested on the fact that the participants at the table shared a common set of economic constraints.
That system no longer describes the market. In 2025, non-traditional investors, including sovereign wealth funds, corporate strategics, hedge funds, and mutual funds, participated in roughly 30% of US venture deals but accounted for 83% of all investment value. By Q1 2026, that share had risen to nearly 90%. The capital flowing into venture-stage companies has never been larger. The share of it subject to venture economics has never been smaller.
This is not a story about too much capital. It is a story about who that capital belongs to, what it wants, and what happens to the rest of the market when the price-setter has a fundamentally different mandate.
THE PRIVATE MARKET ATE THE PUBLIC MARKET
The venture capital industry did not grow into its current scale by accident. The companies it finances simply stopped leaving.
In the 1980s, a venture-backed technology company stayed private for roughly five years before going public. Today, the median is twelve to fourteen years. The median company now goes public with $218 million in revenue, up from $16 million in 1980, at a market value of $1.33 billion, up from $105 million (inflation-adjusted). Andreessen Horowitz showed in 2025 that the most recent IPO cohort generated over 50% of their total market capitalisation while still private, a near-inversion of the 2014–2019 cohort, where 80% or more was created after listing.
The aggregate picture is even more striking. Private technology unicorns, roughly 1,300 companies valued above $1 billion, now represent approximately $4.7 to $5.2 trillion in combined value. There are roughly six times more private unicorns than public companies with a $1 billion market capitalisation. The number of US publicly listed companies has fallen from approximately 8,800 in 1997 to under 4,000 today.
What once would have been public equity is now private equity. When a company needs to finance fourteen years of private growth rather than five, it exhausts the capacity of traditional venture fund structures built on ten-year lifecycles. The companies that grow the fastest hit a scale where no venture fund can write the cheque they need. That is the opening through which sovereign wealth funds, hyperscalers, and corporate strategics entered. Their arrival is not an anomaly. It is a structural consequence of the private market absorbing what used to be public market territory.
The dynamic is self-reinforcing. Companies stay private longer, outgrow VC fund sizes, and attract non-venture capital. That capital has different return requirements, which inflates pricing beyond venture math. Higher private valuations reduce the incentive to IPO. Companies stay private even longer. The cycle deepens with each turn.
THE NEW MARGINAL BUYER
In 2025, non-traditional investors accounted for 83% of all US venture deal value. They are not playing the same game.
The NVCA’s 2026 Yearbook quantifies the shift precisely. Non-traditional investors accounted for 83% of the $320 billion deployed in US venture in 2025. In 2024, the equivalent figure was 42%. The trajectory from 42% to 83% in a single year is not a cyclical fluctuation. It is a structural reordering of who finances innovation.
Sovereign wealth funds are deploying capital against national strategic mandates. SWF-backed deal value surged 198% in 2025 to $199.9 billion. GIC led Anthropic’s $30 billion round. Temasek participated in OpenAI’s $122 billion raise. These funds collectively manage assets exceeding $13 trillion and operate over multi-generational time horizons. Their return requirements differ fundamentally from a venture fund that needs to return 3x net in ten years.
Corporate strategics are funding their own customers. Nvidia committed over $40 billion in equity investments in 2026 alone, with private equity holdings rising from $3.4 billion to $22.3 billion in a single fiscal year. Bloomberg documented the dynamic in detail: Nvidia and Microsoft invest in AI companies, and those companies commit to purchasing Nvidia hardware and Microsoft Azure compute with the invested capital. Anthropic received investment from both, then committed to $30 billion in Azure capacity and contracted to buy Nvidia’s next-generation chip systems. The investment is the purchase order.
This matters because the marginal buyer sets the price. In venture capital, the price of the last round becomes the mark for every existing investor in the company. It flows through to fund-level NAV, to TVPI, to LP reporting, and to fundraising narratives. When that price is set by a buyer whose return threshold is lower, whose time horizon is longer, or whose primary objective is strategic rather than financial, the mark carries information about that buyer’s mandate, not about the asset’s likely return to a venture investor. Every co-investor in the syndicate inherits a mark that was established under a set of economic assumptions they do not share. This is the mechanism through which non-venture pricing contaminates venture economics. It does not require fraud or misjudgment. It requires only that the buyer setting the price has a different definition of what the investment is for.
Consider the numbers. AI companies trade at 24 times revenue against 3.7 times for traditional software. Anthropic’s valuation moved from roughly $380 billion to $965 billion between Q1 and Q2 of 2026, a near-tripling in under three months. OpenAI closed $122 billion at an $852 billion post-money valuation. These are not prices that clear on traditional venture return math. They are positions, established by buyers underwriting something other than a financial return multiple.
The distribution data confirms the gap between paper marks and cash. Cumulative net cash flows to US venture LPs have been negative by approximately $197 billion since 2022. Distribution yield troughed at 7.5% in 2023 against a historical average of roughly 15–18%, recovering only modestly to 11.2% in 2025. Sixty percent of LPs now prioritise DPI over TVPI in fund evaluations. Only 537 US venture funds closed in 2025, the fewest in a decade. The fundraising environment is not rewarding paper marks; it is rewarding cash.
THE ECHO OF 2021, AND WHY THIS TIME IS DIFFERENT BUT NOT SAFER
We have been here before. The last time non-traditional capital flooded venture, it ended with the sharpest correction in a decade.
In 2021, crossover investors poured into venture at unprecedented speed. Tiger Global alone participated in 288 deals, deploying from a $12.7 billion fund near-fully invested within months. Crossover capital accounted for roughly 60% of US venture deal value. Then rates rose, the IPO window shut, and the correction arrived. Tiger’s portfolio was marked down by roughly a third. SoftBank’s Vision Fund posted a $26.2 billion loss. Fewer than 30% of the unicorns minted in the boom raised again over the following three years, and half of those that did accepted down rounds.
The 2026 market rhymes with 2021 in structure but diverges in character. The concentration is more extreme: five deals accounted for 73% of US deal value in Q1 2026. The share of capital influenced by non-traditional investors is higher: 83% versus roughly 60%. But the nature of the buyer is fundamentally different.
In 2021, the crossovers were at least attempting to generate venture-style returns, and they failed. In 2026, the marginal buyers are not even trying to generate venture returns. A sovereign wealth fund underwriting national AI infrastructure, or a chipmaker funding the ecosystem that buys its products, is optimising for something else entirely. The prices they set are rational for them. They are not rational for a fund-of-funds that needs to underwrite a return path to its own LPs.
The strongest counterargument deserves a fair hearing. OpenAI generates roughly $24 billion in annualised revenue. Anthropic grew from $1 billion to $30 billion ARR in approximately fifteen months. SpaceX priced the largest IPO on record. These are extraordinary businesses. But an extraordinary company can still be a poor investment at the wrong price, and the capital arriving latest, the crowding consensus capital, is precisely the capital that pays the highest price. That has been true in every cycle.
WHAT THIS MEANS FOR ALLOCATORS
The venture industry is not broken. But the pricing mechanism at its highest-profile layer has been structurally compromised.
We are not arguing that sovereign wealth funds and corporate strategics are unsophisticated. Many are among the most capable institutional investors in the world, and their entry into venture reflects a rational response to the migration of value creation from public to private markets. What we are observing is that their presence has changed the information content of venture pricing at the top of the market. When the price-setter has a different objective function, the price carries different information, and that information cascades through co-investor marks, fund-level TVPI, LP reports, and fundraising narratives.
For allocators, the implications are fourfold.
First, headline venture data no longer describes a single market. The NVCA itself notes that the 2025 venture market was “really two markets”: an AI market of roughly $220 billion dominated by sovereign and strategic capital, and everything else at roughly $100 billion operating at levels that resemble a normal venture year. Treating these as one market is a category error.
Second, TVPI built on marks set by non-venture buyers requires additional scrutiny. The question is not whether the mark is defensible when it is set, but whether the path from that mark to a cash realisation delivers a return that justifies the illiquidity and the fee structure of venture as an asset class.
Third, the market below the mega-deals, the early-stage, pre-consensus segment, continues to operate under recognisably venture economics. This is where the power law has the most room to operate and where the outlier that carries a portfolio is most likely to emerge.
Fourth, and perhaps counter-intuitively, the early-stage venture investor may be the most natural beneficiary of this structural shift. Non-traditional capital enters overwhelmingly at the latest stages, setting marks that flow back through the cap table to investors who entered at a fraction of the price. For a fund that backed a company at Seed or Series A, a sovereign-anchored mega-round years later is not a threat to their economics. It is the mark-up event.
The conditional risk is that the mark must eventually convert to cash. But the growth of the venture secondary market is changing that calculus. With direct secondary volume exceeding $90 billion in 2025, company-sponsored tender offers becoming routine, and secondary pricing recovering to roughly 94% of NAV, early-stage investors increasingly have a functioning mechanism to crystallise positions without waiting for an IPO. The secondaries market, in this context, is the bridge between non-venture pricing and venture DPI. It remains concentrated in a narrow set of names, as we explored in our previous Venture Pulse. But for the managers who hold positions in the companies that do trade, it transforms an unrealised mark into cash in the fund.
CONCLUSION
The venture capital industry is undergoing a structural repricing of who sets the terms. The private market has absorbed what used to be public market territory, creating companies that attract capital with mandates venture economics were never built to accommodate. That capital now dominates the highest-profile layer of the market, setting marks that cascade through the system.
This is not a temporary dislocation. It is the logical consequence of companies staying private longer, scaling beyond venture capacity, and attracting buyers whose objectives are strategic, sovereign, and industrial rather than purely financial. For the late-arriving consensus capital that pays the price these buyers set, the return arithmetic is uncertain. For the early-stage investor who entered years before, at venture prices and with venture discipline, that same capital is the exit liquidity.
The discipline that survives this transition is the one the market is currently abandoning: enter early, diversify across cycles, and let the consensus capital that arrives later validate the position you already hold. Understanding who sets the price, and where you sit relative to them, is the most important analytical question for any allocator in the asset class today.
SOURCES & NOTES
- NVCA 2026 Yearbook (data provided by PitchBook): US VC deal value $320 billion across 15,352 deals in 2025. NTIs participated in ~30% of deals and 83% of all investment value. NTI capital flows estimated at $80–$100+ billion. AI captured 65.4% of deal value. April 2026.
- PitchBook-NVCA Venture Monitor, Q1 2026: US deal value $267.2 billion; top five deals = 73.2% of deal value; five firms = 73.1% of new commitments; crossover investors 89.9% of deal value / 23.4% of deal count. April 2026.
- PitchBook-NVCA Venture Monitor, Q4 2025 / PitchBook 2025 Annual US VC Secondary Market Watch: Cumulative negative cash flows $197 billion since 2022. Distribution yield 7.5% trough (2023), 11.2% (2025), ~15–18% historical average. Funds closed 2025: 537.
- PitchBook (April 2025): 2024 crossover investor participation: 4.9% of deals, 42% of deal value.
- S&P Global Market Intelligence (January 2026): SWF-backed deal value $199.9 billion in 2025, up 198% from $67 billion in 2024.
- EY-Parthenon: SWFs invested $46 billion in AI ventures in the first eight months of 2025.
- CFA Institute / Bain & Company / State Street: SWF AUM $13–15 trillion as of 2025. Private market allocations rose from ~25% in 2020 to ~30% by end-2025.
- Andreessen Horowitz (“Private Markets Are the New Public Markets,” September 2025): Private tech unicorns ~$4.7 trillion aggregate value; ~6x more private unicorns than public companies at $1B+ market cap; median IPO age 14 years.
- Jay R. Ritter, University of Florida (cited by VanEck, CNBC): VC-backed tech companies private ~5 years (1980s) to ~12–14 years (2022–2024). Median IPO revenue: $16 million (1980) to $218 million (2024).
- CNBC (May 2026) / Bloomberg (March 2026): Nvidia $40B+ in equity investments in 2026. Circular financing documentation.
- Crunchbase / Tiger Global reporting (various, 2022–2024): Tiger Global fund losses; SoftBank Vision Fund $26.2 billion loss. Carta: 2021-vintage unicorn data.
