Startups, venture capital & private markets

Startups, venture capital & private markets

Industry ID: 11 | Slug: startups-vc | Researched: 2026-09-15 | Analyst: agent

CROSS-CUTTING SECTOR WARNING — READ BEFORE USING ANY NUMBER IN THIS DOSSIER. The industry roster classifies id 11 as a capital layer, not a vertical. Almost every funding figure here is the same money already counted in another sector's dossier. The $412.7B of US venture deployed in H1 2026 (PitchBook-NVCA) is overwhelmingly the same dollars that industry 01 (AI & foundation models) reports as AI investment, that industry 03 reports as semiconductor capital, that industry 06 reports as biotech financing. Never sum sector funding totals across dossiers. Section 3.4 below lists precisely which metrics double-count and which are unique to this layer.

1. Definition and boundaries

In scope. The supply side of private capital for company formation and growth: institutional venture capital (pre-seed through growth), angel and accelerator capital, growth equity and crossover investing, buyout and middle-market private equity, private credit to the extent it finances startups and sponsor-backed companies, the limited- partner allocator layer (endowments, pensions, insurers, sovereign wealth, family offices, and now defined-contribution retirement plans), the secondaries market (LP-led, GP-led, continuation vehicles, direct and employee tender offers), fund-of-funds and GP stakes, the private-markets service and data infrastructure (cap-table platforms, fund administration, marketplaces, benchmark providers), the exit environment (IPO, M&A, continuation, wind-down), and the geography of startup ecosystems including formation and failure rates.

Explicitly out of scope, and who owns it.

• What the money buys. The technologies, products and end-markets that startups build belong to the vertical dossiers. OpenAI's and Anthropic's model capabilities sit in 01 AI & foundation models; this dossier owns only the financing structure around them. When we say "AI took 86% of venture dollars," that is a statement about the capital layer's concentration risk, not an AI-market claim.
• Public equity markets in general. Only the IPO window as an exit mechanism is in scope. Secondary-market trading of seasoned public companies is not.
• Broad private credit. Direct lending to non-sponsor mid-market borrowers, asset-based finance, CLOs and consumer credit belong to a financial-services dossier. Only venture debt, sponsor-backed leveraged finance and NAV/GP financing are in scope here.
• Real estate and infrastructure funds. In the private-markets AUM tables they sit beside PE, but their drivers are property and energy, not company formation. Cited only where they explain LP liquidity.
• Crypto token fundraising other than tokenized exposure to private equity of real companies (which is in scope, and is a live 2026 story — see §8).

Boundary disputes worth naming.

1. Is private credit "venture"? Venture debt hit a record $68.8B in 2025 across ~1,000 transactions (Runway Growth/PitchBook, 2026-05-28) and is increasingly used by scaled late-stage borrowers. PitchBook counts it separately from equity; Crunchbase and KPMG treat some debt-and-equity packages inconsistently. This is one mechanical driver of the Crunchbase/KPMG gap in §3.
2. Are corporate strategic investments "venture capital"? In 2026 they dominate: corporate VCs accounted for a record 87.9% of US AI VC deal value so far in 2026 (PitchBook, 2026-07-21). NVIDIA is the largest corporate investor by deal value. If a chipmaker's investment in a model lab is counted as venture, then "venture returns" partly measure vendor financing, not third-party LP capital at risk. We keep it in scope but flag it everywhere.
3. Is an evergreen semi-liquid interval fund a "private markets" product or a retail mutual fund? Regulators are actively deciding this (§8, §11).
4. Where does a continuation vehicle sit? It is simultaneously an exit (for the old fund) and a purchase (for the new one). Counting it as both inflates both sides.

2. Subcategories

3. Market structure

3.1 Concentration: winner-take-most, and getting more so, at every layer

This sector is now among the most concentrated in the economy, and 2026 set records at four separate layers simultaneously:

The 93.5% figure is the single most important number in this dossier. It means the "record exit environment" described in the macro brief is, from the perspective of the median fund, largely not an exit environment at all.

3.2 Where the margin actually sits

Margin in this value chain has migrated decisively away from fund returns and toward fee streams on permanent capital:

• Carried interest requires realisations. With 2021-vintage funds averaging 0.05x DPI — the lowest five-year DPI multiple this century (PitchBook, 2026-08-04) — carry is, for a large cohort of managers, a theoretical asset.
• Management fees on perpetual vehicles do not require realisations. Evergreen fund AUM approached $500B by Q1 2025 and grew more than 30% in the twelve months to September 2025, with wealth investors now one-fifth of that AUM (MSCI, 2026-03-03). US semi-liquid PE vehicles alone raised $204B in 2025, more than double 2023's $92B (McKinsey, 2026-02-10).
• Fees on other people's liquidity problems. Secondaries advisory, continuation-vehicle sponsorship, NAV lending and GP-stakes all monetise the liquidity drought itself.
• Infrastructure and data. Cap-table platforms, fund administrators, marketplaces and benchmark vendors earn recurring revenue regardless of DPI — which is why the most reliable public data on this sector comes from firms whose commercial interest is in the sector appearing large and active. Treat all vendor data accordingly.

Pricing power sits with (a) a handful of mega-managers who can raise on terms, (b) the three or four companies that everyone needs exposure to, and (c) the secondaries buyers who set the clearing price for LP liquidity. It sits nowhere near the median GP or the median founder.

3.3 Barriers to entry

Historically low (a fund is a legal structure and a track record). As of 2026, effectively high: first-time fund formation is on pace for its lowest year since 2016 (PitchBook- NVCA, 2026-07-08), and the median time between fund closes has stretched to 1.7 years. An LP base that has received negative $202B of net cash flow since 2022 (PitchBook, 2026-08-04) has no recycled capital with which to back new managers. The barrier is not regulatory or technological — it is arithmetic.

3.4 Market size — with the double-counting map

Figures below are each attributed to a named modeller with a date, per contract §3.

Rule for downstream analysis: the deployment rows double-count; the fundraising, secondaries, continuation, evergreen and AUM rows do not. Sum only the second group.

4. Who matters

Leading managers (listed alternatives). Blackstone (NYSE: BX, https://www.blackstone.com), KKR (NYSE: KKR), Apollo Global Management (NYSE: APO), Blue Owl Capital (NYSE: OWL), Ares Management (NYSE: ARES), Carlyle Group (NASDAQ: CG), TPG (NASDAQ: TPG), EQT AB (STO: EQT), Brookfield Asset Management (NYSE: BAM), StepStone Group (NASDAQ: STEP). These are the only entities in the sector with continuous public disclosure — and are therefore the highest-value monitoring targets for Phase 2.

Venture firms that set the market. Andreessen Horowitz (https://a16z.com), Thrive Capital (https://thrivecap.com), Founders Fund (https://foundersfund.com) — together 48.1% of US venture fundraising in H1 2026; plus Sequoia Capital, General Catalyst, Lightspeed Venture Partners, Accel, Index Ventures, Insight Partners, Greenoaks, SoftBank Vision Fund, Khosla Ventures, Uncork Capital (seed).

Secondaries and liquidity specialists. Lexington Partners (Franklin Templeton), Coller Capital, Ardian, HarbourVest, Pantheon, AlpInvest (Carlyle), ICG Strategic Equity, StepStone, Adams Street Partners, Industry Ventures (venture-specific), Jefferies and Evercore private-capital advisory (the two firms whose volume reports are the market's scoreboard).

Notable private companies whose marks define the asset class. OpenAI, Anthropic, SpaceX (now public, Nasdaq: SPCX), xAI, Anysphere/Cursor, Databricks, Stripe, DeepSeek, StepFun. Roughly 950+ private companies are valued at $1B+, with 113 new unicorns minted in 2026 (PitchBook, 2026-08-04).

Platforms and infrastructure. Carta (https://carta.com), AngelList, Sydecar, Allocations, iCapital, Forge Global (NYSE: FRGE), Hiive, EquityZen, Upmarket, Nasdaq Private Market, Preqin (BlackRock), MSCI/Burgiss, PitchBook (Morningstar), Crunchbase, Dealroom (NL), Tracxn (IN), Cambridge Associates.

Accelerators. Y Combinator (https://www.ycombinator.com), Techstars, Antler, Entrepreneur First, Plug and Play.

Regulators. US SEC (https://www.sec.gov) — Chairman Paul Atkins; US Department of Labor / EBSA (fiduciary rules for DC-plan alternatives); FINRA; CFIUS and the outbound- investment regime; UK FCA; ESMA and the EU ELTIF 2.0 regime; India SEBI; China CSRC.

Government statistical agencies. US Census Bureau Business Formation Statistics (https://www.census.gov/econ/bfs/) — the only genuinely independent, free, high-frequency public series on company formation; BLS Business Employment Dynamics.

Trade organisations. National Venture Capital Association (https://nvca.org), Institutional Limited Partners Association (ILPA — the LP-side standard-setter for fee and DPI reporting), American Investment Council, Invest Europe, BVCA (UK), Indian Venture and Alternate Capital Association.

Research institutions. Cambridge Associates (US PE/VC Benchmark), MSCI Private Capital Solutions (formerly Burgiss), Stanford GSB Venture Capital Initiative, Harvard Business School Private Capital Project, NBER corporate-finance programme.

Consumer and civil-society counterparties. Better Markets, Americans for Financial Reform and AARP have all engaged on retail/401(k) access to private assets; the AFL-CIO Office of Investment engages on PE portfolio-company labour practices. These groups matter more in 2026 than in any prior year because the sector is, for the first time, asking to be sold to ordinary retirement savers.

5. Products, business models, technologies, customers

Major products. The closed-end 10-year commingled fund (2%/20%, still the default); the continuation vehicle; the evergreen/interval/tender-offer fund; the non-traded BDC; the SPV or feeder into a single company; the venture debt facility; the NAV loan against a fund's portfolio; the GP stake; the employee tender offer.

How money is actually made — and how that is changing. The textbook model is carry on realised gains. In practice, for the 2020–2022 vintages, realised gains barely exist: Cambridge Associates records that since the start of 2022, US VC managers have called 1.6x more capital than they have distributed, reversing the 2012–21 decade in which distributions exceeded calls by 1.3x (Cambridge Associates, 2026-01-05). The business has therefore shifted toward fee income on permanent and semi-permanent capital, toward advisory and structuring fees on liquidity engineering, and toward balance-sheet and strategic capital that does not need a DPI at all. The rise of corporate VC to 87.9% of US AI venture deal value is the extreme form of this: at the frontier, the marginal dollar is no longer a fiduciary's dollar seeking a return; it is a strategic buyer's dollar seeking supply, distribution or optionality.

Technologies that matter to the sector itself (as distinct from what it funds): cap- table and waterfall automation; fund administration and ILPA-template reporting; secondary marketplaces with transfer-restriction enforcement; SPV formation software that has cut the cost of syndication to near zero (and thereby created the grey market in §8); NAV-based valuation and smoothing methodologies, which are now a first-order accounting question; and tokenization rails, which in 2026 produced the sector's most visible enforcement flashpoint.

Customers and what they buy on. GPs sell to LPs on track record, DPI, access and (increasingly) liquidity terms. Founders "buy" capital on price, speed, signalling and follow-on capacity — and in 2026, at the frontier, on compute access, which is why corporate investors win deals fiduciary funds cannot. Wealth channels buy on access narrative and reported volatility; the reported low volatility of private marks is itself the product, which is precisely the feature under scrutiny.

6. Geography

Capital formation concentrates in the US: the Americas took 66% of Q2 2026 global funding (macro brief), and the US absorbed $412.7B in H1 2026. Within the US, the Bay Area, New York and Austin are the top three pre-seed metros, with Texas now at 9% of US pre-seed investment (Carta, 2026-08-13).

Europe — the structurally weak leg, and increasingly a single-country story. European tech raised €44.1B across 1,740 deals in H1 2026, with the UK alone at €18.7B across 423 deals, Germany €6.3B, France €6.0B, Sweden €2.8B, Netherlands €1.9B, Spain €1.7B (tech.eu, Amsterdam, 2026-07-30 — regional source). Dealroom (Amsterdam) puts the UK at a record 48.4% of European VC in 2026 YTD against a decade-long 33–37% norm. Europe's problem is not only that it raises less than the US; it is that its recovery is concentrated in one city. European closed-end PE fundraising fell 41% in 2025 to $118B (McKinsey).

Asia — a genuine China-led rebound. Asia took $42.8B in Q2 2026 on Crunchbase's count, with China just over $30B (+424% y/y, +76% q/q), India $3.3B, and AI companies taking over 60% of Asian funding. Largest rounds: DeepSeek $7.4B at a $50B valuation, StepFun $2.5B, DayOne (Singapore) $2.5B (Crunchbase, 2026-07-16). This conflicts with KPMG's $50.8B Asia figure in the macro brief — see §13.

India — a counter-cyclical contraction. Tracxn (Bengaluru) reports FY26 Indian tech funding of $11.7B, down 18% from $14.3B, with late-stage down 38% and seed down 15% even as early-stage rose 33%; mega-rounds fell from 23 to 13. But India produced 47 IPOs in FY2025-26 against 31 the prior year — the clearest example anywhere of a market where domestic public listing rather than trade sale is the working exit route (Tracxn via Analytics Insight, 2026-04-09 — regional source).

Demand and regulation diverge sharply from capital. Capital sits in the US; regulatory initiative on retail access also sits in the US (SEC/DOL); but regulatory initiative on investor protection in private markets sits in the EU and UK. This split is the sector's main jurisdictional arbitrage and the main reason a Phase 2 pipeline must monitor the SEC rulemaking agenda, the DOL fiduciary docket, and ESMA/FCA consultations separately.

Startup formation itself is cooling in the US. Census BFS recorded 531,728 seasonally adjusted business applications in August 2026, down 7.8% month-on-month, with projected business formations at 28,501, down 4.6% (US Census Bureau, released 2026-09-11). This is the only Tier-A, free, high-frequency formation series available and it is pointing down while venture dollars point up — another instance of the aggregate/median divergence.

7. Historical trend patterns

Over 25 years this sector has run four distinguishable cycles, and — critically — its own past false positives are well documented:

1. 1999–2002 dot-com. Peak-to-trough US VC investment fell roughly an order of magnitude. The durable lesson was not "the internet was fake" but "the exit assumption was fake": companies were funded against an IPO market that closed for three years. This is the exact structure of the 2021 cohort's problem.
2. 2004–2008 the "Web 2.0 / clean-tech" split. Web 2.0 delivered; clean-tech 1.0 was venture's single largest documented false positive, with most of roughly $25B deployed 2006–2011 written off. The tell at the time was that returns depended on policy subsidies and commodity prices rather than on unit economics — an indicator directly transferable to today's AI-infrastructure exposure.
3. 2009–2015 the mobile/SaaS build. Genuine, broad, and the last period in which the median venture fund, not just the top decile, returned capital.
4. 2016–2021 the ZIRP melt-up. Crossover investors (Tiger, SoftBank, mutual funds) compressed diligence to days and inflated late-stage marks. Vintage 2021 is the resulting artefact.
5. 2022–2024 the reckoning that was postponed rather than taken. Down rounds rose but never to 2001 levels; instead, companies took bridges, cut burn and did not reprice. Marks were held. The bill came due as a liquidity problem rather than a valuation problem, which is why the 2026 symptom is 0.05x DPI rather than mass markdowns.
6. 2023–2026 the AI concentration. Currently indistinguishable, on the evidence available, from either (a) a mobile/SaaS-style genuine platform build or (b) a crossover-style mark inflation. The discriminating evidence will be realisations, and as of 2026-09-15 there are almost none outside four companies.

Sector-specific false positives worth remembering. Clean-tech 1.0 (2006–11). Consumer-social after 2012 (hundreds of funded apps, a handful of outcomes). On-demand / "Uber for X" (2014–16). ICOs (2017–18). SPACs (2020–21) — a liquidity innovation sold as a democratised IPO route that produced systematic post-merger underperformance and is the closest structural analogue to today's retail-access products. Direct-to-consumer brands (2018–21). The Vision Fund model (2017–19). In every case, the misleading indicator was volume of capital deployed, and the correct indicator was cash returned per dollar called.

8. What is changing now (as of 2026-09-15)

The headline story and the real story are opposite in sign.

The headline story. H1 2026 was a record half-year for capital deployed ($510B Crunchbase / $560.4B KPMG globally; $412.7B in the US), Q2 2026 global exit value hit $1.9T exceeding the previous annual record, SpaceX executed the largest IPO in history, and VC posted a 17.1% one-year horizon IRR, among the highest of any private capital strategy.

The real story. That IRR is a valuation number, not a cash number. Fund asset values grew 21.6%, driven by AI markups (PitchBook via NLR, 2026-08-04). Meanwhile:

• 2021-vintage funds average 0.05x DPI — the lowest five-year DPI multiple this century.
• Net cash flow to LPs has been negative $202B since 2022, even as AUM rose.
• Cash distributed in the period ran at 11.7–12.8% of asset value against a long-run average of 18.7%; McKinsey's PE-side equivalent shows distributions at ~6% of AUM for the six months to June 2025 versus a 14% historical average, and a five-year rolling DPI-to-AUM ratio of ~10%, the lowest ever recorded.
• Carta's fund-level data shows 2019 and 2020 vintages with median DPIs "barely over zero" and fewer than half of funds having returned any capital at all; even 2017–18 vintages have under 20% of funds at 1x DPI.
• Cambridge Associates independently confirms the mechanism: calls have exceeded distributions by 1.6x since the start of 2022.

Four independent organisations (PitchBook, Carta, Cambridge Associates, McKinsey), using four different samples and two different methodologies, agree. This is the most robustly triangulated finding in the dossier.

The exit window is real but narrow. 44 US VC-backed IPOs had priced by 2026-07-30, on pace to exceed 2025's 50 — but the annual count has been under 50 every year since 2022, and four companies account for 93.5% of 2026 exit value. Aftermarket performance is poor: SpaceX was down 30.3% from its debut and Cerebras down 34.7% as of late July, against a Renaissance IPO Index up 12.7%. PitchBook's own summary is blunt: "A record year for liquidity has not been a record year for the stock performance of newly public companies." Morgan Stanley (2026-05-27) argues the opposite case — that Q1 2026 global issuance rose 43% to $256.8B with breadth across AI infrastructure, aerospace/defence, real estate, healthcare, metals and mining, and consumer retail. Both can be true: broad by sector count, narrow by value.

Secondaries have become the primary liquidity mechanism. H1 2026 secondary volume exceeded $120B, up 20% on the prior record, with GP-led deals at 53.7% of volume for the first time and single-asset continuation vehicles alone at $34B (Evercore, 2026-07-21). Secondaries dry powder fell 10% in six months, with the capital-overhang multiple approaching 1.0x — meaning the buy side is close to fully deployed, which caps how much more liquidity this channel can manufacture. Continuation vehicles reached $115B in 2025 against $35B in 2020 and now account for an estimated 14% of all sponsor-backed exits, which LPs expect to reach 20% now and 29% within five years (McKinsey). Venture-specific secondary trading runs at roughly $100B annually.

The median company's experience is the inverse of the aggregate. In Q2 2026 Carta recorded seed round counts down 20% while seed capital rose 37%, Series A counts down 12% with capital up 16%, Series B capital down 30%, and Series E+ rounds up 14% with capital up 142%. Pre-seed dollars were flat year-on-year ($3.19B vs $3.22B) across 11,500+ instruments versus 14,825 — a ~22% collapse in deal count with unchanged dollars, and a record $276k average instrument. Most damning: Crunchbase's graduation data shows the share of seed companies reaching a further round fell from 55%+ through 2020 to 24% for the 2023 cohort and 16% for the 2024 cohort, with median seed-to-A time now over two years. The AI/non-AI valuation gap is a chasm: median AI Series A pre-money ~$300M against ~$55M for non-AI (Carta, Q1 2026).

Exit economics have inverted. The median company exiting above $500M had raised $323.7M, against $157M a decade earlier, and the median valuation step-up at exit is now 15.5%, down from 62.8% in 2021. Companies are consuming twice the capital to produce a quarter of the step-up.

Retail access is the sector's strategic bet — and it is already stressed. The 2025-08-07 Executive Order "Democratizing Access to Alternative Assets for 401(k) Investors" and a subsequent DOL proposed rule establishing process-based fiduciary safe harbours target a ~$14T DC-plan pool covering 90M+ Americans. SEC staff in May 2025 withdrew a 23-year-old position that had capped registered closed-end funds at 15% exposure to private funds or restricted them to accredited investors with $25,000 minimums; the SEC's 2026 rulemaking agenda adds "Enhancing Retail Exposure to Private Markets" with an October 2026 target that the K&L Gates analysis calls "a statement of ambition rather than a schedule." Blackstone, Vanguard and Wellington announced joint public-private investment solutions on 2026-07-22.

But the liquidity promise in these vehicles broke in Q1 2026. Blue Owl Capital Corp. II closed quarterly redemptions on 2026-02-18 and instead returned ~30% of NAV through asset-sale-funded capital distributions. BlackRock TCP Capital took a 19% NAV writedown in Q4 2025. Blackstone's BCRED raised its quarterly redemption limit from 5% to 7.9% and executives contributed personal funds to meet redemptions in full. The $33B Cliffwater Corporate Lending Fund received redemption requests for 14% of shares against a 7% cap. Non-listed BDC redemptions hit 4.71% of NAV in Q4 2025, nearly tripling quarter on quarter, and funds with $1B+ NAV saw redemptions up 217% q/q. Blue Owl's own shares were down 60% year-on-year to $9.02 by 2026-03-25. The product being sold to retirement savers demonstrated its gating mechanics before it was broadly distributed.

An unauthorised grey market in pre-IPO AI shares emerged and was slapped down. On 2026-05-12 Anthropic declared void any sale or transfer of its stock or any interest in it offered through a named list of platforms including Open Doors Partners, Unicorns Exchange, Pachamama Capital, Lionheart Ventures, Sydecar, Upmarket, and new offerings on Hiive and Forge Global, explicitly prohibiting SPV acquisition of its stock. Tokenized Anthropic and OpenAI exposure fell roughly 40% the following day. One platform claimed aggregate demand exceeding $1T — an unverifiable marketing figure that nonetheless indicates the scale of retail appetite for exposure to four companies.

Manager formation has stalled. First-time fund formation is on pace for its lowest year since 2016. Funds under $50M were 67.7% of closings but 4% of dollars. Private debt is the only strategy with year-to-date fundraising growth (+27.3%); real estate is on track for its slowest year since 2012. Roughly $5T of NAV — 39% of all private-market NAV — now sits in funds seven years or older, after three consecutive years of negative net cash flow.

9. The five lists

Five most important current trends

1. The DPI drought: record marks, record IRR, almost no cash returned (T-11-01).
2. Concentration at every layer — company, deal, manager, exit (T-11-02).
3. Secondaries and continuation vehicles as the default liquidity mechanism (T-11-04, T-11-05).
4. The seed-to-Series-A graduation collapse beneath the mega-rounds (T-11-03).
5. Corporate and strategic capital displacing fiduciary VC at the frontier (T-11-06).

Five fastest-growing signals

1. Single-asset continuation vehicles ($34B in H1 2026 alone) (T-11-05).
2. Evergreen and semi-liquid wealth-channel vehicles (+30% AUM y/y; US semi-liquid PE $204B in 2025) (T-11-07).
3. DC-plan / 401(k) access to private assets (T-11-09).
4. Pre-IPO SPV and tokenized-share grey markets — and issuer enforcement against them (T-11-10).
5. China's AI-led funding rebound (+424% y/y in Q2 2026) (T-11-13).

Five trends most likely to affect businesses

1. Exit-cost inflation: $323.7M raised for a $500M exit, 15.5% step-up (T-11-15).
2. The graduation collapse — 16% of the 2024 seed cohort progressing (T-11-03).
3. Venture debt and structured capital as a default part of the stack (T-11-08).
4. Corporate strategic capital as the price-setter at late stage (T-11-06).
5. Manager consolidation: fewer, larger funds; sub-$50M funds effectively unbanked (T-11-14).

Five trends most likely to affect consumers

1. 401(k) and DC-plan exposure to illiquid, self-marked assets (T-11-09).
2. Retail evergreen vehicles and their gating mechanics (T-11-07, T-11-16).
3. Tokenized and SPV-wrapped pre-IPO exposure sold to non-institutional buyers (T-11-10).
4. Poor aftermarket performance of the 2026 IPO cohort, borne by public buyers (T-11-19).
5. Startup formation and employment effects as US business applications decline (T-11-03, T-11-18).

Overlaps, stated explicitly. T-11-03 appears in both "most important" and "affects businesses" because it is simultaneously the sector's structural story and the operative constraint on every founder. T-11-07 appears in both "fastest-growing" and "affects consumers" because the growth is the consumer exposure. T-11-05 and T-11-04 are the same underlying liquidity substitution viewed from the GP and LP sides respectively.

10. Overhyped / overlooked / cooling / reversing

Most overhyped

1. "The IPO window has reopened." The specific evidence that the hype outruns the substance: 44 VC-backed US IPOs YTD, with the annual count below 50 every year since 2022; four companies representing 93.5% of 2026 exit value; SpaceX down 30.3% and Cerebras down 34.7% from debut. A window through which four companies fit is a door.
2. "Democratised access to private markets." Marketing frames DC-plan and evergreen access as extending institutional returns to ordinary savers. The measured institutional return being extended is a 0.05x five-year DPI on the most recent large vintage, and the liquidity mechanism being extended demonstrably gated in Q1 2026 (Blue Owl OBDC II, Cliffwater 14%-vs-7%, BCRED 5%→7.9%). Label this claim marketing wherever it appears.
3. A 17.1% one-year IRR as evidence of recovery. IRR on unrealised, self-reported marks in a period when AI markups drove a 21.6% increase in fund asset values is a restatement of the markup, not independent evidence about it.

Most overlooked

1. ILPA-standard DPI reporting and the absence of a public benchmark. There is no independent, regulator-published, fund-level performance dataset. Every DPI figure in this dossier comes from a commercial vendor with a different, self-selected sample. Attention has missed this because the vendors are excellent at marketing their data and nobody's business model depends on pointing out that it is not audited.
2. The sub-$50M fund segment. 67.7% of fund closings and 4% of dollars. This is where most first cheques into most companies originate, and it is being defunded in silence because it is irrelevant to aggregate totals.
3. Corporate VC at 87.9% of US AI venture deal value. Almost all commentary treats 2026 as a venture story. It is substantially a corporate-strategic story, which has entirely different persistence properties: strategic capital withdraws on a strategy review, not on a fund cycle.
4. India's listing route. 47 IPOs in FY2025-26 against 31 the prior year, in a market whose total funding fell 18%. The exit mechanism and the funding cycle have decoupled in a way no Western market has managed.
5. Capital-overhang exhaustion in secondaries. Dry powder fell 10% in six months and the overhang multiple is approaching 1.0x. The liquidity valve everyone is relying on has a measurable limit, and almost nobody is tracking it.

Cooling — with the indicator that turned

1. Semi-liquid private credit's liquidity promise. Indicator: non-listed BDC redemptions tripled to 4.71% of NAV in Q4 2025; a $33B interval fund received requests for double its cap in Q1 2026.
2. First-time fund formation. Indicator: on pace for the lowest year since 2016; median time between fund closes now 1.7 years.
3. Non-AI venture. Indicator: seed instrument count down ~22% y/y with flat dollars; Q2 2026 global deal count the lowest since Q3 2017 (KPMG); non-AI Series A median pre-money ~$55M against ~$300M for AI.
4. US startup formation. Indicator: Census BFS business applications down 7.8% m/m in August 2026.
5. Real estate private funds. Indicator: on track for the slowest fundraising year since 2012.

Trends that may reverse — and the mechanism

• Concentration could reverse violently if AI marks are cut. Mechanism: a single large down round at a top-five private AI company forces auditors to question comparable marks across the complex; NAV falls; the 21.6% asset-value growth reverses; evergreen vehicles priced at NAV face redemptions into falling marks. The early indicator is a secondary transaction in a top AI name at a discount to the last primary round.
• Retail access could reverse on a single enforcement event. Mechanism: a gated retirement-plan product or a mis-sold tokenized SPV produces investor losses and Congressional attention; the SEC agenda item stalls; DOL safe harbours are narrowed. Indicator: the first FINRA or SEC enforcement action against a pre-IPO SPV promoter.
• Secondaries pricing could reverse from a seller's to a buyer's market. Mechanism: the overhang multiple crossing below 1.0x means buyers can dictate; discounts widen; LPs stop selling; the liquidity valve closes. Indicator: Jefferies/Evercore reporting average LP-led pricing falling below ~85% of NAV.
• Corporate VC could withdraw faster than any fund could. Mechanism: a hyperscaler capex reset, or an antitrust/accounting challenge to circular vendor-financing structures. Indicator: a quarter in which CVC share of AI deal value falls below 60%.

11. Risks and major uncertainties

Sector-specific risks.

• Mark integrity. Private marks are GP-determined, lightly audited, and currently rising on a small number of comparables. The FOMC has explicitly flagged "high AI-firm equity valuations and increased leveraged financing of infrastructure buildout" as a financial- stability concern (2026-07-29), which is a central bank naming this sector's exposure.
• Liquidity mismatch in retail vehicles. Daily-or-quarterly redemption rights against 7–10 year assets. Already tested, already gated.
• Rate sensitivity. The macro brief's sticky 3–4% inflation and hawkish-lean Fed mean the discount rate on long-duration private assets is not falling. Any thesis assuming a rate-driven re-rating of 2021 marks is unsupported.
• Denominator and duration risk at LPs. $5T of NAV in funds seven years or older after three years of negative net cash flow; endowments under proposed excise-tax pressure and already selling (Yale up to $6B, Harvard ~$1B, NYC pensions $5B).
• Concentration risk that is now systemic. A sector in which two companies took 43% of global funding and four account for 93.5% of exit value has no diversification left.
• Legal risk in the grey market. Transfer-restriction violations, void transfers, and tokenized exposure sold to non-accredited buyers.
• Adverse selection in continuation vehicles. A GP selling an asset to itself sets both sides of the price.

Genuine unknowns — and the distinction that matters.

Things we don't know but could find out: the true fund-level DPI distribution across the whole market (vendors hold it, samples differ, nobody publishes an audited universe); venture-specific secondary pricing as a discount to NAV; the actual number of venture-backed company shutdowns in 2026; the size of the SPV/tokenized pre-IPO market; the real revenue of the private companies carrying the sector's marks.

Things nobody can know: whether the 2023–26 AI cohort resolves like mobile/SaaS or like clean-tech 1.0 — the discriminating evidence is realisations that have not yet occurred and cannot be observed early; whether DC-plan money, once admitted, behaves patiently or panics, since it has never been tested; and whether a four-company exit market can broaden before the current cohort's fund lives expire.

12. Scenarios to 2030

Base — "liquidity engineering as a permanent feature." Secondaries, continuation vehicles, NAV lending and evergreen structures institutionalise as a parallel liquidity system. DPI recovers slowly and partially; the 2021 vintage ends below 1x for the median fund. Manager count shrinks; the top ~20 platforms take the majority of fee income; venture splits into a mega-platform oligopoly and a subscale seed cottage industry. Falsifiable early indicator: continuation vehicles exceed 20% of sponsor-backed exits by 2027, matching the LP forecast, while median 2021-vintage DPI stays below 0.5x through 2028.

Upside — "the realisation wave broadens." The AI cohort converts: 2027–28 sees 80–120 VC-backed IPOs annually with value spread across more than twenty issuers, aftermarket performance turns positive, and distributions normalise toward the 18.7% long-run rate. Seed graduation rates recover above 30%. LP recycling restarts and first-time fund formation recovers. Falsifiable early indicator: the top four companies' share of annual exit value falls below 50% in any calendar year, and the Renaissance IPO Index outperforms the S&P over a rolling twelve months.

Downside — "the marks come down." A large private AI company raises a flat or down round, or trades in the secondary market at a discount to its last primary. Auditors force comparable markdowns; the 21.6% NAV growth reverses; evergreen vehicles gate; DC-plan rulemaking stalls; secondaries buyers, already near 1.0x overhang, step back and discounts widen past 20%. The 2021–2023 vintages resolve as the worst since 2000. Falsifiable early indicator: two or more of the top-ten private AI companies transact in secondaries below their last primary mark within the same quarter.

Disruption — "the capital layer disintermediates itself." Corporate balance sheets, sovereign funds and tokenized retail rails supply so much of the marginal dollar that the closed-end fund becomes a minority structure at the frontier. CVC's 87.9% share of US AI deal value is already most of the way there. GPs become advisers and structurers rather than principals. Falsifiable early indicator: corporate and sovereign capital exceeds 50% of total late-stage deal value across all sectors, not just AI, for two consecutive years.

Regulatory — "the retail door opens, then is regulated shut." SEC completes its "Enhancing Retail Exposure to Private Markets" rulemaking and DOL finalises safe harbours; DC money flows in; a valuation or gating incident follows; Congress responds with disclosure, liquidity and valuation mandates that make the products uneconomic. Net effect is a compliance-heavy channel with lower margins than promised. Falsifiable early indicator: the first SEC or FINRA enforcement action against a retail private-markets feeder or pre-IPO SPV promoter, or a DOL final rule materially narrower than the proposal.

Failure — "a vintage-scale write-off with retail attached." The clean-tech 1.0 pattern repeats at ten times the scale, but this time with retirement savers in the vehicle. Net cash flow to LPs stays negative through 2028; endowments and pensions cut private allocations structurally; venture fundraising falls below $50B annually in the US; the sub-$50M fund segment disappears; startup formation compounds its decline. Falsifiable early indicator: a fourth and fifth consecutive year of negative aggregate net cash flow to LPs, combined with a major public pension announcing a structural reduction in private markets target allocation.

13. Data gaps and limitations

1. No independent DPI benchmark exists. Every distribution figure here comes from a commercial vendor — PitchBook, Carta, Cambridge Associates, McKinsey — each with a different, self-selected, survivorship-affected sample and no external audit. No regulator publishes fund-level performance. The direction is triangulated across four organisations; the level is not verifiable to Tier-A standard. This is the single largest gap in the sector.
2. Private marks are self-reported and untested. The 21.6% growth in fund asset values and the 17.1% one-year IRR rest on GP-determined valuations of companies with no public price. No independent mark-to-market audit trail exists.
3. No reliable startup shutdown statistic. Search results for startup failure rates were dominated by Tier-C content farms recycling a decades-old "90% fail" claim; these were excluded entirely. Carta's dissolution series was not published for 2026 at the time of research. Census BFS measures formation, not venture-backed failure, and BLS Business Employment Dynamics lags by roughly two quarters. The down-round and shutdown rate beneath the AI mega-rounds is therefore the least-evidenced part of this dossier — we have Carta's 11.4% Q1 2026 down-round rate and nothing comparable for Q2 or Q3.
4. Conflicting global totals. Crunchbase $510B vs KPMG $560.4B for H1 2026 global VC; Crunchbase 5,000+ vs KPMG 8,440 Q2 deals. Definitional, not erroneous, per the macro brief — but it means no single "global VC" number should be quoted without its source.
5. Conflicting Asia totals. Crunchbase puts Asia Q2 2026 at $42.8B; KPMG (macro brief) at $50.8B. Both are cited; neither is treated as truth.
6. SpaceX IPO valuation is reported three different ways. $1.5T at the $135 fixed offer price with a ~$2.1T day-one close (Crunchbase, macro brief), $1.77T (Crunchbase exits coverage, 2026-07-02), and ~$1.7T (PitchBook/secondary commentary). The divergence is almost certainly basic-versus-fully-diluted share count and offer-versus-close timing, but it is unresolved and we assert no single figure.
7. Secondaries pricing is not public. Volume is reported by Jefferies and Evercore; pricing as a percentage of NAV, especially for venture, sits behind registration walls and licensed PDFs. Phase 2 will need budget here.
8. The SPV/tokenized pre-IPO market is unquantified. The only figure available — one platform's claim of ">$1T aggregate demand" — is unverifiable vendor marketing and is labelled as such.
9. No Chinese-language primary sources were consulted. The +424% y/y China figure rests on a single Anglophone vendor (Crunchbase). Given that China was the largest single driver of Asia's rebound, this is a material gap. Regional sources used were tech.eu (Amsterdam), Dealroom (Amsterdam) and Tracxn (Bengaluru).
10. DOL rule chronology is uncertain. One law-firm alert is dated 2026-02-19 while describing a proposed rule dated 2026-03-30 with a comment period closing 2026-06-01. The executive order (2025-08-07) and the ~$14T/90M-participant scale are solid; the exact rule date is flagged unverified.
11. Anthropic and OpenAI listing status unconfirmed as of 2026-09-15, per the macro brief's own caution. PitchBook refers to "planned offerings"; we assert nothing.
12. The SpaceX/Anysphere (Cursor) $60B acquisition is reported by a single source (Crunchbase, 2026-07-02) and is recorded as single_source pending corroboration.
13. Blackstone Q2 2026 AUM not verified from primary IR. The IR press release URL returned 404 and the news index did not surface the figure; a secondary headline cited $1.34T, which is not asserted here.
14. Search budget. 24 of ~25 permitted WebSearch calls were used; verification was weighted toward primary and platform-native fetches. Accelerator-cohort data and sovereign-wealth VC allocation data were deprioritised and remain thin.

14. Ranking scorecard

15. Sources

1. "Venture capital's current recovery is all IRR, no DPI," PitchBook, https://pitchbook.com/news/articles/venture-capitals-current-recovery-is-all-irr-no-dpi, 2026-08-04, Tier B.
2. "Q2 2026 PitchBook-NVCA Venture Monitor," PitchBook & National Venture Capital Association, https://pitchbook.com/news/reports/q2-2026-pitchbook-nvca-venture-monitor, 2026-07-09, Tier A.
3. "Q2 2026 Global Private Market Fundraising Report," PitchBook, https://pitchbook.com/news/reports/q2-2026-global-private-market-fundraising-report, 2026-09-03, Tier B.
4. "VC Fund Performance: Q1 2026," Carta, https://carta.com/data/vc-fund-performance-q1-2026/, 2026-06-04, Tier A.
5. "State of Private Markets: Q1 2026," Carta, https://carta.com/data/state-of-private-markets-q1-2026/, 2026-05-29, Tier A.
6. "State of Pre-Seed: Q2 2026," Carta, https://carta.com/data/state-of-pre-seed-q2-2026/, 2026-08-13, Tier A.
7. "US Private Equity & Venture Capital Benchmark Commentary (1H 2025 data)," Cambridge Associates, https://www.cambridgeassociates.com/wp-content/uploads/2025/12/2026-01-US-PE-VC-Benchmark-Commentary-1H2025-Data-PUBLIC.pdf, 2026-01-05, Tier A.
8. "Global Private Markets Report 2026: Private Equity," McKinsey & Company, https://www.mckinsey.com/industries/private-capital/our-insights/global-private-markets-report/private-equity, 2026-02-10, Tier B.
9. "The Venture Market Is Not Recovering — It Is Reorganizing: What the Q2 2026 PitchBook Data Shows," National Law Review, https://natlawreview.com/article/venture-market-not-recovering-it-reorganizing, 2026-08-04, Tier B.
10. "5 charts: VC-backed IPOs hit a record 2026, but the aftermath is dimmer," PitchBook, https://pitchbook.com/news/articles/5-charts-vc-backed-ipos-hit-a-record-2026-but-the-aftermath-is-dimmer, 2026-07-30, Tier B.
11. "Global startup exits from IPOs and M&A soar on AI in Q2/H1 2026," Crunchbase News, https://news.crunchbase.com/venture/global-startup-exits-ipo-ma-soar-ai-q2-h1-2026/, 2026-07-02, Tier B.
12. "In Charts: Seed Deals Keep Getting Bigger As Odds Of Reaching Series A Fall Dramatically," Crunchbase News, https://news.crunchbase.com/seed/data-bigger-deals-longer-seriesa-2026/, 2026-05-26, Tier B.
13. "China And AI Lead Asia's Startup Funding To Multiyear Peak In Q2," Crunchbase News, https://news.crunchbase.com/venture/data-china-ai-lead-asia-startup-funding-peak-q2-2026/, 2026-07-16, Tier B.
14. "Continuation funds drive record H1 for secondary market" (Evercore data, via Secondaries Investor), Yahoo Finance, https://finance.yahoo.com/markets/stocks/articles/continuation-funds-drive-record-h1-223014268.html, 2026-07-21, Tier B.
15. "Secondaries volume extends record-breaking run in H1," Secondaries Investor (PEI Group), https://www.pei-secondariesinvestor.com/secondaries-volume-extends-record-breaking-run-in-h1/, 2026-07-21, Tier B (paywalled body).
16. "The Ascendance and Implications of Evergreen Funds in Private Markets," MSCI, https://www.msci.com/research-and-insights/blog-post/the-ascendance-and-implications-of-evergreen-funds-in-private-markets, 2026-03-03, Tier B.
17. "Private Credit Confronts the Limitations of the 'Semi-Liquid' Label," WealthManagement.com, https://www.wealthmanagement.com/alternative-investments/private-credit-confronts-the-limitations-of-the-semi-liquid-label, 2026-03-25, Tier B.
18. "Democratizing Access to Alternative Assets for 401(k) Investors" (Executive Order), The White House, https://www.whitehouse.gov/presidential-actions/2025/08/democratizing-access-to-alternative-assets-for-401k-investors/, 2025-08-07, Tier A.
19. "Department of Labor Advances Proposed Rule Expanding 401(k) Access to Private Capital," Shulman Rogers, https://www.shulmanrogers.com/legal-alert-department-of-labor-advances-proposed-rule-expanding-401k-access-to-private-capital/, 2026 (exact date unverified), Tier B.
20. "Investing in Private Funds Beyond 15%: SEC Staff Opens the Door for Retail Closed-End Funds," Stradley Ronon, https://www.stradley.com/publications/investing-in-private-funds-beyond-15-sec-staff-opens-the-door-for-retail-closed-end-funds, 2025-05-22, Tier B.
21. "The SEC's New Rulemaking Agenda: A Deregulatory Road Map for Advisers and Funds," K&L Gates, https://www.klgates.com/thought-leadership/The-SECs-New-Rulemaking-Agenda-A-Deregulatory-Road-Map-for-Advisers-and-Funds-7-8-2026, 2026-07-08, Tier B.
22. "Anthropic warns investors against secondary platforms offering access to its shares," TechCrunch, https://techcrunch.com/2026/05/12/anthropic-warns-investors-against-secondary-platforms-offering-access-to-its-shares/, 2026-05-12, Tier B.
23. "Anthropic, OpenAI tokens plunge as AI firms say pre-IPO share transfers are invalid," CoinDesk, https://www.coindesk.com/markets/2026/05/13/anthropic-openai-tokens-plunge-nearly-40-as-ai-firms-warn-spv-transfers-are-invalid, 2026-05-13, Tier B.
24. "Inside The Murky Market Selling Pre-IPO SpaceX And OpenAI Shares," Forbes (Phoebe Liu), https://www.forbes.com/sites/phoebeliu/2026/05/26/inside-the-murky-market-selling-pre-ipo-spacex-and-openai-shares/, 2026-05-26, Tier B.
25. "Runway Growth Capital and PitchBook Release 2025-2026 Venture Debt Review," Runway Growth Capital / PR Newswire, https://runwaygrowth.com/runway-growth-capital-and-pitchbook-release-2025-2026-venture-debt-review-venture-debt-hits-record-68-8-billion/, 2026-05-28, Tier B.
26. "More capital. Fewer deals. What H1 2026 tells us about European tech," Tech.eu (Amsterdam), https://tech.eu/2026/07/30/more-capital-fewer-deals-what-h1-2026-tells-us-about-european-tech/, 2026-07-30, Tier B — regional source.
27. "UK captures record 48% of European VC funding in 2026," Dealroom.co (Amsterdam), https://dealroom.co/resources/uk-share-european-vc/, 2026 (undated page), Tier B — regional source, undated: true.
28. "Indian Startup Funding Falls 18% to $11.7B in FY26" (Tracxn India Tech Annual Funding Report 2026), Analytics Insight, https://www.analyticsinsight.net/amp/story/news/indian-startup-funding-falls-18-to-117b-in-fy26-tracxn-report, 2026-04-09, Tier B — regional source.
29. "Business Formation Statistics, August 2026," US Census Bureau, https://www.census.gov/econ/bfs/current/index.html, released 2026-09-11, Tier A.
30. "Q3 2026: Fewer Deals, Bigger Bets — How Corporate Capital Is Concentrating US AI Venture Activity," PitchBook, https://pitchbook.com/news/reports/q3-2026-fewer-deals-bigger-bets-how-corporate-capital-is-concentrating-us-ai-venture-activity, 2026-07-21, Tier B.
31. "A Larger, Broader IPO Market Takes Shape in 2026," Morgan Stanley, https://www.morganstanley.com/insights/articles/ipo-market-scale-breadth-2026, 2026-05-27, Tier B.
32. "US Endowment Dislocation Creates Compelling Secondary Opportunity," Adams Street Partners, https://www.adamsstreetpartners.com/insights/us-endowment-dislocation-creates-compelling-secondary-opportunity/, 2025-05-28, Tier B (GP with commercial interest in secondaries).
33. "NYC pension system sells $5 billion in private equity assets on the secondary market," Pensions & Investments, https://www.pionline.com/pension-plans/nyc-pension-system-sells-5-billion-private-equity-assets-secondary-market-shedding/, 2025-05-29, Tier B.
34. "Carta Data Points to Larger Venture Rounds and Tighter Selection in Q2 2026," TipRanks (reporting Carta primary data), https://www.tipranks.com/news/private-companies/carta-data-points-to-larger-venture-rounds-and-tighter-selection-in-q2-2026, 2026-08-11, Tier C (underlying data Tier A — Carta).
35. "Blackstone press releases," Blackstone, https://www.blackstone.com/news/press/, accessed 2026-09-15, Tier A.
36. Macro Context Brief (internal), verified 2026-09-15, carrying Crunchbase, KPMG Venture Pulse and Federal Reserve FOMC figures, Tier A/B composite.

More research in this sector · Explore trend records