The DPI drought: whether the private-capital layer returns cash to LPs, and by what mechanism

Six alternative branches. These are subjective probabilities from the seed, not observed frequencies or investment recommendations.

Why this trend

Load-bearing assumption

That distributions to paid-in capital, rather than internal rate of return or net asset value, is the variable that determines LP re-up behaviour, manager formation and the supply of new capital to every downstream sector.

Preconditions

Not recorded.

base · 40% · band

Liquidity engineering becomes a permanent feature rather than a bridge. Secondaries, continuation vehicles, NAV lending and evergreen structures institutionalise as a parallel liquidity system that converts some NAV into cash at a discount, and the 2021 vintage ends below 1x DPI for the median fund. Manager count shrinks, the top twenty platforms take the majority of fee income, and venture splits into a mega-platform oligopoly and a subscale seed cottage industry. Net cash flow to LPs turns modestly positive around 2028 without returning to the 18.7% long-run distribution rate.

Mechanism

GPs, unable to exit through IPO or strategic M&A outside four companies, sell assets to vehicles they also manage or to secondary buyers. LPs accept discounts because they need cash for existing commitments. Because the buy side is funded by the same institutions, the system recycles rather than injects capital, so aggregate net cash flow improves slowly and the distribution rate stays structurally below history. The decelerating element is secondaries capacity: dry powder has already fallen 10% in six months with the overhang multiple near 1.0x.

Preconditions, early indicators and assumptions

Early Indicators

Indicator
Continuation vehicles exceed 20% of sponsor-backed exits, matching the LP forecast, while median 2021-vintage DPI stays below 0.5x
Source
S-11-10 McKinsey Global Private Markets Report; S-11-03 Carta Data Desk
Would Be Visible By
2028-02
Indicator
Cambridge Associates reports calls still exceeding distributions for US VC on 1H 2026 data
Source
S-11-04 Cambridge Associates Benchmark Commentaries
Would Be Visible By
2027-03
Indicator
Global secondary volume between $220bn and $280bn for calendar 2026 with GP-led above 50% of volume
Source
S-11-09 Evercore and Jefferies secondary market reviews
Would Be Visible By
2027-03

Affected Industries

What Businesses Should Do

Founders should plan for a seven-to-ten-year private life and price equity accordingly; LPs should model commitments off distributions actually received rather than off NAV; emerging managers should assume re-ups are funded from allocation headroom, not from recycled cash, and size funds to that.

Precedence Note

Distinguished from upside by the breadth of realisations, not their size: base allows a record exit-value year provided the top four companies still account for more than half of it. Distinguished from downside by the direction of marks, which hold in base and fall in downside.

Would Change Our Mind

The top four companies' share of annual exit value falling below 50% in any calendar year while aggregate exit value also rises.

upside · 13% · band

The realisation wave broadens. The AI cohort converts into cash: 2027-28 produce 80 to 120 VC-backed IPOs a year 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% and first-time fund formation restarts as LPs receive recyclable capital.

Mechanism

The named constraint that releases is aftermarket performance. SpaceX at minus 30.3% and Cerebras at minus 34.7% from debut against a Renaissance IPO Index up 12.7% is what currently deters the second tier of issuers from filing. If the 2026 cohort's aftermarket turns, bankers reopen the pipeline to companies below the mega-cap tier, exit value broadens by issuer count rather than by size, and LP cash flow turns positive quickly because distributions are stock as well as cash.

Preconditions, early indicators and assumptions

Early Indicators

Indicator
The Renaissance IPO Index outperforming the S&P 500 over a rolling twelve months
Source
S-11-02 PitchBook News & Analysis
Would Be Visible By
2027-06
Indicator
More than 80 US VC-backed IPOs priced in a calendar year, with the top four companies below 50% of exit value
Source
S-11-01 PitchBook-NVCA Venture Monitor
Would Be Visible By
2028-02
Indicator
Seed-to-Series-A graduation for the 2025 cohort reported above 25%
Source
S-11-07 Crunchbase News
Would Be Visible By
2028-06

Affected Industries

What Businesses Should Do

Companies with two years of runway should prepare listing-ready financials now rather than waiting for confirmation, because this branch opens and closes faster than an audit cycle; LPs should hold pacing discipline rather than chasing the reopening.

Precedence Note

Requires broadening by issuer count. A record exit-value year driven by two or three additional mega-listings resolves to base, not upside.

Would Change Our Mind

Two consecutive quarters in which VC-backed IPO count rises while median first-day-to-90-day performance stays negative.

downside · 20% · band

The marks come down. A large private AI company raises a flat or down round, or trades in the secondary market below its last primary, and auditors force comparable markdowns across the complex. The 21.6% growth in fund asset values reverses, evergreen vehicles priced at NAV face redemptions into falling marks, DC-plan rulemaking stalls, and secondaries buyers already near a 1.0x overhang step back so discounts widen past 20%. The 2021-2023 vintages resolve as the worst since 2000.

Mechanism

Private marks are GP-determined and currently rest on a small number of comparables. One observable transaction below the last primary in a top-ten AI name removes the comparable that supports the others. NAV falls; evergreen vehicles that sold daily-or-quarterly liquidity against seven-to-ten-year assets face redemptions into a falling price; the transmission channel the FOMC named on 2026-07-29 — high AI-firm equity valuations and leveraged infrastructure financing — runs through exactly this path.

Preconditions, early indicators and assumptions

Early Indicators

Indicator
Two or more of the top-ten private AI companies transacting in secondaries below their last primary mark within the same quarter
Source
S-11-09 Evercore and Jefferies secondary market reviews; S-11-02 PitchBook News
Would Be Visible By
2027-09
Indicator
Average LP-led secondary pricing falling below 85% of NAV
Source
S-11-09 Evercore and Jefferies secondary market reviews
Would Be Visible By
2027-06
Indicator
Carta's quarterly down-round rate rising above 18% from the 11.4% recorded for Q1 2026
Source
S-11-03 Carta Data Desk
Would Be Visible By
2027-06

Affected Industries

What Businesses Should Do

Any company whose last round was priced in 2024-2026 should model a flat round as the base case and secure eighteen months of runway before the comparable repricing arrives; LPs holding evergreen vehicles should read the gating mechanics before the queue forms, not after.

Precedence Note

Requires an observable transaction below a prior primary mark. Slower fundraising, longer processes and flat rounds without a secondary print stay in base.

Would Change Our Mind

A top-ten private AI company raising an up round at a premium above 40% with a named new lead investor and disclosed terms.

disruption · 12% · band

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. Corporate venture arms already account for a record 87.9% of US AI venture deal value, and NVIDIA is the single largest corporate investor by deal value. GPs become advisers, structurers and co-investment syndicators rather than principals, and the management-fee pool contracts even as deployment grows.

Mechanism

Strategic capital buys access to supply, distribution or compute rather than financial return, so it clears at prices a fiduciary fund cannot match. Two independent classes of buyer are already pursuing this — hyperscalers and infrastructure vendors on one side, sovereign wealth funds on the other — which meets the framework's test that one company is a project and two is a sector. As the marginal price-setter stops being a fund, the fund's role narrows to smaller cheques and earlier stages, and its fee base narrows with it.

Preconditions, early indicators and assumptions

Early Indicators

Indicator
Corporate and sovereign capital exceeding 50% of total late-stage deal value across all sectors, not just AI, for two consecutive years
Source
S-11-01 PitchBook-NVCA Venture Monitor
Would Be Visible By
2028-08
Indicator
CVC share of US AI deal value staying above 80% through four consecutive quarters
Source
S-11-01 PitchBook-NVCA Venture Monitor; S-11-07 Crunchbase News
Would Be Visible By
2027-08
Indicator
Form D filings showing a rising share of offerings with a single strategic investor as sole purchaser
Source
S-11-06 SEC EDGAR Form D filings
Would Be Visible By
2027-12

Affected Industries

What Businesses Should Do

Founders taking strategic capital should price the commercial terms separately from the equity terms and assume the investor's strategy review, not its fund life, sets the horizon; fund managers should build co-investment and structuring capability rather than defending fund size.

Precedence Note

Requires the shift to extend beyond AI. CVC dominance inside AI alone, with fiduciary VC still setting late-stage prices elsewhere, stays in base.

Would Change Our Mind

A quarter in which CVC share of AI deal value falls below 60% while total AI deal value holds.

regulatory · 10% · band

The retail door opens, then is regulated shut. The SEC completes its 'Enhancing Retail Exposure to Private Markets' rulemaking and the DOL finalises process-based fiduciary safe harbours; defined-contribution money begins flowing toward a roughly $14 trillion pool covering more than 90 million participants. A valuation or gating incident follows, Congress responds with disclosure, liquidity and valuation mandates, and the channel ends up compliance-heavy with lower margins than promised. Direction here is genuinely two-sided: the same rulemaking could as easily be narrowed at proposal stage as completed.

Mechanism

The instruments are the 2025-08-07 Executive Order, the DOL proposed rule establishing safe harbours, the SEC's 2026 agenda item with an October 2026 target, and the May 2025 staff withdrawal of the 23-year-old 15% closed-end-fund position. The four-stage distinction matters: an agenda item is not a proposal, a proposal is not a final rule, and a final rule is not enforcement. The product's gating mechanics were demonstrated in Q1 2026 before broad distribution, which is what makes a post-inflow incident foreseeable rather than speculative.

Preconditions, early indicators and assumptions

Early Indicators

Indicator
An SEC final rule under the 'Enhancing Retail Exposure to Private Markets' item published in the Federal Register
Source
S-11-12 SEC rulemaking agenda and proposed rules
Would Be Visible By
2027-12
Indicator
The first SEC or FINRA enforcement action against a retail private-markets feeder or pre-IPO SPV promoter
Source
S-11-12 SEC rulemaking and enforcement releases
Would Be Visible By
2027-12
Indicator
A DOL final rule materially narrower than the proposal, or its withdrawal
Source
S-11-13 DOL / EBSA regulatory agenda
Would Be Visible By
2027-06

Affected Industries

What Businesses Should Do

Distributors should build valuation and liquidity disclosure to a standard that survives a mandate rather than to the standard the current rules require; plan sponsors should treat the 0.05x five-year DPI on the most recent large vintage as the return being extended, and document that they did.

Precedence Note

Regulatory outranks base when a final rule issues, even if flows look base-like. A staff statement or a withdrawn position is not a rule and does not move this branch.

Would Change Our Mind

The SEC agenda item lapsing across two consecutive Reg Flex agendas with no proposal issued.

failure · 5% · band

A vintage-scale write-off with retirement savers attached. The clean-tech 1.0 pattern repeats at roughly ten times the scale: net cash flow to LPs stays negative through 2028, endowments and pensions cut private allocations structurally rather than tactically, US venture fundraising falls below $50bn annually, the sub-$50m fund segment disappears and startup formation compounds its decline. The asset that is left is a large stock of seven-year-plus NAV with no buyer at the carried price.

Mechanism

The thesis that fails is that the 2023-26 AI cohort resolves like mobile and SaaS rather than like clean-tech 1.0, where most of roughly $25bn deployed 2006-2011 was written off and the tell at the time was that returns depended on policy and commodity prices rather than unit economics. Holders are LPs, and increasingly DC-plan participants, in funds already seven years or older holding roughly $5tn of NAV — 39% of all private-market NAV. Recovery of that asset class historically takes a full cycle, roughly seven to ten years, as it did after 2001.

Preconditions, early indicators and assumptions

Early Indicators

Indicator
A fourth and fifth consecutive year of negative aggregate net cash flow to LPs
Source
S-11-04 Cambridge Associates; S-11-02 PitchBook News
Would Be Visible By
2028-03
Indicator
A major public pension announcing a structural reduction in its private-markets target allocation
Source
S-11-23 University endowment and public pension disclosures
Would Be Visible By
2027-12
Indicator
US venture fundraising below $50bn for a calendar year
Source
S-11-01 PitchBook-NVCA Venture Monitor
Would Be Visible By
2029-02

Affected Industries

What Businesses Should Do

Operating companies should assume the next round is the last one and reach cash-flow breakeven; institutions should stress-test their private-markets allocation against a 2001-style realisation drought rather than against a drawdown in marks.

Precedence Note

Failure outranks downside on occurrence. Downside is marks falling with the system still functioning; failure is the allocation decision itself being reversed at the institutional level.

Would Change Our Mind

Two consecutive years of positive aggregate net cash flow to LPs, at any level.

Additional scenario notes

Scope Note

Sector 11 is a cross-cutting capital layer, not a vertical. Its deployment figures are the same dollars already counted inside sectors 01, 03, 06 and others and must never be summed with them; only fundraising, secondaries, continuation-vehicle, evergreen and AUM measures are unique to this layer. Every branch below is about cash reaching limited partners, not about marks.

Probabilities Sum

What Must Be True To Grow

Realisations, not markups, reach LPs as cash or freely tradable stock, because only distributions can be re-committed
Exit value broadens beyond the four companies that accounted for 93.5% of 2026 exit value
Secondaries dry powder is replenished faster than it is deployed, since the overhang multiple is already near 1.0x
Seed-to-Series-A graduation recovers materially from 16% for the 2024 cohort, or the funnel that feeds later stages empties
The cost of capital stops rising: the FOMC held at 3.50-3.75% on 2026-07-29 with three dissents in favour of a hike, and the 2026-09-16 decision was unresolved at the corpus date

What Could Stop It

Capital cost (framework §9.4): a rate environment in which long-duration private assets do not re-rate, already visible in sub-$50m funds being defunded
Measurement revision: every DPI figure in the corpus comes from a commercial vendor with a self-selected, survivorship-affected sample and no external audit
Demand deflation: the pipeline of exit candidates is speculative outside a handful of names, as the 44 IPOs and 93.5% value concentration show
Regulatory reversal: the DC-plan and retail-access door narrowing after an incident
Attention withdrawal: corporate strategic capital withdraws on a strategy review rather than on a fund cycle, and it is 87.9% of US AI venture deal value

Uncertain Assumptions

Text
Vendor DPI figures approximate the true fund-level distribution, despite four different self-selected samples and no audit
Confidence
low
Load Bearing
false
If Wrong
The level of the drought is unknown in either direction; the direction survives because four organisations agree, but every threshold forecast in this set becomes unresolvable
Text
Secondary and continuation-vehicle proceeds continue to be counted as distributions when cash reaches the LP, and not before
Confidence
medium
Load Bearing
false
If Wrong
DPI recovers on paper without LPs receiving spendable cash, and base and failure become observationally similar for two to three years
Text
Distributions, not IRR, drive LP re-up and manager formation behaviour
Confidence
high
Load Bearing
true
If Wrong
The 17.1% one-year horizon IRR sustains fundraising on its own, first-time fund formation recovers without a cash cycle, and the whole scenario set is built on the wrong variable
Text
The 2023-26 AI cohort is currently indistinguishable from either a genuine platform build or a crossover-style mark inflation
Confidence
high
Load Bearing
false
If Wrong
If it is in fact distinguishable today, probability mass moves decisively to either upside or failure and the base case is too wide

Authored

2026-09-15