Across 2024–2025, median VC TVPI is holding near 1.6x while median DPI has fallen to roughly 0.3x — a 5x gap between paper value and cash actually returned, the widest divergence between the two metrics in over a decade.
That gap is the whole story of this cycle. Funds look fine on a TVPI line and broke on a DPI line. The longer answer — by vintage, by quartile, and by what LPs do with it — is more interesting.
2024–2025 Venture Capital TVPI vs DPI Benchmarks: The Side-by-Side
TVPI counts total value — cash returned plus remaining paper NAV — divided by capital called. DPI counts only the cash. In 2024–2025 the median VC fund carries about 1.6x TVPI but just 0.3x DPI, meaning roughly 80% of reported value is still unrealized marks. Top-quartile funds reach 2.5x TVPI yet only 0.6x DPI. The two numbers have never been further apart in the modern data set.
| Attribute | TVPI | DPI |
|---|---|---|
| What it measures | Total value (cash + paper NAV) ÷ paid-in | Cash distributed ÷ paid-in |
| 2024–25 median | ~1.6x | ~0.3x |
| 2024–25 top quartile | ~2.5x | ~0.6x |
| 2024–25 bottom quartile | ~1.0x | ~0.1x |
| Can be inflated by marks? | Yes — relies on GP valuations | No — cash is cash |
| What LPs trust most | Directional upside signal | Realized, unforgeable return |
| Peaks in fund life | Years 4–7 (mark-ups) | Years 7–12 (exits) |
Figures are blended estimates from Carta, PitchBook, and Cambridge Associates benchmark data through year-end 2025. You can explore the live ranges on the VC Performance dashboard.
TVPI vs DPI Benchmarks by Vintage Year
The TVPI–DPI gap is almost entirely a function of fund age. Older vintages have had time to exit, so their DPI has caught up to TVPI. The 2018–2021 vintages — which raised at peak valuations and then hit a frozen exit window — show the widest spread. Here is how the benchmark ranges look across vintages as of late 2025.
| Vintage | Median TVPI | Median DPI | Gap |
|---|---|---|---|
| 2012–2013 | 2.4x | 2.0x | 0.4x |
| 2014–2015 | 2.1x | 1.5x | 0.6x |
| 2016–2017 | 1.9x | 0.9x | 1.0x |
| 2018–2019 | 1.7x | 0.4x | 1.3x |
| 2020–2021 | 1.4x | 0.15x | 1.25x |
| 2022–2023 | 1.2x | 0.05x | 1.15x |
Read down the DPI column and the exit drought is obvious. A 2013 vintage has returned 2.0x in cash; a 2021 vintage has returned 15 cents on the dollar after four-plus years. The 2020–2021 cohort is the danger zone — high paper marks set at 2021 peaks, almost no realized cash, and a clock that is running. If those marks compress on the way to exit, the 1.25x gap closes downward, not upward.
Why DPI Cratered While TVPI Held Up
TVPI stayed near 1.6x because GPs kept marking AI and late-stage positions at or above last-round prices, and a handful of AI mega-rounds dragged blended marks higher. DPI fell because there was nowhere to sell. Three structural forces drove the divergence:
Frozen IPO window
Tech IPO volume stayed roughly 80% below the 2021 peak through 2024, so the largest source of DPI simply wasn't open.
M&A slowdown
Higher rates and antitrust scrutiny cut strategic acquisitions; total US VC distributions ran near $150B in 2024 versus far higher contributions.
Mark discipline lag
GPs hold private marks at last-round prices for quarters after public comps fall, propping TVPI even as exit value erodes.
AI mark concentration
A few AI names carried at 10–50x revenue inflate blended TVPI while contributing zero realized cash to DPI.
Net of all this, the LP cash-flow picture in 2024–2025 was negative: institutions wrote more capital calls than they received in distributions for the third straight year. That is why DPI — not TVPI — became the metric that decides re-ups. You can cross-check the public side of this drought on the Tech IPO tracker.
TVPI vs DPI: Which Number LPs Actually Trust Now
After the 2021–2022 markdowns erased billions in paper TVPI, institutional LPs stopped treating TVPI as a return and started treating it as a hypothesis. The reframe is simple: TVPI tells you what a GP believes; DPI tells you what they delivered. In diligence today, a fund pitching 2.0x TVPI with 0.1x DPI in year five gets read as unproven, not strong.
What Builds LP Trust
- ✓ DPI above 0.5x by year six
- ✓ Realized exits, not just step-up rounds
- ✓ Marks within 20% of nearest public comps
- ✓ Consistent distribution cadence across funds
What LPs Now Discount
- ✕ High TVPI driven by one unrealized markup
- ✕ Marks held flat while comps fell 40%+
- ✕ Zero DPI past year seven
- ✕ "Paper" IRR with no cash behind it
What the 2024–2025 Benchmarks Mean for Your Fund
If you are a GP, the benchmark math is unforgiving but clear. A 1.6x TVPI is now table stakes; it no longer signals a strong fund because everyone marked up in 2021. Your differentiation is the DPI line. Returning even 0.5x in cash by year six — through secondaries, strip sales, or disciplined early exits — puts you in rare company this vintage.
For LPs, the takeaway is to underwrite the gap, not the headline. A 2020 fund at 1.5x TVPI and 0.1x DPI is a bet that the 1.4x of paper survives the trip to liquidity. Some will; the 2020–2021 marks set at peak are the most likely to compress. Anchor your re-up decisions on realized cash and distribution cadence — compare any fund against its vintage peers on the VC Performance dashboard before you trust a single TVPI number.
TVPI is a forecast. DPI is a fact.
In 2024–2025, the median fund forecast 1.6x and delivered 0.3x — and only one of those numbers gets you a re-up.
Track TVPI, DPI, and IRR benchmarks by vintage on the VC Performance Dashboard at Value Add VC. Originally published in the Trace Cohen newsletter.