VENTURE CAPITAL DEEP DIVE · VC-ONLY DATA · 11 YEARS · 25 UNICORNS

The $500 Billion
Illusion

Venture capital funds deployed $880B+ during ZIRP. They claim $1.2 trillion in assets. But only ~$200B came back as actual cash. The rest is paper.

This analysis focuses on venture capital specifically — not private equity buyout or hedge funds. VC has a unique return profile: the J-curve means funds typically take 5-10 years to return capital. Young vintages (2020-2022) should have low DPI. The question is whether ZIRP-era valuations will ever convert to real exits.

TVPI Erosion
$0B

Peak to current paper loss (VC only)

Still on Paper
$0B

Unrealized RVPI (2018-22 VC)

Actual Cash Back
$0B

DPI on $1,064B deployed

2021 Unrealized
0%

Of TVPI is paper value

Sources: Cambridge Associates US VC Index, CalSTRS FYE 2025, CalPERS PEP Review, PitchBook-NVCA, Carta Q4 2025, Preqin, SEC FilingsData as of Q4 2025
VC ONLY: 2021 VINTAGE 0.5% MEDIAN IRR$329.6B VC DEPLOYED AT PEAK ZIRPONLY 25% OF 2021 VC FUNDS DISTRIBUTING60% OF LPs NOW WEIGHT DPI OVER TVPIWeWork: $47B → $0 (BANKRUPT)Hopin: $7.75B → $15M (99.8% LOSS)VC DISTRIBUTION RATE: 5% vs 16.8% AVGOpenAI: $29B → $300B (THE EXCEPTION)$311.6B VC DRY POWDER SITTINGVC IPO MULTIPLES: 5x vs 12.3x IN 2021

First, Understand the J-Curve

VC funds are designed to lose money early and make it back later. The 'J-curve' describes this arc: capital calls and fees drag returns negative in years 1-3, then exits drive value in years 5-10. Low DPI in young funds is expected — the question is whether these particular vintages will ever climb out.

Standard VC industry model — typical 10-year fund life with 2-3 year investment period, 5-7 year harvest

The Typical VC J-Curve (Idealized Net TVPI)

How a well-performing VC fund's returns typically evolve over its 10-year life

Key Phases

Years 0-3: The Dip

Capital calls, management fees, no exits. DPI near zero is normal.

Years 4-6: Inflection

First exits. Value creation becomes visible. TVPI climbs. DPI starts trickling.

Years 7-10: Harvest

IPOs, M&A, secondaries. DPI accelerates. TVPI gap should narrow as paper converts to cash.

Why this matters: 2020-2022 funds are in years 3-5 — low DPI is structurally expected. The alarm is that TVPI is already declining before the harvest period even begins, and exit markets remain frozen.

VC Fund Performance: TVPI vs DPI by Vintage

TVPI includes unrealized paper value. DPI is actual cash returned to LPs. The gap between them is the illusion — and it's growing with every recent vintage. All data below is venture capital only.

Cambridge Associates US Venture Capital Index, Carta LP Benchmarks Q4 2025, Preqin

VC Only

Median TVPI vs DPI

The orange line shows how much of TVPI is unrealized paper

Reading this chart: Green bars = actual money returned to LPs. Red bars = total claimed value (including paper). The orange line = the gap between them. For 2021 vintage, LPs got $0.06 back for every $1 invested — the rest is a GP's spreadsheet estimate.

% of Total Value That Is Unrealized

Color = risk level. 2021-2022 vintages: nearly everything is paper.

J-curve context: 97% unrealized for 2022 (age 3) is structurally normal — these funds are still deploying. The concern is 2017 at 57.5% unrealized after 8 years — by now, most value should be realized cash.

VC Vintage Risk Assessment

Risk ratings account for fund age (J-curve position), not just raw DPI. A 3-year-old fund with low DPI is normal. A 7-year-old fund with low DPI is alarming.

VintageAgeJ-Curve PhaseTVPIDPIGap% Unrealized% DistributingIRRRisk
201510 yrsMature — past harvest period2.60x1.85x0.75x28.8%88%18%LOW
20169 yrsMature — harvest winding down2.30x1.60x0.70x30.4%84%16.5%LOW
20178 yrsShould be distributing heavily2.00x0.85x1.15x57.5%79%14%MEDIUM
20187 yrsEntering harvest — DPI should accelerate1.55x0.55x1.00x64.5%68%10.5%HIGH
20196 yrsLate J-curve — exits should be starting1.65x0.45x1.20x72.7%62%12%HIGH
20205 yrsMid J-curve — low DPI expected1.35x0.18x1.17x86.7%42%6.5%VERY HIGH
20214 yrsEarly J-curve — but TVPI already declining1.10x0.06x1.04x94.5%25%0.5%CRITICAL
20223 yrsEarly J-curve — near-zero DPI normal1.05x0.03x1.02x97.1%18%0.1%CRITICAL

Show Me the Money: Realized vs Unrealized

For every dollar of TVPI claimed, how much is real cash (DPI) vs still sitting on a GP's spreadsheet? The J-curve means young funds should be mostly paper — but 2017 at 8 years old is still majority unrealized.

Cambridge Associates US VC Index, Carta Q4 2025

VC Only

Stacked: Cash vs Paper by Vintage

What's normal vs what's not: 2022 being 97% red is fine at age 3. 2015 being 29% red at age 10 is fine — mostly harvested. But 2017 at 57% red after 8 years means exits stalled. And 2018-2019 at 65-73% red in years 6-7 suggests the harvest window is closing without adequate distributions.

% of VC Funds Distributing Any Cash

Blue line = fund age. The older the fund, the more it should be distributing.

2021 vintage: Only 25% of funds have returned any cash at all after 4 years. For comparison, 2015 vintage had 88% distributing by year 10. The concern: exit markets (IPO + M&A) remain largely frozen for VC-backed companies.

The ZIRP Bubble: VC Capital Deployed 2014–2024

$329.6B deployed into venture capital in 2021 alone — nearly double any prior year. This is VC-only deployment data (excludes PE buyout, growth equity, and hedge funds).

PitchBook-NVCA Venture Monitor Q4 2024, NVCA Yearbook 2024

VC Only

Annual US Venture Capital Deployed ($B)

Red bars = ZIRP era (Fed Funds rate near 0%). Orange line = deal count.

ZIRP context: The Fed cut rates to 0-0.25% in March 2020 and held until March 2022. With risk-free returns at zero, institutional capital flooded into VC seeking yield. The 2021 spike — $329.6B across 17,054 deals — is unprecedented and may never repeat.

VC Capital by Era

ZIRP Era Total$736.9B
2021 Alone$329.6B (44.7%)
2021 vs 2014+280%

IRR: The Return Mirage

Median IRR for 2021 VC vintage: 0.5%. Top quartile saved face, but the median fund underperforms a savings account. Note: IRR is particularly unreliable for young VC funds because it's heavily influenced by unrealized paper markups.

Cambridge Associates US VC Index, Carta Q3-Q4 2025

VC Only

VC IRR by Vintage vs S&P 500

Why compare to S&P 500? LPs pay VC managers 2% management fee + 20% carry. If the median VC fund can't beat a passive index fund, the illiquidity premium isn't justified. The 2021 vintage at 0.5% median IRR is currently trailing even Treasury bills.

Power Law: VC Dispersion Across Quartiles

VC is a power-law game. Top-decile funds are 3-6x better than median. But most LPs hold the median, not the best.

The power law trap: VC industry returns are dominated by a handful of exceptional funds. Top-decile 2019 vintage shows 4.5x TVPI while median shows 1.65x. When GPs pitch "VC returns 3x," they're citing top-quartile — most LPs see the median.

The Writedown Wall: $820B in Paper Erosion

Between 2018–2022, peak VC paper values hit $2.18 trillion. Current values: $1.36 trillion. That's $820B evaporated — and $1.16 trillion still unrealized. These are VC-only estimates based on Cambridge Associates median TVPI benchmarks applied to PitchBook deployment data.

Methodology: PitchBook VC deployment × Cambridge Associates median TVPI benchmarks. Approximation, not actual fund losses.

VC Only

The Money Stack: Deployed → Returned → Still on Paper

Green = cash returned. Red = still unrealized. Orange = value that already evaporated.

VC TVPI Compression: Peak → Current

2021 saw the biggest drop: from 2.4x to 1.1x. The J-curve says TVPI should be rising at age 4, not falling.

This is the real alarm: In a normal J-curve, TVPI at year 4 should still be climbing as portfolio companies grow. But 2021 vintage TVPI is declining — from 2.4x peak to 1.1x — which means the underlying portfolio is being marked down, not up. This breaks the standard VC model.

Writedown Wall Detail (VC Only)

VintageVC DeployedPeak TVPICurrent TVPIPeak ValueCurrent ValueErosionCash ReturnedStill Unrealized
2018$153.1B1.8x1.55x$276B$237B-$38.3B$53.6B$183.7B
2019$173.7B2.2x1.65x$382B$287B-$95.5B$55.6B$231B
2020$166.6B2.1x1.35x$350B$225B-$125B$41.6B$183.3B
2021$329.6B2.4x1.1x$791B$363B-$428.4B$39.6B$323B
2022$240.7B1.6x1.05x$385B$253B-$132.4B$9.6B$243.1B
TOTAL$1,064B$2,184B$1,364B-$820B$200B$1,164B

Unicorn Reality Check: Peak vs Now

25 major VC-backed unicorns tracked. 14 lost more than half their peak value. 6 are AI-era exceptions. These are the portfolio companies inside the VC funds above — their markdowns flow directly into the TVPI compression.

Crunchbase, CB Insights, PitchBook, SEC S-1 filings, Fidelity/T. Rowe Price mutual fund disclosures

The ZIRP Casualties (sorted by % decline from peak private valuation)

How to read this: Peak valuations are the highest private-round price. Current values are latest public market cap (if IPO'd), most recent funding round, or secondary market estimates. WeWork ($47B → $0) and Hopin ($7.75B → $15M) represent total loss scenarios that directly hit LP returns.

The Exceptions: AI & Space Outliers

The AI exception: These companies are almost entirely AI or space — sectors with generational tailwinds. OpenAI and Anthropic alone account for most of the positive value delta. They will rescue the specific funds that invested in them, but most VC portfolios don't hold these names.

Full Unicorn Tracker (with sources)

CompanySectorPeak ($B)Peak YrCurrent ($B)ChangeStatusKey InvestorsSource
WeWorkPropTech$47B2019$0B-100%Bankrupt Nov 2023SoftBank ($18.5B invested)WSJ, bankruptcy filing
HopinEvents$7.75B2021$0.015B-99.8%Sold $15M to RingCentralAtomico, IVP, General AtlanticAxios Aug 2023
BoltCommerce$11B2021$0.5B-95.5%Restructured 2023Activant CapitalWSJ, SEC
Better.comMortgage$7.7B2021$0.75B-90.3%SPAC IPO 2023SoftBank, GoldmanBloomberg, SEC
PelotonHealth$8B2020$1.4B-82.5%Public (PTON)Tiger Global, True VenturesNasdaq
InstacartDelivery$39B2021$8.2B-79%IPO Sep 2023a16z, D1 CapitalSEC S-1
GoPuffDelivery$15B2021$3.5B-76.7%RestructuredSoftBank, Accel, D1PitchBook
LyftMobility$15.1B2018$4.5B-70.2%Public (LYFT)Fidelity, a16z, RakutenNasdaq
KlarnaFinTech$45.6B2021$14.6B-68%IPO filed 2025SoftBank, Sequoia, PermiraSEC S-1 2025
SumUpFinTech$22B2022$8B-63.6%ActiveGoldman Sachs, BlackRockCB Insights
Checkout.comPayments$40B2022$15B-62.5%Active (est.)Tiger Global, Insight, DSTBloomberg
ChimeFinTech$25B2021$11.6B-53.6%IPO Jun 2025Sequoia, SoftBank, TigerCNBC, PitchBook
BrexFinTech$12.3B2022$7B-43.1%ActivePeter Thiel, YC, DSTPitchBook
RedditSocial$10B2021$6B-40%IPO Mar 2024Fidelity, Sequoia, TencentBloomberg, SEC
CanvaDesign$40B2021$26B-35%ActiveT. Rowe Price, General AtlanticBloomberg
StripePayments$95B2021$65B-31.6%Active (2024 raise)a16z, Tiger GlobalCrunchbase, Fidelity
ByteDanceSocial$250B2021$220B-12%ActiveGeneral Atlantic, SoftBankVisual Capitalist
AnthropicAI$4.1B2023$61.5B+1400%Active (2024 raise)Google, Spark, SalesforceBloomberg 2024
OpenAIAI$29B2023$300B+934.5%Active (2025 raise)Microsoft, Thrive, TigerReuters 2025
SpaceXSpace$137B2023$350B+155.5%ActiveGoogle, Fidelity, a16zReuters 2025
DatabricksData/AI$43B2023$100B+132.6%Active ($10B raise)a16z, NEA, BatteryBloomberg 2024
RevolutFinTech$33B2021$45B+36.4%Active (2024 raise)SoftBank, Tiger GlobalBloomberg 2024
RobinhoodTrading$11.7B2021$15B+28.2%Public (HOOD)Ribbit, DST, NEANasdaq

Sector Analysis: Where the VC Damage Landed

FinTech got crushed — the sector most inflated by ZIRP (low rates = cheap capital for lending startups). AI defied gravity. These are VC-backed company valuations, not public market cap.

Crunchbase, PitchBook, SEC filings, Fidelity/T. Rowe Price mutual fund markdowns

Peak vs Current Valuation by Sector ($B)

Sector Value Change (%)

Sector context: FinTech was the most rate-sensitive sector — BNPL, neobanks, and lending startups were valued on growth enabled by cheap capital. When rates rose, their unit economics collapsed. AI is the inverse — a secular technology shift that attracts capital regardless of rates.

VC vs PE: CalSTRS Portfolio Split

CalSTRS manages 377 funds across both PE buyout and VC/growth. Here's how they compare — PE buyout has historically higher DPI (buyouts generate cash sooner via leverage recaps), while VC has wider dispersion and longer J-curves.

CalSTRS Private Equity Portfolio Performance Report, FYE June 30, 2025. DPI/RVPI/TVPI calculated from reported cash flows and NAV. Report states values are not GP-approved.

VC OnlyPE / Buyout

DPI Comparison: VC/Growth vs PE Buyout

PE buyout (orange) consistently returns more cash than VC (blue), especially in recent vintages

Why PE returns cash faster: Buyout funds acquire controlling stakes, use leverage, and often do dividend recaps within 2-3 years. VC funds take minority stakes in startups that may not exit for 7-10+ years. Different structures, different DPI timelines. Both 2018-2025 vintages show compressed DPI, but PE is ahead.

TVPI Comparison: VC/Growth vs PE Buyout

VC claims higher TVPI in recent vintages — but with much lower DPI, it's mostly paper

The TVPI illusion in action: 2021-2025 VC shows 2.0x TVPI but only 0.17x DPI — meaning 91.5% is paper. PE same vintage shows 1.56x TVPI and 0.49x DPI — still mostly paper, but 3x more cash has come back. This is why LPs increasingly prefer PE buyout.

CalSTRS Portfolio Summary

VC/Growth Funds

48

$3.71B contributed | 0.93x DPI

PE Buyout Funds

329

$50.8B contributed | 0.97x DPI

2018-25 VC Unrealized

82%

19 funds, mostly paper value

Pre-2005 VC DPI

2.0x

Mature funds fully harvested

CalPERS PE Funds: What 'Normal' Looks Like

These are PE buyout funds, not venture capital. They're included as a comparison benchmark — mature PE funds show what healthy DPI looks like: multiples close to or exceeding TVPI, meaning paper value converted to real cash.

CalPERS PEP Fund Performance Review, Jun & Sep 2025

PE / Buyout

CalPERS PE Funds: IRR vs Investment Multiple

Scatter shows PE buyout performance — these are not VC funds. Included for LP comparison.

Pre-2010 2010-2014 2015+

CalPERS PE Fund Detail

Note: These are all PE/buyout funds. For mature funds, DPI ≈ Investment Multiple means paper value was fully realized.

FundTypeVintageMultipleIRRDPIStatus
Apollo IVPE / Buyout19981.7x8.5%1.65xRealized
Blackstone IPE / Buyout20001.2x6.5%1.22xRealized
CalPERS Corp PartnersPE / Buyout20011.6x9.3%1.56xRealized
CVC Europe IIIPE / Buyout20012.6x41%2.54xActive
Carlyle Europe IIPE / Buyout20031.6x19.7%1.6xActive
TPG Partners IVPE / Buyout20031.9x15.3%1.9xActive
Permira Europe IIIPE / Buyout20041.7x26.6%1.73xActive
Advent GPE V-DPE / Buyout20052.4x42.7%2.42xActive
Silver Lake IVPE / Buyout20142.8x20.9%N/AActive
Blackstone BCP VIIPE / Buyout20171.8x15.3%N/AActive
Key PE vs VC difference: Notice that for realized PE funds, DPI is very close to the investment multiple — meaning almost all paper value converted to cash. This is what healthy fund performance looks like. VC funds rarely achieve this parity because of the power-law return distribution.

Vintage DNA: Multi-Dimensional VC Comparison

How do key VC vintages compare across TVPI, DPI, IRR, and distribution rate? The 2021 vintage collapses on every axis — even accounting for its young age.

Cambridge Associates US VC Index, Carta Q4 2025

VC Only

VC Vintage Radar: 2015 vs 2017 vs 2019 vs 2021

All metrics normalized. 2021 (red) collapses on every dimension.

Age-adjusted view: 2021 is only 4 years old, so some metrics will naturally be low. But compare to where 2019 was at age 4: 2019's TVPI was still climbing, while 2021's is already declining. That's the J-curve breaking.

Top Quartile vs Median DPI: The Power Law Gap

The best VC funds mask the median reality. Top-quartile 2019 returned 1.2x DPI while median returned 0.45x.

Selection risk: LPs who selected top-quartile 2015 managers got 3.1x DPI — life-changing returns. LPs who selected median 2021 managers got 0.06x DPI — a 94% unrealized portfolio. Access to top managers is the entire VC game, and most LPs don't have it.

Industry Benchmarks: The Numbers That Matter

Key VC-specific metrics from the most trusted industry data providers. Each number is cited to its source.

25yr VC Pooled Net Return12-15%Cambridge Associates US VC Index — blends all vintagesSource: Cambridge Associates
5yr VC Return (to 2023)8-12%Compressed by 2022-23 markdowns. Below S&P 500 5yr return.Source: Cambridge Associates
VC Calendar Year 20246.2%PE returned 8.1%. Both below S&P 500.Source: Cambridge Associates
Total VC AUM (end 2023)$1.21TMostly unrealized paper value in portfolios.Source: NVCA Yearbook 2024
VC Dry Powder (end 2023)$311.6BRecord high. GPs waiting for better entry points.Source: NVCA Yearbook 2024
Distribution Rate 2022-24<5%vs. 16.8% decade average. LP cash famine.Source: Preqin/Cambridge
LPs Weighting DPI over TVPI60%Post-2021 disillusionment with paper marks.Source: 2024 LP Survey
Secondary Market Vol (2025)>$210BLPs manufacturing liquidity via secondary sales.Source: EY/Greenhill
VC IPO Price/Sales (2024)~5xvs. 12.3x median in 2020-21. Decade low.Source: PitchBook
2021 Vintage Median IRR0.5%Up from negative. Essentially flat after 4 years.Source: Carta Q3 2025

The Reckoning: 2026–2028 Timeline

Most VC funds have a 10-year life with two 1-year extensions. The ZIRP-era funds are entering the harvest window — years 5-8 when LPs expect DPI to accelerate. If exits don't materialize, extension requests and GP/LP conflicts will spike.

2026

2019 funds turn 7

Entering peak harvest. Extension requests for struggling funds begin. LPs scrutinize RVPI vs DPI on every quarterly report.

2027

2020 funds turn 7

ZIRP-deployed capital faces the exit test. Currently 42% distributing — needs to reach 70%+ to match historical norms.

2028

2021 funds turn 7

The $329.6B vintage. 0.5% median IRR. 25% distributing. Most consequential year — this vintage determines if ZIRP marks were real.

2029

GP fundraising wall

Fund III/IV raises require strong DPI track records. LPs with bad DPI block re-ups. VC AUM likely contracts for first time since 2009.

2030+

New equilibrium

Survivors: AI winners, disciplined managers with real DPI. Casualties: tourist crossover funds (Tiger, Coatue), spray-and-pray strategies.

Why years 7-10 matter: VC funds typically have a 10-year life. The investment period (deploying capital) is years 1-4. The harvest period (exits and distributions) is years 5-10. Funds that can't distribute by year 10 must either extend (requiring LP consent) or sell remaining positions at fire-sale prices on secondary markets. The ZIRP-era funds are now entering this critical window.

Key Insights: What the Data Says

Six takeaways from analyzing VC-specific fund performance, 25 unicorn valuations, $1 trillion+ in deployed venture capital, and the widest TVPI-DPI gap in VC history.

$820B Eroded

Peak VC paper values hit $2.18T for 2018-22 vintages. Current: $1.36T. That's $820B in TVPI compression — and the markdown cycle isn't over. (Source: Cambridge Associates TVPI benchmarks × PitchBook deployment)

19¢ on the Dollar

~$200B returned on $1,064B deployed in VC (2018-22). After 3-7 years, LPs have seen less than 20% back in cash. The J-curve explains part of this for young funds — but 2018 at 7 years old should be much higher.

2021 = Worst VC Vintage

$329.6B deployed at peak ZIRP. 0.5% median IRR. 94.5% unrealized. Only 25% distributing. At age 4, some paper-heaviness is expected — but TVPI is already declining, which is abnormal. (Source: Carta Q3 2025)

Distribution Drought

VC distribution rate: <5% of NAV in 2022-24, vs. 16.8% decade average. LPs are in a cash famine — 8 consecutive quarters of near-zero distributions. IPO window remains largely closed. (Source: Preqin, Cambridge Associates)

AI Saves the Lucky Few

OpenAI (+935%), Anthropic (+1400%), SpaceX (+156%), Databricks (+133%). These outliers will rescue the specific VC funds that hold them — but most portfolios don't. Power law in action. (Sources: Reuters, Bloomberg, Crunchbase)

LP Behavior Shift

60% of LPs now weight DPI over TVPI in manager evaluation (2024 LP survey). Secondary market volume hit $210B+ (EY/Greenhill). LPs are manufacturing liquidity because GPs aren't providing it.

Methodology, Sources & Caveats

This analysis combines public institutional LP disclosures with industry benchmark data. All VC and PE data is clearly labeled. Numbers are estimates based on the best available public data.

Data Sources

  • Cambridge Associates US VC Index — Pooled net returns, vintage year TVPI/DPI/IRR benchmarks. The standard VC performance benchmark.
  • Carta LP Benchmarks Q3-Q4 2025 — Fund-level quartile data, DPI progress, vintage-year distributions.
  • CalSTRS FYE 2025 — 377 fund records (PE + VC mixed). DPI/RVPI/TVPI calculated from raw cash flows. Report states values are not GP-approved.
  • CalPERS PEP Review 2025 — PE/buyout fund performance. Used as comparison benchmark, not VC data.
  • PitchBook-NVCA Venture Monitor Q4 2024 — VC-only deployment totals and deal counts, 2014-2024.
  • Preqin — Fund performance quartiles, AUM data, distribution rates.
  • SEC Filings, Crunchbase, CB Insights — Individual company valuations from S-1 filings and funding round data.
  • Fidelity/T. Rowe Price disclosures — Mutual fund markdowns of private holdings (required quarterly disclosures).

Key Definitions

  • DPI (Distributed to Paid-In) — Cash actually returned to LPs ÷ capital contributed. The only metric that pays bills. A 1.0x DPI means LPs got their money back.
  • TVPI (Total Value to Paid-In) — DPI + RVPI. Includes unrealized paper value. The number GPs put in fundraising decks. Can be gamed via generous markups.
  • RVPI (Remaining Value to Paid-In) — Unrealized NAV still on GP's books ÷ capital contributed. Subjective, based on GP estimates, rarely audited.
  • IRR (Internal Rate of Return) — Time-weighted return. Sensitive to timing and unrealized gains. Can look great on paper markups that never become real exits.
  • J-Curve — The typical return pattern of a VC fund: negative in early years (fees, capital calls), rising in later years (exits). A 10-year VC fund typically returns most cash in years 5-8.
  • ZIRP — Zero Interest Rate Policy. Fed funds rate 0-0.25% from March 2020 to March 2022. Created historically cheap capital that inflated VC valuations.

Important Caveats

  • • Writedown Wall estimates are approximations (deployed × TVPI change), not actual fund-level losses.
  • • CalSTRS data includes some growth equity funds tagged as VC — categories are based on keyword matching.
  • • Unicorn "current values" for private companies are secondary market estimates, not audited valuations.
  • • 2020-2022 vintages are still in early J-curve — final outcomes won't be known until 2028-2032.