Venture Capital's AI Boom Has a Liquidity Test By Dor Arad | October 4, 2026 An investor can own a share of the future and still struggle to get capital back. That is the tension I see in today's venture capital market. AI has created compelling opportunities, enormous financing rounds and ambitious expectations. Yet the strength of a technology trend does not automatically resolve the economics of the funds investing in it. The distinction matters to founders, fund managers and the institutions financing them. A company raising money, a fund reporting a higher valuation and an LP receiving a distribution are three different events. A credible recovery needs progress across all three. THE HEADLINE AND THE UNDERLYING MARKET NVCA's public first-half 2026 overview reports more than $400 billion raised by U.S. startups. It also describes investment and fundraising concentrated among relatively few companies and managers. These are first-half figures, not final Q3 results. [2] PitchBook analyst Emily M. Zheng's public summary gives another perspective: AI accounts for nearly 80% of U.S. venture deal value over the trailing twelve months, versus 43.2% of deal count. [3] Those figures can support enthusiasm about AI while raising questions about the rest of the market. Does the typical manager own the assets driving the headline? How much exposure overlaps across funds? Can founders outside the largest rounds obtain capital on workable terms? The aggregate alone cannot answer those questions. PAPER VALUE NEEDS A PATH TO REALISATION A new funding round can extend a startup's runway and provide evidence for a higher valuation. If it is a primary financing, however, the company receives the proceeds. Existing investors do not automatically receive cash. Even a genuine exit needs careful interpretation. A fund may retain shares after an IPO. Acquisition proceeds may arrive in stages. A secondary sale may cover only a small part of a holding. For an LP, the useful discussion follows the entire path from company value to proceeds received after fund costs. Two measures help: TVPI includes distributions and remaining portfolio value relative to capital paid in; DPI covers distributions relative to paid-in capital. Distributions may include securities, so cash liquidity requires a further breakdown. [5] A high reported multiple can be informative. It still needs an explanation of what remains unrealised, how it was valued and when it might become available. TIME IS PART OF THE RETURN Carta's public Q2 2026 summary reports median DPI of 0.37x for 2017-vintage funds, 0.15x for 2018 and 0.04x for 2019. The figures describe Carta's covered funds and should not be generalised to every venture portfolio. [4] Limited distributions do not establish that the remaining assets are worthless. They do make timing a serious investment question. An illustrative threefold return received after five years is about 24.6% annualised; received after ten years, it is about 11.6%. This assumes one initial investment and one final payment, before additional fees or taxes. The same multiple can deliver a very different outcome. Young funds need time to build companies. As funds age, managers should explain changes to expected exit dates and the additional funding required to reach them. “Be patient” becomes more persuasive when it comes with milestones, alternatives and a clear account of the cost of waiting. AI SUCCESS DOES NOT REMOVE PRICE RISK I am wary of treating the market as a choice between “AI changes everything” and “AI is a bubble.” A transformative business can be an unattractive investment at the wrong price. A company can grow rapidly while later financing dilutes an earlier investor. A useful product can face enough competition that its customers capture most of the economic benefit. Concentration adds another dimension. Owning several funds may offer less diversification than expected if they hold the same companies or depend on the same financing conditions. Investment committees should test a scenario in which AI adoption succeeds but investor returns disappoint. That possibility is more revealing than assuming the technology must fail for the portfolio to struggle. ACCOUNTABILITY SHOULD BE VISIBLE IN THE TERMS Low distributions are not proof of misconduct. Nor do these sources justify a sweeping claim that managers face no consequences. But investors are entitled to ask whether the manager shares the burden of delay. If a fund requests more time, what happens to its fees? If assets move to a vehicle controlled by the same manager, how is pricing independently tested? If cash distributions are financed by borrowing, what debt and costs remain? ILPA's NAV-facility guidance recommends LP advisory committee approval for using such facilities to fund distributions. The guidance concerns private equity and does not establish that every VC fund uses this financing. [7] These distinctions matter because identical cash receipts can leave different obligations behind. My preference is for fund terms that make extensions, conflicts and compensation transparent. Where circumstances warrant, investors should negotiate changes to fees, stronger oversight and a clearer exit plan. The appropriate terms depend on the fund agreement and the actual situation. WHAT I WOULD LOOK FOR NEXT A more convincing recovery would show distributions spreading across mature funds, credible transactions supporting a wider range of valuations and more viable businesses obtaining funding beyond the largest names. Fund reports should separate realised results from remaining estimates and explain both after costs. LPs should be able to see how much of the outcome depends on a few holdings, further fundraising or an optimistic exit timetable. Founders should ask a parallel question: will this round help build a business that can eventually support itself? AI may create exceptional value. Venture capital earns its place by financing that value creation and delivering an attractive outcome to the investors who supplied the capital. The question I would put to any investment committee is this: What evidence would convince us that higher valuations are becoming durable investor returns? Read the full analysis: https://doraradinsights.com/en/ai-venture-capital-valuation-liquidity-gap Sources and scope This is original analysis prompted by the public title and summary of TheMarker's October 3 article. The full subscription text was not available for review. Figures below refer to their stated periods and samples; numerical examples are illustrative. [1] TheMarker: Success or illusion: is the AI boom saving venture capital or concealing its problems? — public title and summary https://www.themarker.com/d-academy/2026-10-03/ty-article-magazine/.premium/000001a0-f2b1-dbb2-a9f8-fbfb8f580000 [2] National Venture Capital Association: PitchBook-NVCA Venture Monitor — public Q2 2026 overview https://nvca.org/pitchbook-nvca-venture-monitor/ [3] Emily M. Zheng, PitchBook, via LinkedIn: Public summary accompanying Quantitative Perspectives on US VC https://www.linkedin.com/posts/emzheng_new-report-our-latest-quantitative-perspectives-activity-7507831173193269248-LLDX [4] Carta: VC Fund Performance: Q2 2026 — public summary https://carta.com/data/vc-fund-performance-q2-2026/ [5] Institutional Limited Partners Association: Private Equity Glossary — distributions, DPI and TVPI https://ilpa.org/resources-tools/private-equity-101/private-equity-glossary/ [7] Institutional Limited Partners Association: Guidance on transparency around NAV-based facilities https://ilpa.org/news/new-ilpa-guidance-encourages-lp-gp-dialogue-transparency-around-nav-based-facilities/ #VentureCapital #ArtificialIntelligence #PrivateMarkets #CapitalAllocation #InvestorRelations