Venture Liquidity, Endowments and AI Investing in 2025

Harry Stebbings joins investors Rory O’Driscoll and Jason Lemkin to discuss a reported Windsurf deal, venture-fund liquidity, university endowments and AI business models.

The guests discuss a reported approximately $3 billion OpenAI acquisition of Windsurf. O’Driscoll says the transaction had not happened to his knowledge and was still a rumor; the panel evaluates a hypothetical deal, not a completed acquisition.

O’Driscoll thinks a coding product could be strategically useful to a model provider with a large valuation. Lemkin questions whether the transaction will actually happen; neither knows the final terms or outcome.

Lemkin compares potential acquisition prices with an acquirer's overall value and argues coding is an important competitive use case. His assessment of which model provider leads is an opinion from April 2025.

O’Driscoll describes why a fast-changing market may justify making strategic bets despite uncertainty. He contrasts successful historical acquisitions with deals that failed, emphasizing that the winner is not obvious in advance.

Lemkin says an acquirer could continue supporting other applications alongside its own product. O’Driscoll agrees this is possible but sees a longer-term risk for independent applications if a platform favors its own tools.

Stebbings argues that multistage venture firms are increasingly competing with dedicated seed funds. He cites a reported early investment in Windsurf, but any fund-return calculation depends on an uncompleted deal and unknown ownership.

O’Driscoll praises the investor's selection while cautioning against inferring a systemic result from one company. Lemkin says smaller ownership stakes and larger seed funds could make a single exit less able to return a fund.

The guests compare difficulties across seed, Series A and later stages. O’Driscoll says abundant capital and funds crossing stages can raise competition throughout the market, not only at seed.

Stebbings describes seed rounds in which founders seek less dilution. O’Driscoll says strong demand for a deal can improve the founder's bargaining position; these are their experiences rather than a universal market statistic.

The panel considers a company valued at a reported high revenue multiple. Lemkin asks about forward revenue; O’Driscoll says a multiple is incomplete without the growth rate, period and assumptions behind it.

O’Driscoll explains that early companies often have high current revenue multiples and must grow into their valuations. Continued growth, future financing and the eventual exit price remain uncertain.

The investors debate multistage firms' ability to back companies over several rounds. Large funds may offer continuity, while early investors face the challenge of retaining meaningful ownership.

The panel discusses whether seed investors should specialize or expand into later stages. Their answers reflect their own fund sizes, access and incentives rather than a single optimal model.

Stebbings asks how large firms compete for talent and deals. The guests describe partner autonomy, brand and capital as potential advantages but note that individuals also assess their own career opportunities.

O’Driscoll discusses the incentives for an established partner to start a new firm. A strong prior track record can help fundraising, but starting independently also introduces execution and fundraising risks.

They consider whether established platforms or emerging managers will attract limited-partner commitments. O’Driscoll says the more fundamental test is performance, while Stebbings reports that some allocators are waiting.

The panel differentiates a broad boom in new funds from a smaller number of experienced investors spinning out. It remains uncertain whether institutional appetite will support as many new managers in the next cycle.

Stebbings raises pressure on university endowments, including uncertainty about government funding and tax treatment. The guests discuss possible consequences as conditions and forecasts from April 2025, not confirmed losses for every institution.

O’Driscoll says an endowment with uncertain cash needs may place a higher value on liquid assets. His example of potential revenue disruption is a scenario for cash planning, not an established reduction at a named university.

Stebbings asks how limited partners balance liquidity with relationships at established venture firms. O’Driscoll says allocators may prioritize existing commitments over new managers, but decisions vary by institution.

The panel mentions a reported possible Yale secondary sale of private assets. O’Driscoll explicitly says the report may be untrue and offers several possible explanations, including managing cash needs rather than abandoning the endowment model.

Lemkin says some limited partners told him their cash-distribution plans were too optimistic even before the latest policy uncertainty. He believes delayed private-equity distributions may be a larger pressure than venture alone; this is based on limited conversations.

O’Driscoll explains the denominator effect: when liquid public assets fall while private marks adjust more slowly, private assets can become a larger share of the portfolio than the target allocation.

He distinguishes the denominator effect from a more immediate cash shortfall and from uncertainty about long-term returns. A delayed exit can reduce annualized IRR even when an investment's eventual multiple is unchanged.

The guests discuss universities' ongoing spending obligations and the difficulty of forecasting sudden changes to revenue. Their statements are scenarios and opinions, not institution-specific financial statements.

Stebbings says some endowments he spoke with had high allocations to private assets. O’Driscoll notes that long-lived institutions can tolerate illiquidity when they are compensated for it, but still need enough cash for obligations.

The conversation returns to capital-efficient startups raising large rounds. Lemkin says high valuations and investor demand can make founders willing to take capital even when the business may not need it immediately.

O’Driscoll calls it a paradox that investors want capital-efficient companies yet compete to invest large amounts. The panel treats this as an observation about incentives, not a rule for every financing.

Stebbings asks about AI-enabled rollups of services firms. O’Driscoll cites his firm's investment in a predecessor of Nuance as an example of a long-running acquisition strategy, while remaining skeptical that all similar rollups will work.

His concern is that acquired customers did not choose the buyer's new AI product. Some may fit the automation well, while others require different service, making integration and retention harder than a simple revenue multiple suggests.

Lemkin agrees the fit between an acquired customer base and a new product is a useful question. Both guests emphasize that buying revenue does not automatically create a repeatable sales motion.

The discussion shifts to competition. Lemkin says he will consider a crowded market when a founder has strong growth and a clear area to win, even though he would prefer less competition.

O’Driscoll argues that competition can pressure customer-acquisition costs and margins. The investors differ on how much competitive analysis to do before investing and acknowledge the risk of being wrong.

The panel considers whether one outstanding founder and a fast-growing product can overcome a crowded field. They frame the answer as a judgment about a particular deal rather than a promise that growth guarantees a winner.

The guests discuss fund construction and the significance of ownership. A return on one investment must be evaluated against the size of the fund and the outcomes of the rest of the portfolio.

Lemkin argues that an investor with genuine high conviction in a potential fivefold return should find a way to invest. The fivefold outcome is a hypothetical; the conversation provides no way to guarantee it.

O’Driscoll distinguishes an individual's flexibility from building an institutional firm with consistent decision processes. A single exception can be harder to manage across a team even when an opportunity looks compelling.

He says well-informed conviction deserves weight because the future is uncertain, but adds that a firm's job is to execute its stated strategy. The panel does not equate conviction with certainty.

O’Driscoll says he is more concerned about missing a strong company in his firm's core enterprise-software focus than a large deal outside its expertise. He then leaves the conversation.

Stebbings and Lemkin discuss the concentration of technology talent in the San Francisco Bay Area. Lemkin says local proximity helps his own relationship-driven approach; this is his experience rather than a requirement for every founder.

Lemkin explains that he prioritizes exceptional growth and a founder he believes in, and says he has not always conducted extensive direct competitor interviews. He describes a personal method, not a recommendation to skip diligence.

Stebbings challenges whether some past portfolio companies looked like obvious outliers at the time. Lemkin cites his recollection of their growth and acknowledges the narrowness of his investment criteria.

They discuss investment in a competitive market, including Owner, in which Lemkin has invested. He says the company's founder and technical team matter to his thesis, and his expectations for its success are forecasts.

Lemkin contrasts his own limited customer checks with a larger firm's extensive diligence. Stebbings suggests detailed work can also help a growth investor earn a meeting with a founder.

Lemkin says he uses AI to screen incoming company materials and alert him when certain metrics meet his criteria. He describes his process and its intended use, not a validated guarantee of investment performance.

Stebbings thanks Lemkin and closes the conversation.