Oren Zeev on AI, Venture Fund Economics and Investing Against Consensus

Investor Oren Zeev joins Harry Stebbings to discuss contrarian decisions, AI disruption, venture-fund economics and his own portfolio. Zeev has financial interests in companies and funds he describes; his performance claims, private valuations and forecasts are his account, not independently verified results or guarantees.

Zeev says an investment that looks strange can face less immediate competition, giving founders time to build an advantage. Contrarianism alone is not enough: the investor must also be right. He tries to avoid markets crowded with ten or twenty competitors because each company has a smaller chance to become the leader.

He sees AI as a major source of both value creation and destruction but says his investing fundamentals have not changed. A new question is whether each company is likely to benefit from AI; a business that is merely neutral may also be less attractive in his framework. That is an investment screen, not a prediction that every neutral business will fail.

Asked about Navan, a major holding, Zeev says he is convinced it will benefit from AI and rejects the possibility of disruption. His confidence reflects his substantial investment and is a forecast, not an established outcome. He believes public investors have not yet distinguished sufficiently between incumbents that benefit and those that lose.

Zeev says Navan’s gross margin was around 50% three years earlier and claims support automation has improved it, without giving a current figure. He expects AI to handle much more support and improve customer experience. The historical figure and benefits are his account, and the future share of automated support is a prediction.

More generally, he argues that distribution, operational complexity, integrations, data and, in some industries, regulation can protect incumbents even when software becomes easier to build. He also says companies that fail to adapt may struggle. These factors can be advantages, not guarantees that any specific incumbent is safe.

Stebbings asks whether crowded AI customer-support markets contradict Zeev’s preference for unconventional opportunities. Zeev says one or two new companies may succeed but he does not trust himself to pick them from a very large field at the earliest stage. His numerical illustration is rhetorical rather than a verified count of competitors.

Stebbings asks whether extraordinary growth expectations make formerly impressive companies harder to finance. Zeev replies that doubling each year for five years still compounds to 32 times the starting level; the central questions are sustainability and unit economics. Compounding arithmetic is not a forecast that any company will achieve it.

Zeev describes a portfolio company he says is growing from $20 million to $40 million in annual recurring revenue with healthy economics, and predicts another doubling. Stebbings argues investors have competing opportunities in faster-growing businesses. The company’s numbers are the investor’s account; ARR is a run-rate metric, not necessarily recognized revenue, and the next year’s growth is a forecast.

Zeev warns that rewarding growth alone can encourage unsustainable transactions, illustrating the problem with two companies buying each other’s products. His example is hypothetical and not an allegation against a named business. He says revenue should be assessed alongside economics and genuine demand.

He acknowledges that in some winner-take-most contests, such as his historical example of ride-hailing competition, companies may need to pursue share before margins. When businesses have a choice, he prefers sustainable growth, but says the right timing depends on the market. Growth and margin tradeoffs cannot be reduced to a single rule.

Asked whether he trusts the market’s judgment, Zeev says both his largest successes and mistakes involved disagreeing with it. He tries to suppress social proof and invest according to his own conviction, while accepting that individual decisions will be wrong. His discussion of asymmetric venture outcomes is a framework, not a promise of returns.

On follow-on investing, he argues that investors need intellectual honesty and must change their minds when evidence changes instead of defending previous decisions. He contrasts this with the temptation to seek information that confirms an existing belief. His assessment of his own discipline remains self-reported.

Zeev describes an unnamed property-technology investment whose revenue run rate rose from about $2 million to $30 million; a further move to $100 million was a projection, not an achieved result. He invested again after stress testing interest-rate increases, but says the rate rise was faster than his modeled worst case and the company did not survive.

He cautions against judging a decision solely by its outcome: a reasonable probability-based bet can still lose. In hindsight he says his stress test was not severe enough, but does not conclude that every future investment should be excessively conservative. This analogy does not establish the original decision was objectively correct.

Stebbings asks about concentration in Navan and the pressure of the pandemic. Zeev recalls an earlier dispute with an airline that the company resolved and says he trusted CEO Ariel Cohen through COVID. He describes rapid cost and product changes as his account of management’s response, not an independent evaluation of every affected worker or customer.

He says his approximate limit for one company is 20% of a fund and argues that his limited partners can diversify across managers. The approach increases exposure to individual company outcomes, and his assertion that diversification at the fund level offers no LP benefit is his opinion, not a universal portfolio conclusion.

Stebbings relays concerns from LPs that Zeev deploys capital too quickly. Zeev accepts that unpredictable deployment makes planning harder and says some LPs have opted out. He prefers to invest when he sees opportunities, while acknowledging that in retrospect he invested too quickly in 2021.

He estimates he paid three to four times what he should have for many 2021 deals and expects one peak-vintage fund to be acceptable rather than outstanding. He says good companies can still produce winners despite high entry prices. Those are his retrospective estimates and predictions, not realized audited fund returns.

Zeev argues that long-term LPs may diversify across successive funds even when one fund has a concentrated vintage. He says two 2021–22 funds exceeded $500 million and later reduced fund size, describing a roughly $250 million fund and an unfinished next raise. His target and expected outcomes are not commitments.

On venture-firm structure, he largely agrees that very large platforms and differentiated boutique or solo managers have distinct advantages, while an undifferentiated middle may struggle. He describes his own speed and personal relationships as a solo investor’s strengths and acknowledges that large firms can offer things he cannot.

Zeev estimates that at least half of funds may be unable to raise, unsure they can raise, or delaying a test of the market, and thinks many will not succeed. These are different states, not a measured 50% closure rate. His estimate and prediction are not based on a survey presented in the interview.

On private marks, he says valuation methods leave managers discretion and that LPs should consider incentives and the manager’s character as well as methodology. He suggests a firm confident of raising may feel less pressure to inflate marks. This is his general assessment, not proof that any particular manager or auditor misstates valuations.

He observes that LPs have faced limited liquidity and are paying more attention to distributions relative to paid-in capital, or DPI, than to paper value alone. He speculates that large IPOs could improve liquidity in 2026–27, but the companies and timing are not established in the interview. He regards the preference for DPI as potentially cyclical.

Zeev generally avoids selling strong private holdings through secondary transactions because he prefers to retain potential upside. Stebbings argues that earlier liquidity can reduce execution and IPO risk; Zeev agrees a sale can make sense if the expected incremental return does not justify the risk and wait. Neither approach guarantees a better outcome.

Zeev says he is the largest LP in each fund, typically contributing about 13–14%, and also receives 30% carry. These are distinct economic interests and do not imply a fixed share of future proceeds. He says he seeks alignment by optimizing for long-term LP value.

He clarifies that his funds charge a low management fee but that he reinvests it in the fund and takes no personal income from it. He says the arrangement places outside LP repayment before his own distributions. This is his description of his fund terms, not a claim that no fee is charged or that all funds have the same waterfall.

Zeev argues that large management fees can create incentives to prioritize raising the next fund over maximizing long-term performance. He illustrates a $10 billion fund with a 2% annual fee as a simplified example, without accounting for variations in fee basis, expenses or timing. His claims about other managers’ motivations are general views.

He also argues that partners in large firms may face career incentives to favor their own investments and hesitate to acknowledge mistakes. He contrasts this with his solo structure. His claim of stronger alignment is his perspective and does not establish that any firm or investor is free of conflicts.

Stebbings asks about expensive Series A rounds. Zeev says the label matters less than whether progress since the prior round has reduced risk: a product, office and early logos may look impressive without proving durable product-market fit. He advises examining customer commitment and renewals rather than appearances.

On preemptive financing, Zeev says founders may reasonably accept an attractive offer but should spend according to market evidence rather than the cash balance or board pressure. He acknowledges that this is easier said than done and that excess funding can reduce focus. His advice is not a guarantee that a particular financing is beneficial.

Zeev says founders ask him for advice because he does not force it and will support their decisions even when he disagrees. He thinks the way advice is delivered affects whether someone can hear it. He recalls dismissing an LP’s suggestion to reduce fund size, then later deciding the suggestion had been right.

Stebbings asks about ownership targets after describing opportunities to buy small stakes in fast-growing startups. Zeev says he avoids promising rigid investment rules to LPs because he wants flexibility. This is his approach to portfolio construction, not a claim that investors should ignore their actual contractual obligations.

Zeev recounts investing in Decart at the founder stage through a capped SAFE despite normally avoiding such instruments, ending with roughly 5% ownership. He says the company became profitable quickly and he could not increase the position, while later investors included Sequoia and Benchmark. The investment remains ongoing; the outcome and financial claims are his account.

He says he previously invested his own money for eight years but prefers managing a fund because he wants to be a significant backer rather than participate in more deals. He describes roughly 15 companies per fund and overlapping holdings across funds, explaining why the total number of unique companies is smaller than a simple multiplication.

In a personal and political aside, Zeev disputes the assumption that wealthy people are inherently bad and says many people he knows seek positive impact. He criticizes political attitudes toward wealth. These are his observations and political opinions, not a measured account of wealthy people or policy outcomes.

Stebbings expresses both excitement about investing in AI and concern about labor displacement. Zeev shares both feelings and thinks the transition could be unusually powerful; he hopes his fears prove wrong. Neither offers a validated forecast of job losses or a settled view of how benefits will be distributed.

Zeev recalls meeting a founder once, passing, and later investing after a second meeting showed progress and greater confidence. He presents the example as a personal assessment of a portfolio company, not evidence that its future success is assured. He says the process moved quickly despite other term sheets.

On missed investments, Zeev says seeing few famous misses can indicate an investor did not see enough opportunities. He recalls proposing early Facebook at a prior partnership and wanting to take Audible private after a share-price decline, but says he could not obtain partnership approval. These are his recollections, and hypothetical returns cannot be established.

He describes an attempted introduction to Wiz that went to an email address the founder was not checking, but says he likely would not have secured the investment anyway. He warns against assuming that merely seeing a deal meant an investor could have won an allocation. Stebbings notes that smaller checks can sometimes make access easier.

Zeev praises another investor’s authenticity and leadership while distinguishing that approach from his own. He closes optimistic about opportunities to build businesses with AI, but concerned about political unrest and people being left behind. The simultaneous optimism and concern are his judgments, not certain social or financial outcomes.