Startup Funding, OpenAI and AI Competition: A Venture Investor Roundtable

Harry Stebbings and investors Jason Lemkin and Rory O’Driscoll discuss startup financing, venture returns, AI companies and how AI may change business software and work.

Stebbings introduces a new funding round for Owner. Lemkin says he led its seed investment and cites a $120 million round and company growth figures; his account reflects his investment interest and the numbers are not independently verified in the discussion.

Lemkin describes a fast fundraising process with multiple term sheets. He finds the speed uncomfortable but says companies with strong metrics can attract substantial capital even when other startups struggle.

O’Driscoll says an investor asked to decide in two days is often competing with someone who already knows the founder. He recommends building relationships and doing work ahead of a financing rather than assuming two meetings are sufficient.

Stebbings says his team is considering preempting rounds soon after seed funding. Lemkin sees a case for acting when an investor already knows the company and new evidence supports the price; O’Driscoll is uneasy about paying much more after only a short interval.

O’Driscoll notes that an unsolicited round may solve the investor's problem more than the company's need. He also says rapidly growing companies can make productive use of more capital, so the answer depends on circumstances.

Stebbings cites a reported fall in Series A activity. The guests distinguish early belief in a team from later evidence of traction, although their views about what deserves funding are judgments rather than a complete explanation of the data.

Lemkin says Owner had periods when raising was difficult. He argues founders should focus on building a compelling company, while O’Driscoll says statistics do not determine an individual startup's outcome.

O’Driscoll shares his experience of stage-four colon cancer and treatment. He says it clarified that he enjoyed his work and wanted to continue; this is a personal account, not medical guidance.

The panel discusses whether investments in OpenAI and Scale could improve a Tiger Global fund's results. O’Driscoll says the answer depends on position sizes, entry prices and future outcomes, which he does not know.

He illustrates how even a large multiple on a small position might not return a very large fund. A concentrated position could have more impact, but the panel does not know Tiger's actual allocation.

The guests consider the importance of vintage and timing: an investment made before a strong exit window can have a different outcome from one made at peak prices. Their comparisons are illustrative, not an audit of any fund.

Stebbings discusses FTX-related investments in Anthropic and Cursor and suggests they might have improved financial outcomes under a different scenario. That counterfactual is speculative and does not establish what creditors would have recovered.

The guests acknowledge Sam Bankman-Fried's fraud conviction while discussing reported investments. Lemkin's remark about committing only a little fraud is explicitly identified by O’Driscoll as irony, not advice or an endorsement.

O’Driscoll calls the early AI investments striking, while separating that assessment from the fraud case. The discussion does not independently verify investment amounts, proceeds or a counterfactual recovery.

Asked whether Bankman-Fried might receive funding in the future, Lemkin speculates someone could back him; O’Driscoll thinks a criminal conviction would substantially raise the barrier. A pardon and any future financing are hypothetical.

Stebbings raises a reported Coatue vehicle aimed at a wider group of investors with a stated lower minimum and some liquidity features. The panel considers the structure as described, without verifying final terms or eligibility.

Lemkin is skeptical about marketing illiquid venture exposure to individuals. O’Driscoll says other private-equity managers seek similar sources of capital, while questioning why companies increasingly stay private.

O’Driscoll thinks investors may be attracted by access to private technology companies. Lemkin says fees and carry can be difficult to understand and may look different when compared with public funds rather than traditional venture funds.

The panel distinguishes three risks: manager selection, the underlying asset class and the fund structure. O’Driscoll emphasizes uncertain venture performance and redemption limits; he does not guarantee any manager's results.

Lemkin says he generally discourages individuals from committing money they may need to illiquid venture funds. This is his personal view, not individualized investment advice.

Stebbings asks about reports that Klarna planned to hire people after an AI support push. Lemkin argues that earlier accounts of the company's approach were oversimplified, and the guests debate which support tasks can be automated.

They discuss the distinction between handling simple cases automatically and providing humans for more complex requests. The panel's interpretations are not an independent measurement of Klarna's staffing or service quality.

The conversation turns to Microsoft's reported workforce reduction. Lemkin argues companies may need to reorganize more aggressively around AI; this is his management opinion, not proof of why particular employees lost jobs.

O’Driscoll cautions that translating technology into organization-wide productivity is difficult. The guests differ over the speed and scale of future job changes, and their estimates are predictions.

They discuss how leaders decide which tasks need people and where tools can help. Lemkin emphasizes urgency, while O’Driscoll focuses on sequencing and the practical work of implementation.

The panel discusses an OpenAI leadership appointment and whether a rapidly changing technical company needs its chief executive to be an engineer. Lemkin says his experience makes him skeptical of nontechnical leadership, while acknowledging OpenAI's success.

O’Driscoll argues that recruiting, empowering technical teams and providing resources are also important executive skills. He does not treat a leader's background alone as decisive evidence of a future outcome.

The guests consider possible consumer and coding applications for OpenAI. Their expectations about future products and market reach are forecasts, not company announcements.

Stebbings suggests a much higher future OpenAI valuation and asks about concentrating a fund. O’Driscoll sees a large opportunity but says his limited partners pay him to pursue a different strategy; the valuation is speculative.

O’Driscoll says longer private holding periods may require more portfolio companies to end up with enough winners. The arithmetic is his portfolio-construction hypothesis, not a proven rule.

He warns that a good company can still be bought at too high a price or too early. Forecasting several years of rapid growth from leaked or reported figures creates substantial uncertainty.

The panel discusses OpenAI's nonprofit and public-benefit structure and the complicated relationship with Microsoft. O’Driscoll describes his understanding of a proposed structure, not a completed transaction or definitive legal interpretation.

He says changing a complex agreement can require negotiations among stakeholders and may face legal questions. The speakers' estimates of costs and ownership are not verified contract terms.

Stebbings cites a rumored $500 million Perplexity round at a $14 billion valuation. Lemkin estimates possible revenue growth and questions the price; both the financing and his revenue estimate are unconfirmed in the conversation.

O’Driscoll frames the investment as a high-risk bet on a potentially enormous search market, not an equal probability that any one competitor will win. His discussion of a trillion-dollar outcome is a hypothetical investor thesis.

The guests consider both the possibility of competing with larger providers and a potential acquisition. They acknowledge that regulation, execution and pricing could change the outcome.

Stebbings discloses that he is an investor in Perplexity and is asked whether he would invest more at the rumored price. He says that round size and valuation are outside his fund's usual strategy.

The panel turns to a reported Clay transaction. Stebbings raises the possibility that Clay could challenge Salesforce; Lemkin says he sees strong demand among marketers but also many new competitors.

Lemkin recounts conversations with marketers interested in AI and predicts competition for Clay will be intense. His anecdotes and proposed financing strategy do not establish customer motives or the company's actual plans.

O’Driscoll describes Clay's original data-combination workflow and its subsequent AI features. He says his firm has investments in adjacent sales and marketing automation, a relevant commercial interest in the discussion.

The guests debate whether customers buying out of fear of missing AI will remain if products fail to deliver. O’Driscoll argues that a company must eventually provide the value it promised.

Lemkin shares his observations about marketing teams and their interest in AI tools. His proposed pace of staffing changes is an opinion; the panel does not establish company-wide employment outcomes.

The speakers discuss AI video editing and the increased supply of content. Stebbings emphasizes the challenge of discovery, while Lemkin discloses an investment in Opus and describes a potential feature.

O’Driscoll expects people who learn to use AI tools to remain valuable, while Lemkin predicts faster efficiency gains. They close with different views on the pace and extent of workforce change.

Stebbings thanks the guests and closes the conversation.