Navan’s IPO, Harvey’s Valuation and the New Economics of Venture

Harry Stebbings opens with Navan’s IPO and a decline in its shares. Jason Lemkin sees the debut as a reminder that a strong software company can face a difficult public valuation, while Rory O’Driscoll emphasizes the company’s survival through the pandemic and the importance of a completed listing.

Lemkin cites reported revenue and growth as reasons the reception felt disappointing. Those figures and the market values discussed are time-specific statements from the episode, not independently verified financial results or a forecast of the stock’s future performance.

O’Driscoll argues that a decline after pricing illustrates the risk IPO buyers take, challenging the idea that every IPO allocation is free money. One example cannot establish the average return or resolve the wider debate about offering prices.

Stebbings asks when a venture investor’s reported stake value becomes cash. O’Driscoll describes a typical lockup and a potentially gradual sale or distribution; exact restrictions, timing, taxes and proceeds vary by offering and investor.

Lemkin describes a historical rule of thumb for distributing shares over time after a lockup, while noting that conviction and market liquidity affect the decision. Their examples distinguish a temporary paper value at a first-day high from the proceeds eventually realized.

The group asks how mature public-company multiples should affect venture entry prices. O’Driscoll says growth rates matter: a company expanding much faster than a mature business is not a like-for-like comparison, though its multiple may change as growth slows.

Lemkin says high entry prices and dilution make it harder for a small fund to generate large multiples without a very large exit. His example is a fund-level calculation, not a claim that a particular recent investment will achieve or miss a return.

Stebbings argues that investors can miss strong founders by demanding an obvious multibillion-dollar outcome at the first meeting. Lemkin says portfolio construction and the size of an initial check affect how much uncertainty a fund can tolerate.

O’Driscoll notes that a higher revenue threshold for listing, if sustained, changes the bar for venture-backed companies. The threshold is his rule of thumb rather than an exchange listing requirement, and many paths to liquidity remain possible.

The participants debate whether a roughly $4.5 billion exit is sufficient for investors with different fund sizes and ownership stakes. A founder’s personal outcome, a fund’s multiple and the timing of cash distributions can differ substantially.

Stebbings introduces a reported Harvey financing at an $8 billion valuation. The round size, revenue and growth figures are cited in conversation and are not independently verified; their discussion uses assumptions about future markets and cash generation rather than company guidance.

Lemkin likes financings that raise substantial capital for a relatively small percentage of ownership. That reduces immediate dilution for existing holders, but the panel acknowledges that a high valuation can raise the bar for a later return.

O’Driscoll describes legal work as a potential use case for language models and asks whether spending could expand from legal software into work currently performed by people. The panel does not establish Harvey’s market share, legal accuracy or a measured amount of labor replaced.

He illustrates the math with a hypothetical $24 billion company valued at seven times revenue, implying roughly $3 billion in annual revenue. The calculation is a scenario for evaluating market size and growth, not a prediction that Harvey will achieve it.

Stebbings asks whether venture funds are accepting smaller ownership positions. Lemkin says recent deals have given him less ownership than his target, while O’Driscoll says both capital-efficient startups and capital-intensive model companies can produce small investor stakes for different reasons.

The guests discuss how founders can sequence smaller financings to reduce dilution. Lemkin describes accelerator rounds as one example; the terms and ownership of individual transactions are not verified by the discussion.

O’Driscoll says an optimized fundraising plan for a founder may leave a venture investor below its ownership target. The tradeoff depends on valuation, future financing and portfolio construction; neither side’s preferred percentage guarantees success.

Stebbings raises a public exchange between Sam Altman and Brad Gerstner concerning OpenAI’s infrastructure financing. The panel says the funding question is important but does not claim access to the complete financing plan or the participants’ private motives.

O’Driscoll argues that boards should be able to ask how major commitments can be funded and discuss scenarios with a CEO. Lemkin says investors may hesitate to challenge successful founders. Their comments are general observations and speculation, not evidence of wrongdoing or a finding about OpenAI’s board.

O’Driscoll says ambitious forecasts could create reputational and financial risk if they prove too optimistic. The figures discussed are estimates and the suggested crash scenario is hypothetical, not a claim that the company supplied false numbers or cannot meet commitments.

The guests discuss board oversight: directors can support a founder while also asking for realistic cash-flow scenarios and checking assumptions. They do not assert that any named board failed to perform its duties.

The conversation moves to Amazon and cloud demand. O’Driscoll cites reported AWS growth and argues that demand for AI compute was strong at the time. The quarterly figures and capital-spending comparisons are the speakers’ account, not independently verified financial analysis.

Lemkin questions how much a later compute partnership alone says about Amazon’s competitive position. O’Driscoll distinguishes that opinion from the observed growth figures discussed, and the guests agree that infrastructure demand and future competitive success are different questions.

Asked about public-company pricing, Lemkin says Google looked underappreciated to him relative to Amazon. He acknowledges that public equities are not his primary specialty; the comparison is a time-specific opinion, not investment advice.

O’Driscoll describes Google’s search, cloud and AI capabilities while noting that a new competitor can still put pressure on an incumbent. His assessment of future market outcomes is a forecast rather than proof of which company will win.

Stebbings discloses being a Meta shareholder before asking about capital expenditure and market reaction. O’Driscoll says investors may question how spending will earn a return, while recognizing that he does not know management’s full plan.

The group turns to Twilio and other mature software companies. O’Driscoll contrasts cash-generating businesses with high-growth AI companies and says their public multiples may reflect different growth expectations; the figures are examples, not a valuation recommendation.

Lemkin argues that existing software companies should look for relevant AI-related demand and cites reported voice-AI activity at Twilio. He presents the company statistics as his understanding; they are not independently checked in the episode.

O’Driscoll agrees that even a modest reacceleration in growth can affect valuation, but says incumbents need to find a genuine connection between their product and new customer needs. His multiple examples are rough market comparisons, not guarantees.

Lemkin describes rapid growth at WorkOS as an example of benefiting from AI-related demand, while warning that he may have the numbers wrong. He acknowledges that some adoption can be the result of timing rather than a universally repeatable strategy.

He argues forcefully that companies should change teams if they cannot respond to AI. This is his personal management opinion; the panel offers no evidence that broad layoffs are the right remedy for every company or that employees alone determine a firm’s strategy.

Lemkin says he has used Salesforce Agentforce and considers the product competitive, but questions whether its value will translate into revenue growth. His experience is an anecdote, not a controlled evaluation or evidence of company-wide adoption.

The guests discuss whether software vendors can earn revenue by automating work currently performed by people. Lemkin is optimistic based on his experience; O’Driscoll says the pace and scale of diffusion through the economy remain uncertain.

Lemkin describes using sales agents and says referrals to vendors he uses generated business. The figures are his own account and do not establish independent vendor performance, employment outcomes or a general rate of return.

Stebbings cites OpenEvidence adoption as an example of rapid individual uptake. The number and comparison with Doximity are unverified in the program, and enthusiasm for a medical information tool does not establish clinical accuracy or replace professional judgment.

O’Driscoll says a product may reach a finite professional audience quickly, making subsequent expansion important. He distinguishes individual adoption from the often slower process of enterprise procurement and implementation.

Lemkin says founders who move slowly may find opportunities in industries where adoption is also slower. The panel contrasts individual use with company-wide deployment, while treating the proposed timelines as estimates rather than measured forecasts.

Asked whether Series A investing has become harder, Lemkin argues a larger pool of seed-stage startups can create opportunities for later investors, even though pricing and competition are difficult. He sees seed investing as particularly crowded.

O’Driscoll agrees that the range of AI startups creates opportunities but says capital and competition require investors to build relationships early, decide quickly and understand where they have an advantage. His assessment reflects his work at Scale Venture Partners.

In a final hypothetical comparison, the panel discusses prediction-market companies Kalshi and Polymarket. Valuations and regulatory status are not independently verified, and the speakers acknowledge uncertainty about how future rules may change.

O’Driscoll raises integrity risks around sports micro-bets and information that a participant might influence. His examples are hypothetical or opinions about potential oversight; they do not establish misconduct by any named person or platform, and applicable rules depend on jurisdiction.

The guests close with a discussion of sports and future oversight. Their market preferences and regulatory forecasts are personal views at the recording date, not legal advice or a recommendation to use a particular platform.