Rory O'Driscoll highlights that neoclouds like CoreWeave operate with heavy debt loads and tens of billions in compute CapEx. Because they lack the flexibility of pure software, any slight market slowdown in AI demand threatens their entire debt structure and profitability.
Jason Lemkin and Rory O'Driscoll discuss defense and AI startups trading at 40x revenue while durable public software companies sit below 3x. Unless extraordinary growth continues indefinitely, late-stage venture deals face painful valuation compression when exiting into modern public markets.
Jason Lemkin explains that when managing 5 to 10 billion dollar funds raised every 18 months, small exits cannot move the needle. Mega funds must construct positions capable of yielding at least a billion dollars in gains per deal, pricing out traditional mid-sized exits.
Jason Lemkin argues that LLMs have commoditized writing code, but getting that code to production-intent quality requires immense effort. Tools solving automated testing, sandboxing, and code review address the true operational hurdle holding back AI-generated software deployments.
Rory O'Driscoll observes that public markets reward steady management fee income over volatile carry. To justify public valuations, alternative asset managers must prioritize continuous AUM expansion and new fund products, inevitably shifting their cultural incentives away from pure, high-risk venture bets.
Harry Stebbings and Rory O'Driscoll point out that sky-high seed valuations cut allocations down to 300k to 500k checks. Meanwhile, massive late-stage platforms like SpaceX and Anthropic have expanded by trillions, causing late-stage capital gains to historically outstrip early-stage venture returns.
Jason Lemkin notes that developer tools like Replit operate under tight gross margins and switch default models purely on price and parity. By observing how platforms route prompts away from expensive closed models, investors get an objective view of LLM cost and performance.
Jason Lemkin points out that taking even one venture round complicates dividend distributions and investor buyouts later. Unlike bootstrapped businesses that can distribute millions in annual profits, single-round companies often end up stuck between misaligned venture expectations and illiquidity.
