Anthropic and OpenAI are adding revenue faster than hyperscalers ever did. David George of a16z and David Clark of VenCap put a number on it: the top 1% of AI company exits have grown 10x in 24 months. That is not a rounding error. It is a structural shift in how fast value concentrates at the frontier.
The conversation is worth reading for the mechanics, not just the conclusions. The 40% annual leadership churn stat is the sharpest thing here: four in ten AI market leaders drop off every year, which makes durable winner selection the core problem in venture right now. They also work through the token cost curve, who captures value as inference gets cheaper, and why most enterprise AI apps are still skeuomorphic, built to look like old software rather than exploit what the technology actually does.
The bubble question gets a direct answer, and the framing around SpaceX, OpenAI, and Anthropic IPOs reshaping public market expectations is specific enough to matter for anyone thinking about late-stage or crossover positioning. George's argument that native AI companies run their operations differently from legacy software firms is the thread that ties the whole discussion together, and it is underdeveloped enough in the summary to make the full conversation worth your time.
[WATCH ON YOUTUBE →]