Tan is careful not to trade one orthodoxy for another — “I think there are going to be as many Patrick Collisons as ever” — but reports a distinct shift: “one mega trend that we’re seeing is like a 35, 40, 45 year old founder who has been around the block, built a lot of companies” — and of one such founder, Peter Steinberger, “he’s been around the block, like he knows what to build.” The leverage follows: “if you take that person, like suddenly there’s 400 of those people, like you can outperform an entire department of any Mag Seven.” Earlier in the conversation he makes the multiplier general: with agentic coding, “literally any given person could be 400 of that person.”
The reasoning is arithmetic rather than generational. If the multiplier is roughly constant across people, then what varies is the thing being multiplied, and the scarce input is knowing which thing to build. Experience used to come bundled with a penalty — slower, more expensive, more encumbered — and the multiplier removes most of that penalty while leaving the judgment intact. Tan also frames his own path as evidence that the judgment is learned rather than innate: he calls himself “a late bloomer” whose life is “somewhat like a testament to… I can make mistakes, and then I can introspect, I can reason about them and then make different choices.”
This is AI compresses the distance between idea and execution but not between good and bad judgment followed through to who wins: if judgment is the binding constraint, the advantage accrues to whoever has accumulated the most of it. It is the individual-scale version of Commodity work's terminal value is zero but structured expert judgment compounds indefinitely, where commodity capability trends to zero while expert judgment compounds, and it explains the mechanism behind AI is steel for organizations — when software carries the context, human communication stops being the load-bearing wall from the other direction — one experienced person plus context-bearing agents can now span what previously required an org.