Asked whether the remaining problems are intelligence-bound, Tan says “it’s all human — humans are the ones who are going to slow this down,” then reframes that as good news. Against a discourse he characterizes as “all white-collar jobs are going to go away, there’s going to be a permanent underclass,” he argues the opposite: “all of the bureaucracy and the slowness and the seven plus or minus two of like every company, every institution, every organization, all the middle managers in the world — that is actually the white pill.” His estimate is explicit: “it’s going to be 20 years and that’s not a bad thing. That’s actually a good thing. Society is way slower than you think. Government is way slower than you think. Every company in the world is way slower than you think.”
He then says the quiet part, flagging it himself as heterodox: “there are real moats to these things. Like there are structural reasons why a Microsoft isn’t going anywhere… this is sort of a little antithetical to what most people in Silicon Valley would say, like ‘I want a startup to replace all of them.’ And the reality is like they won’t.”
This sits in direct tension with In AI the threat is layer migration, not a competitor — work relocates across layers when any variable moves, which treats value as continuously and rapidly relocating between layers. Both can be true on different clocks — the layer boundary can move fast while the institution built around the old boundary moves slowly — but they prescribe different strategies, and the tension is the useful part. If Tan is right, the play is not to race an incumbent to its own market but to build inside the gap its inertia creates, which is a stronger version of the claim in The system of work is the moat, not the model — the model is fungible underneath. It also explains why The gains come from redesigning work around AI, not bolting AI onto human workflows describes a redesign that large orgs consistently fail to execute: the failure is structural, not a lack of will.