Most companies have convinced themselves that AI strategy means model selection. Pick the right frontier model, layer it on your operations, and stay close enough to the cutting edge that you don't fall behind. Microsoft CEO Satya Nadella published an essay this week that calls that entire approach a category error.
The argument is disarmingly simple: in a world where every company has access to the same general-purpose models, the model itself stops being the differentiator. What matters is what you build around it. Nadella introduced the term "token capital" — the idea that owned AI capability, built by feeding institutional knowledge back into systems over time, functions like a new class of strategic asset, one that sits alongside human capital on the balance sheet of every serious business. The companies that don't build it, he argued, risk something more serious than falling behind. They risk transferring what they know to the models everyone else uses too.
