Satya Nadella just told every company on the planet that picking the fanciest AI model is not a strategy. It’s a shopping trip.
The Microsoft CEO posted a memo on X on June 14 that racked up tens of millions of views, laying out a framework for how organizations should actually think about AI competitive advantage. The core argument: companies need to build their own proprietary “learning loops” rather than outsourcing their intelligence to whichever foundation model is trending this quarter.
Human capital meets token capital
Nadella introduced two concepts that are worth paying attention to. The first is “human capital,” which covers employee expertise, institutional workflows, and the messy real-world knowledge that lives inside organizations. The second is “token capital,” his term for the proprietary AI assets a company develops and improves over time.
Nadella also coined the term “token-maxing” to describe what he wants companies to avoid. That’s the practice of simply throwing more compute and bigger models at problems without building the feedback infrastructure that makes AI actually useful in context.
