Nadella: Build ‘Token Capital’ or Be Doomed by Big AI

In his June 15th blog post, Microsoft CEO Satya Nadella warned companies to develop their own “token capital” — proprietary AI — in addition to human capital. This means training AI models on your own data, creating your own learning loops, and making your company’s knowledge available via query instead of just relying on third party models.

The reason? If you don’t have your own AI, “you’ll end up with a small number of AI systems capturing all the economic returns, while entire industries have their knowledge commodified,” writes Nadella. And if only a few models benefit, “the political economy will simply not tolerate it.” “You can offload a task, or even a job, but you can never offload your learning.”

1The self-interest angle

Keep in mind that Microsoft itself is currently behind the state-of-the-art on a number of key AI models. In this sense, the idea that the base model is interchangeable and the true value lies in your unique data (and associated tooling) steers companies directly to Microsoft Azure and Office-based AI. Either way, Microsoft is the infrastructure provider. It’s both correct and self-serving.

2Why it matters to a normal person

This thinking applies on an individual level, too. Your long term competitive advantage wasn’t the model itself, since these change frequently and become cheaper over time. Rather, it’s the data you collect, the notes you take, the decisions you make, and the workflows you follow. If you’re using AI to perform tasks without retaining any learning, then you’re paying for a service. But if you capture that reasoning (i.e., your prompts, your edits, your log of what has worked in the past), then you’re building something that persists across model iterations.