Microsoft CEO Satya Nadella thinks AI will reshape what companies fundamentally are.

A "real cognitive loop" is forming between people and digital systems for the first time, Nadella writes in his blog. In addition to human capital, companies will also need "token capital" going forward, meaning AI capabilities they own and control.

Nadella gets specific about what that requires: Companies should build proprietary learning systems. Private evals should track whether AI models are actually getting better at business-relevant outcomes. Internal training setups should improve models with real company data, and institutional knowledge should become queryable and reusable.

"This means the real opportunity is not in picking the best model but instead in building a learning loop on top of models where human capital and token capital compound," Nadella writes. "You can offload a task, or even a job, but you can never offload your learning. The future of the firm is the ability to compound that learning across people and AI."

The real test is whether a company can swap out the base model without losing the knowledge it has built on top, Nadella says. The Microsoft CEO is, of course, arguing his own book here. Microsoft trains its own AI models, but so far they lag the competition. The company is instead trying to keep enterprises locked into its Azure and tooling stack, for example, through AI bundle deals tied to Office products.

Nadella's tone on AI model commoditization is shifting

Nadella's tone has shifted, too. In March 2025, the Microsoft CEO said, "The models are getting commoditized," with the real value living in products and the system stack. Just over a year later, he sounds less sure about the commoditization part.

"The last thing any of us want is a world where every company across every sector is ceding value to a few models that eat everything they see. If all the value is accrued by only a few models, the political economy will simply not tolerate it. There is no societal permission for an AI future that hollows out entire industries," Nadella writes.

What might get commoditized instead, Nadella warns, is company knowledge. He describes "a small number of AI systems capturing all the economic returns, while entire industries find their knowledge commoditized right out from underneath them."

Nadella's warning makes the most sense if model commoditization is moving slower than he had hoped. OpenAI and Anthropic may have models other companies can't easily match, and they're also wrapping entire product ecosystems around those models, where the line between model capability and the "agent" harness keeps blurring. That could create exactly the kind of concentration Microsoft usually claims for itself.

AI News Without the Hype – Curated by Humans

Subscribe to THE DECODER for ad-free reading, a weekly AI newsletter, our exclusive "AI Radar" frontier report six times a year, full archive access, and access to our comment section.

Subscribe now