[AINews] Satya on Loopcraft: Building Frontier Ecosystems

Satya Nadella published his first X article this week, and it crossed 60 million views. The argument is direct: the real competition in AI is not between models but between learning loops. Nadella calls this Loopcraft, framing it as a new theory of the firm where human capital and token capital compound together inside an organization. The CEO of Microsoft is, for the first time since the OpenAI breakup eight months ago, stating his AI strategy in plain terms.

The core claim is that cognitive loops between people and digital systems represent genuinely new organizational infrastructure. You can offload a task or a job, Nadella writes, but you cannot offload your learning. That sentence is doing real theoretical work. It positions Microsoft not as a model vendor but as the provider of the substrate on which institutional knowledge compounds. If you have been tracking the Big Model versus Big Harness debate, you have heard adjacent arguments before. You have not heard a sitting Fortune 10 CEO commit to it this publicly or this precisely.

The full piece is worth reading for the specifics on how Nadella defines token capital, why he frames frontier ecosystems as the priority over frontier models, and how he connects individual organizational learning loops to value flowing across every industry and country. The Latent Space podcast conversation with Nadella from MS Build, linked in the original, gives the technical context the X article compresses. Read both together.