Satya Nadella reshapes Microsoft’s leadership to sharpen its AI strategy

Financial Times reports, that the changes rolled out over the course of the year and touched both engineering and the commercial side of the business.

The most notable external hire is Jay Parikh, a former Meta engineering leader. He was tapped to run CoreAI – Platform and Tools, a group meant to bring Microsoft’s AI platform and developer tooling under one roof. At the same time, Nadella elevated Judson Althoff in the commercial organization and expanded the remit of Ryan Roslansky, who remains CEO of LinkedIn but is taking on additional responsibilities across Microsoft’s ecosystem.

For Nadella, this is not about reshuffling names on an org chart. He wants Microsoft to move faster on its own AI models, improve the Copilot assistant and the services around it, and strengthen developer tools and enterprise applications where AI has to behave in a predictable, controllable way. Nadella also sees the shifts as a way to assess potential successors and give new leaders more room to operate.

Current and former Microsoft employees interviewed by FT connect the overhaul to two drivers.

First, Microsoft wants to build its in-house AI capabilities more quickly, from models and infrastructure to programmer tools and AI features inside apps, so the company is not dependent on a single external technology provider.

Second, competition is heating up. Amazon and Google, once often described as playing catch-up, made visible progress in 2025 in both infrastructure and model development. Inside Microsoft, this period is described as a more hands-on leadership style. Deputy CTO Dee Templeton compared it to “founder mode,” a term popularized by investor Paul Graham. Nadella has been trying to simplify layers and speak with teams more directly so product decisions happen faster.

This is not his first high-speed pivot. GNcrypto notes that after becoming CEO in 2014, Nadella shifted Microsoft away from a Windows-first mindset and mobile ambitions toward Azure and a more open stance on open source. A similar logic is now being applied to AI: keep the partnership with OpenAI, but build Microsoft’s own AI stack in parallel and shorten the path from research to product, even if that creates internal friction.