Microsoft CEO Satya Nadella has launched a personal blog, sn scratchpad, and signalled that 2026 could be a pivotal year for AI as the industry moves from early experimentation to wider real-world deployment. In his first post, dated Dec 29, 2025, Nadella argues that the value of AI will depend less on raw model capability and more on how it is applied to create measurable outcomes for people and organisations.
From “discovery” to “diffusion”
Nadella writes that AI is moving beyond the initial “discovery” phase into “diffusion,” where the focus shifts from impressive demos to practical use—separating “spectacle” from “substance.” He describes a “model overhang,” where capability is advancing faster than the ability to translate it into real-world impact.
‘Bicycles for the mind’—updated for AI
Building on Steve Jobs’ “bicycles for the mind” analogy, Nadella frames AI as scaffolding for human potential, not a substitute for people. “What matters is not the power of any given model, but how people choose to apply it to achieve their goals,” he writes, arguing the industry should move beyond “slop vs sophistication” debates and develop a clearer product and social framework for AI’s role.
From models to systems
A central theme is a shift from standalone models to engineered AI systems—with orchestration across multiple models and agents, plus safeguards around memory, access/entitlements, and tool-use. Nadella notes that models still have “jagged edges,” and argues the next phase will be defined by building reliable scaffolds that deliver consistent value in messy real-world environments.
Why Nadella says 2026 matters
Nadella adds that the coming year will require deliberate choices about how AI is deployed, given constraints such as energy, compute, and talent—and that “societal permission” will hinge on demonstrable real-world impact.
By – Charu Mandhyan