Microsoft's Nadella argues companies must build proprietary AI capabilities paired with human judgment to avoid ceding value to a handful of dominant AI providers.

What happened: Microsoft CEO Satya Nadella outlined a framework for enterprise AI strategy centered on what he calls "token capital"—a company's proprietary AI systems and models—paired with "human capital" (employee knowledge and judgment). He argued that human direction must steer AI systems, and that the real competitive advantage comes from building learning loops that let companies improve their AI over time while retaining control of their intellectual property.

Why it matters: Nadella warned against a future where a few dominant AI models capture most of the value, comparing it to how globalization hollowed out industrial economies. He stated that concentrating AI value in only a few models would lack "societal permission," implying the political and economic stakes are high. Enterprise AI spending is already outrunning corporate forecasts, making the question of whether that investment buys genuine company capability or dependency on external providers increasingly urgent.

What to watch: Nadella's key "test" of control is whether a company can swap its underlying base model without losing the accumulated expertise built into its learning loop. He described this as the measure of "control and sovereignty in the era ahead"—suggesting that companies retaining the ability to shift models while preserving their institutional knowledge will be the winners in the AI economy.