Microsoft CEO Satya Nadella has warned that companies using artificial intelligence risk losing not only money but also their proprietary knowledge. In a personal blog post published on July 15, 2026, Nadella said organizations effectively pay two types of costs when adopting AI—financial and intellectual—and the latter often goes unnoticed. He explained that to use AI effectively, firms must share internal data and expertise, which can become valuable assets for AI providers over time.

To illustrate his point, Nadella referenced economist Kenneth Arrow’s “information paradox,” noting that in the AI era, companies are now giving away their valuable knowledge to providers, creating long-term risks. He emphasized that AI suppliers continuously learn from user data and behavior, while client firms remain unaware of how their information is used.

Nadella urged companies to strengthen control over knowledge generated through AI use, describing this process as a “learning loop.” He called for new “trust boundaries” to ensure customer data and internal processes are not used without permission and advised building flexible infrastructures that reduce dependency on single AI providers.