Microsoft CEO Satya Nadella published a stark warning for enterprises using third-party AI models, arguing that companies are effectively paying for AI intelligence twice — once in token fees and again by handing over the proprietary knowledge that makes those models useful.
In a blog post published Sunday, Nadella warned that the better a company wants an AI model to perform, the more sensitive business information it must feed into the system. “Models learn from ‘exhaust,’ the prompts people write, the tools agents use, and especially the corrections people make when the model is wrong,” he wrote. “Every correction is distilled into institutional know-how.”
Nadella described this as “the kind of knowledge a competitor could never buy,” yet enterprises are voluntarily handing it over to model providers like OpenAI and Anthropic. He joins a growing chorus of voices — including venture capitalist Jason Calacanis and Palantir CEO Alex Karp — warning that AI labs could eventually use customer data to compete against their own clients.
The warning carries added weight coming from Microsoft, which has invested more than $13 billion in OpenAI and integrated its models deeply into Azure and Copilot products. Microsoft has also begun developing its own in-house MAI models to reduce its reliance on external AI providers.
Nadella’s critique comes as enterprises accelerate AI adoption while grappling with questions about data governance, vendor lock-in, and the long-term competitive dynamics of relying on models trained on their own proprietary information.
