Microsoft CEO Satya Nadella has reinforced a warning to businesses adopting AI: companies that rely entirely on proprietary AI providers for their AI needs may eventually lose control of their future.

 

Speaking on CNN’s Fareed Zakaria GPS, Nadella said businesses should carefully manage everything they share with AI providers, including their data, prompts, context, and usage information. He argued that companies should retain ownership of the metadata generated when they use AI so they can eventually train their own models or fine-tune open-weight models.

 

“Any firm that doesn’t have this control, I will claim will not remain a firm because you’ve essentially outsourced your thinking,” Nadella said.

 

His message is clear: enterprises should avoid becoming dependent on a single AI model or provider. Instead, they should build an independent AI infrastructure layer that keeps their prompts, context, memory, and workflows separate from the underlying models.

 

Nadella also encouraged businesses to avoid relying entirely on AI providers’ built-in coding tools, such as Anthropic’s Claude Code and OpenAI’s Codex. By separating the AI application layer from the model itself, companies can use multiple models for different tasks and continue operating even if one provider changes its strategy, raises prices, or disappears.

 

“By keeping the harness separate from the model and the context and memory separate from the model, you absolutely can use multiple models for what they’re great at,” Nadella said. “At the same time, any one model can go away, and you can still continue to be in control of your own destiny.”

 

The warning is notable because Microsoft is a major investor in both OpenAI and Anthropic. Microsoft also benefits from enterprises adopting the cloud infrastructure and AI management tools that Nadella recommends.

 

Still, the broader concern is gaining traction. Enterprises are increasingly adopting multiple AI models, including lower-cost and open-weight alternatives that they can customize and run on their own infrastructure. As a result, companies are looking for ways to manage multiple models and build AI agents that are not tied to a single provider.

 

The issue is about more than cost. Nadella warned that once a company has “outsourced its thinking” to an AI provider, the provider could potentially use its growing knowledge of the company’s workflows and business needs to launch competing products or services. That risk may become even greater as businesses give AI agents access to internal systems, data, and operational processes.

 

This concern has long existed in the startup ecosystem. Founders have worried that major AI platforms could study successful applications built on their models and eventually introduce competing features. Nadella is now extending that warning to large enterprises.

 

However, Nadella drew a distinction between businesses and individual consumers. When asked how everyday users could protect their data, he suggested that sharing some information is part of the value exchange for using digital services, particularly free services.

 

The larger takeaway for enterprises is clear: AI may power the future of business, but companies should retain control of their data, prompts, context, memory, and AI workflows. The goal should not be to depend on one AI model for everything, but to build a flexible, multi-model AI strategy that keeps the organization in control.