On Sunday, Microsoft CEO Satya Nadella reinforced his earlier cautionary message to businesses using AI, intensifying his stance. He foresees that companies entirely dependent on proprietary AI labs for their AI requirements will not endure in the long run.
Nadella expressed this view on CNN’s “Fareed Zakaria GPS.” When asked by Zakaria to clarify what over-sharing with an AI model provider means, Nadella emphasized that businesses must be cautious about all they disclose, from data to prompts.
Nadella advocated for a configuration where “every time you use the model, all of the metadata around it is retained by you, so that you could use all of that to train perhaps your own weights or your own open model.” (Weights refer to a model’s trained parameters, acting like its brain. Nadella’s argument is that companies should preserve their own usage data to eventually develop their own models.)
“Any firm that doesn’t have this control, I will claim will not remain a firm because you’ve essentially outsourced your thinking,” he added.
In summary, Nadella warns that companies lacking their own models—or a layer of AI infrastructure known as AI gateways to keep their prompts separate from the model—face significant risk.
He urges companies to refrain from depending on AI labs’ built-in coding tools, known as harnesses. Examples include Anthropic’s Claude Code and OpenAI’s ChatGPT Codex.
“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. At the same time, any one model can go away, and you can still continue to be in control of your own destiny,” Nadella said.
Notably, Microsoft invests in the two largest AI labs, Anthropic and OpenAI. Coding agents are popular among enterprises using AI models and reportedly generate substantial revenue for the model creators.
Yet, Nadella advises enterprises against excessive reliance on them. Microsoft’s cloud business could benefit from this warning, as it offers the kind of alternative infrastructure Nadella recommends.
Despite appearing self-serving, Nadella’s point is valid. Enterprises increasingly recognize the need for diverse model options, especially more affordable ones, and are turning to open-weight models—those with publicly available underlying code—that they can fine-tune and operate on their own hardware. This shift also necessitates systems for managing multiple models and coding agents independent of a specific model provider.
Nadella’s insights extend beyond financial concerns. He predicts that once a company has “outsourced its thinking” to a model, there’s little to stop the AI lab from later competing with them. This risk escalates as businesses adopt AI agents and allow them deep access to their operations.
This is a long-standing concern in the startup industry: What prevents model makers from eliminating startups by emulating and competing with them?
In May, for example, when OpenAI CEO Sam Altman offered every Y Combinator startup in its latest cohort AI credits, investor Jason Calacanis warned about the risk, stating: “If you take these tokens, there’s a non-zero chance that OpenAI will study exactly what your startup is doing, copy your idea and put your app into their free offering. This is the classic platform playbook — be careful, founders!”
Now, Nadella delivers a similar message to enterprises.
A caveat: Nadella’s concerns about oversharing with AI models apply solely to businesses and not individuals. When Zakaria inquired about everyday people’s protection, Nadella dismissed it, noting that data sharing is the tradeoff consumers make for using a service, particularly free ones.
“To some degree there’s got to be some value exchange in the consumer space where you’re getting something for free, maybe for your data. That’s sort of how the advertising business model has worked,” Nadella said.
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.
