# Microsoft CEO Warns That Companies Embracing AI Could Drive Themselves Out of Business > Source: > Published: 2026-07-29 14:59:06+00:00 The business world has quickly learned that there’s such a thing as using too much AI, as they get slammed with outrageous token bills. Now, Microsoft CEO Satya Nadella is warning of another way that embracing AI [could doom a business](https://futurism.com/future-society/executives-corporations-ai-bills-metered-billing). In an interview Sunday, Nadella cautioned against sharing too much information, ranging from proprietary data to the prompts entered, with a third-party AI model. Instead, he recommended that businesses should make sure they retain all the data on how employees use AI tools so they can eventually train or fine-tune their own [open-weight model](https://futurism.com/future-society/easy-poison-open-weight-ai). This can be accomplished by using AI gateways, which serve as a protective intermediary. “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 on an episode of *CNN*‘s Fareed Zakaria GPS, [as spotted by TechCrunch](https://techcrunch.com/2026/07/27/satya-nadella-says-companies-that-trust-one-ai-for-everything-may-not-survive/). That’s a striking claim — and an ironic one, since outsourcing your thinking is arguably exactly the point of using AI, of which Microsoft is a leading vendor. Nadella, in a nutshell, is saying that heavily relying on proprietary AI software could also be a company’s undoing. He may have an angle in pushing these claims, though. Microsoft hosts the cloud infrastructure that businesses can run their open weight models on. The more people run to hosting their own models, the better for its bottom line. There’s been a lot of hype lately around open-weight models [as a cheaper alternative to those offered by the top AI labs](https://futurism.com/artificial-intelligence/us-realizing-chinese-ai-models-way-cheaper). In AI, “weights” are the numerical values that determine how a model processes information. The models are “open” because these weights are publicly accessible and free to download. Agentic AI tools like Anthropic’s Claude Code, by contrast, are closed-source. They also come in a “harness,” which is essentially the software the underlying model is couched in. Nadella thinks companies should ditch AI harnesses and instead use a custom AI platform that allows you to recruit multiple AI models — like a cheaper, open-weight one. “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 in the interview, per *TechCrunch*. “At the same time, any one model can go away, and you can still continue to be in control of your own destiny.” Getting mixed signals from Nadella here? TechCrunch also notes how he suggests, in pretty hypocritical fashion, that while businesses should be stingy about the data they share with AI models, regular users shouldn’t. “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,” he mused. “That’s sort of how the advertising business model has worked.” **More on AI: *** Sam Altman Says Even the Power of AI Will Never Lead to a Shorter Work Week*