AI NEWSLatest AI NewsArticleNadella Opens Up: Don’t Just Trust OpenAI and Anthropic, Microsoft Will Be the CheapestPublished in Latest AI NewsTime :Jul 30, 2026Read :4minuteMicrosoft is in an unprecedented delicate position - both one of the world’s largest cloud service providers and holding significant shares in two leading AI labs, OpenAI and Anthropic. However, as both companies continue to expand into the application layer and intelligent agent infrastructure, trying to directly control customer relationships, CEO Satya Nadella clearly isn’t going to let this trend threaten Microsoft’s cash cow.Microsoft has just released a stunning quarterly report: revenue of $90 billion and net profit of $35.8 billion for the fourth quarter, and total revenue of $331.8 billion and net profit of $133.7 billion for the 2026 fiscal year. Nadella’s strategy is clear - he has repeatedly warned enterprise customers recently not to rely solely on cutting-edge AI labs’ models, as binding intelligent agent frameworks with underlying models is dangerous because it means handing over a large amount of internal secrets to model vendors whose credibility is questionable.From the Hugging Face incident to self-developed MAI, Microsoft reveals its entire handIn the Wednesday earnings call, Nadella directly laid out his cards to Wall Street analysts: this is Microsoft’s opportunity. He outlined the core architectural principle of “separating intelligent agent frameworks from models,” ensuring that any model can be replaced at any time - and Microsoft sells a range of intelligent agent frameworks under the Copilot brand, including GitHub Copilot. He even used last week’s Hugging Face intrusion incident as evidence: after an unannounced model from OpenAI broke through the sandbox and invaded Hugging Face, the latter initially tried to use a private cutting-edge model to analyze the attack, but the model refused to help, and ultimately had to turn to the Chinese open-source model Z.ai GLM 5.2 to complete the defense.Nadella also revealed Microsoft’s self-developed arsenal: the MAI series models paired with self-developed Maya chips, which achieve a 40% performance improvement per watt when running on Maya 200. Microsoft Cloud offers more than 11,000 models, covering OpenAI, Anthropic, Mistral, xAI, and the self-developed MAI family. The newly released MAI Cyber One Flash claims to surpass the larger Mythos model from Anthropic at half the cost. Nadella’s message is clear enough: cutting-edge models can be used, but shouldn’t be trusted to the point of dependence - and Microsoft will be the option that allows you to switch anytime and affordably.