(Credit: Jonathan Raa/NurPhoto via Getty Images)
The close relationship between Microsoft and OpenAI is no secret. After all, Microsoft picked the GPT family of AI models as the foundation for Copilot and made an 11-figure investment in Sam Altman’s firm. That’s why one announcement from Microsoft's business- and developer-centric Build conference sticks out.
Prior to this morning's keynote, we spotted an interesting paragraph in the ninth section of the publicly available Book of News for this year's Build is the following paragraph:
Azure AI Foundry Models is expanding with new direct, first-party offerings hosted and billed directly by Microsoft, including Grok 3 and Grok 3 mini models from xAI. These models will have all the service level agreements (SLAs) Azure customers expect from any Microsoft product.
Forty-five minutes into the Build opening keynote presentation, after Microsoft CEO Satya Nadella announced that integration (and well after Altman made his own appearance), xAI CEO Elon Musk popped up on-screen for a pre-recorded chat with Nadella.
Grok is a competing AI chatbot integrated into Musk’s X social network. It has a questionable track record, producing bizarre and sometimes dangerous results, which Musk acknowledged in his Build cameo. "We have and will make mistakes, but we aspire to correct them quickly," he told Nadella before urging Microsoft developers to kick the tires on Grok and provide feedback, and teasing the imminent release of Grok 3.5.
Like other big tech companies, Microsoft seems to want to cozy up to the head of DOGE; it's always useful to have a friend in Washington. But Microsoft hasn't switched its main AI tool, Copilot, from OpenAI GPT models to Grok. In fact, at Build, Microsoft announced that Copilot would implement OpenAI’s GPT-4o image-generation model. So, I don't think securing government work is the main reason for the Grok additions.
Besides, it’s not even Microsoft’s first involvement with xAI. Earlier in 2025, the two companies joined forces with Nvidia on an AI infrastructure partnership. Plus, Microsoft previously made the cheap and fast Chinese AI model DeepSeek R1 available to Azure developers and added Meta’s Llama 2 model before that. Clearly, the company is opportunistic when it comes to embracing and integrating generative AI tech.
Google doesn't seem to be in a rush to add Gemini to Azure, but it wouldn't shock me if some of that tech eventually makes its way over. I expect to hear plenty about Gemini at Google I/O, which starts on May 20, and a Microsoft cameo at that event isn't entirely out of the question.
What's Microsoft's End Goal?
Nadella hasn't been subtle about the company's goals. He has repeatedly made the clever remark that Copilot is "the UI of AI." Now it seems Microsoft wants Azure to be the clearinghouse of generative AI development. This seems like a wise goal since AI is still the hottest game in town, even if it means adding tech with an unsavory history.