Microsoft is prepping to host Elon Musk's Grok on Azure
Microsoft is reportedly in talks with xAI to host Elon Musk’s Grok AI model on Azure AI Foundry for developers and internal teams. The move reinforces Satya Nadella's push to turn Azure into the go-to home for every major AI model, even if bringing Musk's tech aboard escalates friction with partner OpenAI.

Microsoft is preparing to bring Elon Musk’s Grok AI to Azure.
According to a report from The Verge, Microsoft has spent the last few weeks in talks with xAI and is already instructing its infrastructure engineers to get ready to host the Grok model on Azure AI Foundry. If the deal closes, developers will be able to build apps with Grok, and Microsoft could even weave the model into its own products.
Nadella wants Azure to host everything
Microsoft CEO Satya Nadella is reportedly the driving force behind the partnership. Nadella wants Azure to be the default home for every buzzy AI model on the market, regardless of who built it.
We saw this playbook earlier this year when Nadella rushed engineers to deploy DeepSeek’s cheap R1 model in just a few days. The deal with xAI appears to focus strictly on inference hosting rather than training; Musk previously walked away from a potential $10 billion Oracle server deal to train future xAI models internally.
Microsoft could announce the Grok partnership as early as its Build developer conference on May 19th.
The widening rift with OpenAI
Hosting Grok is a spicy move internally, not just because of Musk’s controversial political work with the Department of Government Efficiency, but because of Microsoft’s fraught partnership with OpenAI.
Musk and OpenAI are locked in a bitter legal brawl, with OpenAI recently countersuing Musk over claims he is using bad-faith tactics to derail the lab. Meanwhile, The Wall Street Journal recently reported that Nadella and OpenAI CEO Sam Altman are drifting apart.
Nadella brought in Inflection co-founder Mustafa Suleyman last year as an insurance policy against OpenAI, but Microsoft’s in-house model efforts have reportedly stumbled. Copilot consumer adoption has flatlined at roughly 20 million weekly users, according to Newcomer, leaving Microsoft still dependent on OpenAI tech.
To make matters worse, Microsoft was expecting OpenAI to deliver GPT-5 this month. But with model delays and GPU capacity crunched by its upgraded image generator, GPT-5 is unlikely to arrive on schedule. Hosting Grok proves once again that Microsoft is actively diversifying away from Altman.
Why it matters
For enterprise developers, having Grok on Azure AI Foundry makes testing xAI models friction-free. You get enterprise security, unified billing, and the ability to swap between OpenAI, DeepSeek, and Grok without switching cloud providers.
We think this is a smart, calculated pivot by Microsoft. Building frontier models in-house has proven remarkably hard, even with top talent like Suleyman. Instead of waiting around for OpenAI to fix its capacity bottlenecks and hand over GPT-5, Microsoft is positioning Azure as the Switzerland of AI infrastructure.
However, the strategy comes with serious friction. Partnering directly with Musk, OpenAI’s fiercest critic, rubs salt in an already strained relationship. It looks increasingly likely that the exclusive era of the Microsoft-OpenAI alliance is over, and both sides are openly preparing for a post-partnership world.
Sources (1)
- theverge.com(opens in a new tab)Originaltheverge.com
Topic trackerMicrosoft and OpenAI's strategic relationship
- Microsoft adds Elon Musk’s Grok 3 to Azure
- Microsoft is prepping to host Elon Musk's Grok on AzureThis story
- Microsoft turns 50 with Copilot updates and in-house AI ambitions
- Microsoft AI chief wants out of OpenAI’s shadow
- Microsoft and OpenAI tweak their mega-deal through 2030