Microsoft is exploring the possibility of running Grok, an AI model developed by Elon Musk’s company, xAI, on its cloud platform, Azure. According to sources familiar with the situation, the tech giant has already directed its engineers to prepare the necessary infrastructure for the integration, should the deal move forward.
If the collaboration proceeds, Grok would likely be added to Azure AI Foundry, a development hub that provides tools and pre-built models for developers working on AI applications. This would allow developers to incorporate Grok into their own projects, as well as give Microsoft’s internal teams access to the model. While Microsoft declined to comment on the matter, this potential partnership could mark a significant shift in the cloud AI landscape.
Azure has gradually expanded its range of AI models, including some that compete with OpenAI’s offerings, which has been a key partner for Microsoft. Earlier this year, the company quickly onboarded DeepSeek’s R1 model—a low-cost alternative from a Chinese startup—onto Azure AI Foundry in response to growing interest in diverse AI models. This was part of a broader strategy to ensure Azure remains a top platform for hosting AI services.
CEO Satya Nadella has been particularly invested in making Azure a central hub for a wide variety of AI tools. The company’s infrastructure team has been working overtime to ensure that Azure can meet the technical demands of new AI technologies.
Asha Sharma, Microsoft’s corporate vice president of AI platform, recently emphasized how Azure AI Foundry is evolving to support the infrastructure needed for a growing number of AI agents. “All the systems we’ve built for the past 50 years need to apply to AI agents,” Sharma remarked.
However, xAI has historically preferred to keep its model training in-house. Musk had previously walked away from a $10 billion deal with Oracle in 2023, citing plans to build internal infrastructure for training its models. It remains uncertain whether Microsoft would exclusively host Grok or if other cloud providers might also play a role in this arrangement. As of now, Microsoft’s involvement would likely be limited to hosting the model, not training it.
The potential deal has raised some internal questions at Microsoft, particularly regarding Musk’s involvement with the Department of Government Efficiency (DOGE) initiative. Although Musk has announced plans to reduce his role in DOGE this month, any final agreement between Microsoft and xAI could be revealed at Microsoft’s Build conference on May 19.
In addition, the ongoing tension between Microsoft and OpenAI adds another layer of complexity. Microsoft and OpenAI’s relationship has been under scrutiny, especially after OpenAI filed a countersuit against Musk, accusing him of employing delaying tactics. This comes amid growing strains between Musk and OpenAI since his departure from the company he helped establish.
Reports also suggest that Microsoft’s relationship with OpenAI is showing signs of strain. The hiring of Mustafa Suleyman, co-founder of DeepMind, was seen by some as a move to reduce reliance on OpenAI. However, Microsoft’s in-house AI models to compete with OpenAI’s have so far yielded limited success. This leaves Microsoft relying heavily on GPT-based technology, such as the models used in Office and Copilot.
Microsoft had originally anticipated the release of OpenAI’s GPT-5 model this month, but delays in feature rollouts and increased usage have shifted the timeline. As a result, the addition of Grok to Azure would align with Microsoft’s broader goal of diversifying the AI models available on its platform.
Microsoft’s GitHub Copilot, which already supports models from Anthropic and Google alongside OpenAI, is a good example of this strategy in action. If Grok is integrated into Azure, it could set the stage for a future where developers have multiple AI models to choose from, enhancing the flexibility and capabilities of Microsoft’s AI ecosystem.
