Microsoft walks a tightrope: Adds Elon Musk’s Grok to Azure AI Foundry alongside OpenAI

In a move that could redefine its AI alliances, Microsoft has added Elon Musk’s Grok to Azure AI Foundry—making the controversial xAI model available to developers and enterprises through the same platform that powers OpenAI’s ChatGPT.

This is the first time a major cloud provider has partnered with xAI, the company Musk founded in direct opposition to OpenAI. The Grok models, which include the latest Grok-3 and its smaller variant, are now live on Azure, with Microsoft managing billing and infrastructure. Grok joins other third-party models from Meta, Mistral, and Cohere in Azure’s growing multi-model ecosystem.

A tense expansion

Microsoft’s decision to support Grok comes despite its deep financial and strategic ties to OpenAI, in which it has invested over $14 billion. While Microsoft has repeatedly positioned Azure as an open platform for the best-performing models, onboarding Grok sends a clear signal: no model is off limits, not even one that routinely mocks OpenAI and challenges its foundational principles.

Grok has built its identity as an unfiltered, anti-woke alternative to ChatGPT, with Musk explicitly stating it will answer questions other models refuse. That posture has drawn criticism for enabling misinformation and unsafe content. Now, Microsoft is attempting to tame Grok’s raw output by wrapping it in Azure’s enterprise-grade governance and safety systems.

A sanitised Grok for the enterprise

The Grok models deployed through Azure are significantly more locked down than the ones available via X (formerly Twitter). Microsoft says users will benefit from observability, compliance tools, safety filters, and integrated monitoring, making Grok viable for enterprise use.

This sanitised version of Grok still carries impressive technical credentials. Trained on xAI’s Colossus supercomputer, Grok-3 promises advanced reasoning, extended context lengths, and strong performance across benchmarks. Microsoft is offering it through both standard API pricing and soon, through provisioned capacity for production workloads.

The addition of Grok highlights Microsoft’s broader AI strategy: own the infrastructure layer and host every top-tier model, regardless of its origin. It’s a bold bet on model pluralism, where customers pick the right tool for the job, and Microsoft supplies the rails. This strategy has helped Azure rapidly expand its catalogue of foundation models, and CEO Satya Nadella has been personally pushing to onboard cutting-edge systems like DeepSeek and now Grok.

But it also raises questions about Microsoft’s balancing act. Grok isn’t just another model—it’s a direct competitor to OpenAI, often hostile in tone and content. By supporting both, Microsoft risks complicating its partnership with the very company that powers its flagship AI products like Copilot and Bing Chat.

Microsoft is signalling that it wants to be the platform for all frontier models, regardless of ideology or rivalry. That may make Azure indispensable to developers. But it also tests how far Microsoft can stretch its neutrality without fracturing its alliances.