Microsoft on Monday announced a partnership with Elon Musk’s AI startup xAI, bringing its controversial Grok models to the company’s cloud platform Azure.
The news came during the kickoff of Microsoft’s Build developer conference, where both companies presented the deal as a step forward in expanding access to advanced AI models, despite rising criticism over Grok’s recent behavior.
The partnership gives xAI’s Grok models a new distribution channel, while also bolstering Microsoft’s growing portfolio of hosted AI systems.
Grok joins Azure AI Foundry with stricter controls
Microsoft now offers managed access to Grok 3 and Grok 3 mini through its Azure AI Foundry platform. These models come with enterprise-grade features, including service-level agreements and direct billing through Microsoft.
Originally pitched by Musk as an unfiltered, anti-woke AI, Grok has built a reputation for pushing the boundaries of acceptable responses.
It willingly responds to prompts that other models typically reject. According to benchmarks like SpeechMap, Grok 3 is among the more permissive models.
However, the Grok versions hosted on Azure are more tightly controlled than those on X, Musk’s social network.
Microsoft’s offerings include enhanced governance, customization, and integration features designed to support secure enterprise use.
Microsoft cements lead in cloud-based AI infrastructure
This move strengthens Microsoft’s position in the cloud AI market. Azure users can now access over 1,900 AI models, including systems from OpenAI, Meta, and DeepSeek. Models from Google and Anthropic are not yet available on the platform.
The Build conference also highlighted new tools for managing AI agents—systems that can act on behalf of users. These include a model leaderboard and smart model selection features for developers.
“To make AI agents truly effective, they need the ability to connect with everything in the world,” said Microsoft CTO Kevin Scott.
Backlash grows over Grok’s controversial responses
The collaboration follows several high-profile missteps by xAI. Just days ago, its models sparked outrage by referencing “white genocide” in the context of South Africa.
The incident drew sharp criticism across social media and the tech community.
Addressing the controversy, Musk acknowledged the errors but defended the company’s mission. “Our models aspire to truth with minimal error,” he said. “Of course, there’s always going to be some mistakes that are made, but we are working to improve rapidly.”
Microsoft 365 Copilot adds deep research capabilities
Microsoft also introduced major updates to its 365 Copilot product line. The new tools, Researcher and Analyst, focus on in-depth analysis and insight generation. They comply with enterprise-grade data policies and privacy standards.
Researcher can generate detailed strategies and reports using both internal and external data. Analyst operates more like a data scientist, refining its output through chain-of-thought reasoning.
These updates mark a significant evolution in Microsoft’s business-focused AI tools and could set new standards for enterprise productivity.