Microsoft to unveil homegrown AI models at Build, including coding tool

Microsoft is preparing to use its Build developer conference in San Francisco to show off a new batch of homegrown AI models, including a coding tool aimed at sharpening GitHub Copilot. The June 2-3 event, which will run in San Francisco and online, has become one of the clearest signs that Microsoft now sees its developer conference as a strategic AI showcase, not just a software gathering.

The planned models go beyond coding. Microsoft also intends to present systems focused on transcription, reasoning, speech and image tasks, a broader lineup that points to a company building more of its AI capability in-house rather than depending entirely on outside partners. For developers, the most immediate change would be inside everyday workflow: a stronger coding model could mean more accurate autocomplete, better help across large codebases, fewer context switches and faster debugging inside the tools developers already use.

That shift matters because GitHub Copilot is no niche product. Microsoft said in its 2025 annual report that Copilot now has more than 20 million users, giving the company a large base to improve and monetize. A better model under the hood could lift product quality, but it could also make Microsoft more able to control pricing, packaging and the pace of new features across GitHub and the broader developer stack.

The timing also fits a wider reset in Microsoft’s relationship with OpenAI. On April 27, Microsoft said it would keep a non-exclusive license to OpenAI intellectual property through 2032, stop paying revenue share to OpenAI and remain OpenAI’s primary cloud partner. OpenAI said it can serve customers across any cloud provider under the revised terms. Together, the changes suggest Microsoft wants more room to steer its own model supply, integrate products more tightly and reduce the risk of relying too heavily on a single partner.

Investors have taken the message seriously. Microsoft shares rose nearly 3 percent on the news, and market data on May 29 showed the stock up about 3 percent intraday. That reaction reflects how closely Wall Street is watching Microsoft’s AI roadmap as the company tries to turn its product reach into an advantage over rivals.

The company’s scale gives it room to make that bet. Microsoft reported revenue of $81.3 billion for the quarter ended December 31, 2025, and chief executive Satya Nadella said in the January earnings release that the company was only at the beginning phases of AI diffusion. Build now looks like the stage where that diffusion becomes more concrete: more in-house models, more control over the stack and a clearer attempt to close any capability gap while deepening lock-in across Microsoft’s ecosystem.