Microsoft will use its Build developer conference in San Francisco on June 2 to unveil MAI-Thinking-1, its first in-house reasoning model, alongside a Copilot 'super app,' a developer-optimized Windows 11 experience, and a broader push for local AI models running on Windows. The keynote starts at 9:30 AM PT / 12:30 PM ET, with Microsoft CEO Satya Nadella sharing the stage with Nvidia CEO Jensen Huang to discuss the RTX Spark PC class. The company has shrunk Build into a smaller venue this year, a deliberate pivot toward rebuilding developer trust as it reshuffles around AI.
Microsoft AI chief Mustafa Suleyman is expected to lead the model announcements. MAI-Thinking-1 is notable for what it isn't: Microsoft did not use distillation, meaning the reasoning model was not trained on another AI model's outputs. That positions the company to pitch enterprise customers a reasoning system whose lineage is fully its own, separate from its long-running dependence on OpenAI weights. Suleyman is also expected to detail MAI-Image-2.5 and MAI-Image-2.5-Flash, which he teased last week.
For Build's actual audience, the bigger story is Windows. Microsoft will unveil a Windows 11 developer-optimized experience with pre-installed apps, tools, and scripts in a distraction-free environment. Windows chief Pavan Davuluri teased the announcement last week, and the company will also detail more of its ongoing effort to rewrite parts of Windows 11 for performance — a plan it outlined earlier this year and has begun shipping in pieces.
“something new is coming for developers”— Pavan Davuluri, Windows chief
Key facts
- 01Microsoft AI chief Mustafa Suleyman is expected to unveil MAI-Thinking-1, the company's first reasoning model, built without distillation from another model's outputs.
- 02The Build keynote kicks off June 2 at 9:30 AM PT / 12:30 PM ET in San Francisco.
- 03Microsoft is also expected to detail MAI-Image-2.5 and MAI-Image-2.5-Flash, plus a Copilot 'super app' bundling its assistants into one interface.
- 04A new Windows 11 developer-optimized experience will ship with pre-installed dev tools, scripts, and a distraction-free environment.
- 05Satya Nadella will appear with Nvidia CEO Jensen Huang to discuss RTX Spark, with a bigger Build focus on local models running on Windows.
Local AI on Windows gets a dedicated push. Microsoft wants developers running smaller models on-device instead of paying for cloud inference, and the hardware story now extends beyond the Qualcomm and AMD/Intel split. Nvidia's RTX Spark — which the company announced as a PC class built for local AI agents — brings a new Arm-based silicon partner into Windows. Miniature RTX Spark PCs from Microsoft and HP were notably absent from Nvidia's Computex lineup, suggesting Build is the intended reveal window.
Qualcomm is expected to feature as well, alongside continued work on Windows on Arm. Qualcomm and Microsoft laid the groundwork for Arm improvements in Windows 11, which is what allowed Nvidia to return to the platform after the troubled Surface RT era. Microsoft now has to balance two Arm silicon providers in parallel with its long-standing x86 split between AMD and Intel.
The Copilot super app is the consumer-facing piece. It bundles Microsoft's various Copilot AI assistants into a single interface, and an early leaked screenshot from Friday turned out to be a mockup prepared for Build demonstrations. The app also includes a first look at Microsoft Scout, a new agent reportedly based on Microsoft's OpenClaw work. The super app will not ship at Build and is unlikely to enter preview before late summer.
GitHub is the unresolved tension running through the week. Microsoft's developer platform has faced departures, outages, and security incidents, and Build is being driven in part by the GitHub team. The conference is the company's most visible chance to address those issues with the audience that matters most.
“Microsoft desperately needs to win back GitHub trust here, particularly as high-profile developers are starting to sound the alarm.”— Tom Warren, Senior Correspondent, The Verge
The stakes for Microsoft at this Build are unusually concrete. The company is trying to convert its AI investment into a coherent product story across three layers — models (MAI-Thinking-1, MAI-Image-2.5), an interface (the Copilot super app), and a runtime (Windows plus RTX Spark, Arm, and x86 silicon). Each layer has been announced piecemeal over the past year; Build is the first event where Microsoft can present them as a single stack.
The risk is execution timing. MAI-Thinking-1 will need to clear enterprise benchmarks against established reasoning models from OpenAI, Anthropic, and Google to justify the in-house investment, and Microsoft has not yet published evaluation numbers. The Copilot super app remaining months from preview means developers will leave Build with a roadmap rather than a product, and GitHub's reliability problems do not have a quick fix the keynote can ship.
For the AI market, the read is that Microsoft is finally building a model lineup it controls outright, rather than wrapping OpenAI APIs in a Copilot brand. MAI-Thinking-1 trained without distillation is the most strategically meaningful piece of the week — it gives Microsoft optionality on pricing, fine-tuning, and enterprise deployment that a pure OpenAI reseller does not have. If the model holds up on enterprise workloads, the Copilot stack becomes considerably harder to dislodge, and Microsoft's negotiating position with every model vendor — OpenAI included — shifts.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
