Microsoft Build 2026 Highlights: From MAI-Code-1 to Project Solara, Here's What Microsoft Announced

Microsoft held Build 2026 on June 2–3 at Fort Mason in San Francisco. CEO Satya Nadella opened the keynote with a clear frame: the new developer stack isn't just tools and APIs. It's about building a full environment where AI agents can actually do real work. Everything announced flows from that idea.

Here are the highlights worth caring about.

Microsoft Finally Has Its Own AI Models

This one's significant. For years, Microsoft's AI story was basically "we have OpenAI." Build 2026 shifted that.

Microsoft announced a family of in-house MAI models across several categories. MAI-Thinking-1 is the company's first dedicated reasoning model. MAI-Code-1 is a coding model built specifically for GitHub, and it's already live in Copilot and VS Code. MAI-Image-2.5 brings what Microsoft says is a step-change in image editing quality. MAI-Transcribe-1.5 handles speech-to-text across 43 languages, and MAI-Voice-2 adds new voices in 15-plus additional languages.

These models are also going beyond Microsoft's own platforms. They'll be available on Fireworks AI, Baseten, and Open Router. Microsoft is clearly trying to build its own model credibility, separate from the OpenAI relationship.

GitHub Copilot Is Growing Up Fast

The GitHub Copilot desktop app is now in preview, and it's more than just a sidebar in your editor. It's designed to handle full agentic workflows natively, so you can manage, monitor, and direct coding tasks from a standalone app.

Also in preview is Project Rayfin, which tries to close the gap between a working prototype and something actually deployable. It gives developers a managed backend-as-a-service built on Microsoft Fabric, so the jump from demo to production doesn't require rebuilding everything from scratch.

Windows Is Getting Serious About Developers

Microsoft announced a few things here that are genuinely useful day-to-day.

Windows Development Configurations is now generally available. You get a single config file that takes a fresh machine to a ready-to-code setup in minutes, pre-loaded with WSL, PowerShell 7, Git, GitHub CLI, VS Code, Python, and more. Anyone who's spent half a day reimaging a machine will appreciate this.

WSL containers lets developers run Linux containers directly on Windows without a separate VM, using a new command-line tool called wslc.exe. And Intelligent Terminal, currently in experimental preview, is a version of Windows Terminal with native agent integration. It reads your live shell state including command history, current directory, exit codes, and Git context, and can suggest fixes when a command fails.

Microsoft IQ: Context That Actually Matters for Agents

Microsoft IQ is now generally available across GitHub Copilot, Foundry, and Copilot Studio. It's a context layer with a few distinct parts: Work IQ pulls in real workplace knowledge from Microsoft 365 including emails, documents, meetings, and org relationships. Fabric IQ grounds agents in structured business data. And the newly announced Web IQ provides fast, live web grounding.

The Work IQ APIs open up on June 16, giving developers programmatic access to that organizational knowledge layer. This is what makes an agent feel like it actually knows your company, not just the internet.

Project Solara: Agents Beyond the Laptop

This is the conceptual one, but it's worth paying attention to. Microsoft showed two early reference designs for what it's calling Project Solara, a chip-to-cloud platform for agent hardware.

One is a badge-style wearable device, powered by Qualcomm silicon, designed for on-the-go agent interaction when you're away from a desk. The other is a desk companion powered by a MediaTek SoC, always on, always aware of your context. Neither is shipping yet, but the idea is clear: Microsoft wants agents to exist in more places than just inside an app window.

The NVIDIA Partnership and the Hardware Behind It All

Jensen Huang joined Satya on stage to announce a unified computing stack targeting agentic AI, anchored by new Windows PCs powered by NVIDIA RTX Spark. The Surface RTX Spark Dev Box is a physical machine aimed at developers who need serious local AI compute.

On the cloud side, Microsoft's second-gen AI accelerator Maia 200 is already running in data centers in Iowa and Arizona, with Italy, Australia, and South Korea coming next. Cobalt 200-based virtual machines are now in preview across more than 10 global regions.

Majorana 2: Microsoft's Quantum Bet Gets Real

Microsoft announced Majorana 2 at Build 2026, and the numbers are hard to ignore. The new quantum processor uses qubits that are over 1,000 times more reliable than its predecessor, with lifetimes jumping from milliseconds to an average of 20 seconds, occasionally crossing a minute.

The upgrade comes from swapping aluminum for lead in the material stack, which more than doubles the topological gap that protects qubits from errors. Because of this progress, Microsoft has cut its timeline for a scalable quantum computer in half, with 2029 now the target.

Source