On June 2-3, 2026, Microsoft held its annual Build conference at Fort Mason, San Francisco. Satya Nadella's opening keynote message was clear: the new developer stack is no longer just about tools and APIs. It's about building a complete environment where AI agents can actually do real work autonomously.
For enterprises, this edition marks a strategic turning point. Microsoft is no longer talking about AI as an occasional assistance tool — it's a full platform, from silicon to cloud, designed for autonomous agents. Here's our breakdown of the major announcements and their concrete impact for IT leadership.
7
new MAI models announced
1 PFLOP
AI compute in Surface RTX Spark
120B
parameters runnable locally
1. The Age of Agentic AI: Microsoft's Strategic Shift
The headline announcement at Build 2026 is undoubtedly the shift to agentic AI as the central paradigm. Microsoft is no longer just offering copilots that assist — they're introducing autonomous agents capable of chaining tools, data sources and actions to complete complex end-to-end tasks.
Microsoft Agent Framework: AutoGen + Semantic Kernel Convergence
The most structurally significant developer announcement is the production release of the Microsoft Agent Framework, which merges AutoGen and Semantic Kernel into a single, commercially supported SDK for building multi-agent systems. In practical terms, enterprises now have a unified framework to develop, deploy and supervise AI agents in production, with support and security guarantees.
Microsoft Scout: Autopilots Are Here
Microsoft unveiled a new category of agents called Autopilots — always-on, autonomous agents with their own identity that act on your behalf. The first one, Scout, continuously monitors your inbox and Teams sessions to identify actions you need to take, prioritise them and execute when you're ready.
Enterprise impact: Autopilots represent a paradigm shift for CIOs. Autonomous agents with their own Entra ID identity raise fundamental questions around governance, security and compliance. Enterprises that anticipate these challenges will be best positioned to benefit.
Microsoft Execution Containers (MXC): Agent Security
To address security concerns around autonomous agents, Microsoft introduces the MXC SDK — a policy-driven execution layer. Developers can declare what an agent can access (files, network, APIs) with containment boundaries enforced at runtime. This approach offers a spectrum of isolation semantics, dynamically composable based on intent and risk.
2. Copilot Platform: AI Embedded in Every Application
Microsoft launches Copilot Platform — a set of APIs, connectors and a runtime that allows any application to embed Copilot agents. This announcement transforms Copilot from a Microsoft product into an open platform that third-party vendors can integrate into their own solutions.
Microsoft IQ: The Unified Intelligence Layer
At the heart of this platform, Microsoft IQ is the unified intelligence layer that powers the entire ecosystem. It comes in four complementary components:
Work IQ brings workplace intelligence inside the Microsoft 365 trust boundary. Foundry IQ manages enterprise knowledge for agents via Azure AI Foundry. Fabric IQ Ontology provides business semantics. And Web IQ delivers real-time web grounding APIs powering Microsoft Copilot and ChatGPT.
GitHub Copilot App: Next-Generation Developer Tool
The native GitHub Copilot app (Windows, macOS, Linux) is arguably the announcement that will most impact developers' daily workflows. It offers three working modes — Interactive, Plan and Autopilot — with parallel sessions via git worktrees, Agent Merge and mobile handoff via GitHub Mobile. Available with no waitlist for all Pro, Pro+, Max, Business and Enterprise subscribers.
3. New MAI Models: Microsoft's In-House AI
Microsoft announced seven new MAI models, signalling its ambition to offer its own foundation models:
MAI-Thinking-1 is Microsoft's first dedicated reasoning model, designed for tasks requiring deep reflection and complex reasoning chains.
MAI-Code-1 is a coding model specifically trained for GitHub and VS Code, already integrated into Copilot.
MAI-Image-2.5 delivers 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+ additional languages.
Phi-4: Compact Open-Source Models
Microsoft releases Phi-4-Medium and Phi-4-Vision under MIT licence (open weights), joining Llama 4 and Mistral in the Azure AI model catalogue. Phi-4-Silicon, a 3.8 billion parameter model optimised for NPUs, ships with the Windows 2026 Update and can handle tasks like email summarisation, document formatting and schedule negotiation entirely on-device, without sending data to the cloud.
Data sovereignty: the availability of performant local models (Phi-4-Silicon) is a game-changer for European enterprises subject to data sovereignty constraints. Running AI models locally on workstations opens new use cases without compromising confidentiality.
4. Windows: Optimised for Developers
Coreutils: Native Linux Utilities on Windows
Microsoft brings familiar Linux command-line utilities directly to Windows with Coreutils. Developers get the tools they already know (ls, grep, cat, sed, awk, etc.) natively in Windows Terminal, without WSL.
Windows 365 Developer Configuration
Available in public preview, Windows 365 Developer Configuration offers a "ready-to-code in minutes" Cloud PC, pre-configured with VS Code, GitHub Copilot, WSL2 + Ubuntu, PowerShell 7, Git and GitHub CLI. Developers can spin up a complete development environment, streamed from anywhere.
Intelligent Terminal: AI-Augmented Terminal
Microsoft introduces an experimental intelligent terminal that understands the context of your commands, suggests corrections, explains errors and can execute complex workflows on demand. It's the convergence of traditional terminal experience and contextual AI assistance.
5. Hardware: Surface RTX Spark Dev Box and Project Solara
Surface RTX Spark Dev Box: The Compact AI Workstation
The Surface RTX Spark Dev Box is purpose-built for sustained developer and AI workloads. Its specifications are impressive: 1 petaflop of AI compute, 128 GB of unified memory, ability to run models up to 120 billion parameters locally, all within a 100W thermal envelope. It ships with custom-tuned Windows 11 Pro, WSL2, native GPU passthrough, CUDA support, VS Code and GitHub Copilot pre-installed.
Project Solara: The Hardware Platform for AI Agents
Arguably the most visionary announcement at Build 2026 is Project Solara — a chip-to-cloud platform designed from the ground up for devices that run AI agents instead of traditional applications. Microsoft is betting that AI will open entirely new computing scenarios, freed from the constraints of classical software.
Two reference designs were shown, both targeting enterprise workers:
The wearable badge reimagines the standard employee ID card. A fingerprint button wakes an agent in one press. A single tap records and transcribes a conversation. A built-in camera lets the agent act on what the user sees. The badge runs on a new Qualcomm wearable chip.
The desk companion is a desktop hub that sits beside a PC and responds to voice commands, signs users in via facial recognition and surfaces the day's most pressing items. With a monitor attached, it becomes a full Windows machine running in the cloud. The hub runs on MediaTek IoT silicon.
Android, not Windows: notably, the operating system is MDEP (Microsoft Device Ecosystem Platform), an enterprise version of Android that Microsoft developed for Teams Rooms hardware. Microsoft chose MDEP over Windows deliberately, to run on smaller, lower-power devices while keeping the management and security features IT departments expect: OTA updates, device integrity, Microsoft Defender, Intune and Entra ID sign-in.
Several major companies — AccuWeather, Best Buy, CVS Health, Levi's and Target — are expected to begin pilots in the coming months based on these reference designs. This is a strong signal: Project Solara is not a distant concept but a platform Microsoft is already pushing into the field.
For CIOs, Project Solara raises a strategic question: the AI agent as the primary work interface, beyond the screen and keyboard. Enterprises that anticipate this shift — by preparing their Entra ID infrastructure, Intune policies and security frameworks — will be first to benefit.
6. Developer Tools: Visual Studio 2026 and Agent Debugging
Visual Studio 2026 and VS Code both get deep agent debugging experiences. A new agent profiler shows a timeline of reasoning steps, API calls and latency. Developers can set breakpoints on agent thought processes and replay agent runs deterministically.
This is a major advance for multi-agent system reliability in production: being able to inspect, understand and reproduce an agent's behaviour is essential for enterprise-scale deployments.
7. Quantum: Majorana 2 and the 2029 Horizon
Finally, Microsoft announced Majorana 2, its new quantum processor. Qubits are over 1,000 times more reliable than the previous generation, with lifetimes jumping from milliseconds to an average of 20 seconds (occasionally crossing a minute). Thanks to this progress, Microsoft is cutting its timeline for a scalable quantum computer in half, with 2029 as the new target.
8. What This Means for Enterprises
Build 2026 is not just a technical conference — it's a strategic roadmap for the next 3 to 5 years. Here are the concrete implications for IT leaders:
AI agent governance — Autonomous agents with Entra ID identity require rethinking governance, access and compliance models. Who is accountable when an agent makes a decision?
Data sovereignty — Phi-4-Silicon and local models pave the way for on-device AI for sensitive use cases, without sending data to the cloud.
Workstation evolution — Coreutils, Intelligent Terminal and Windows 365 Developer Configuration bring Windows closer to the Linux experience developers demand, while maintaining enterprise management via Intune and Entra ID.
Multi-model strategy — The Azure AI catalogue with MAI, Phi-4, Llama 4 and Mistral models gives enterprises the choice of the right model for each use case, without vendor lock-in.
Quantum readiness — The 2029 horizon for commercial quantum should appear on medium-term transformation roadmaps, particularly for post-quantum cryptography.
EMAZIS can help: as a consulting firm specialised in digital transformation, we help enterprises assess the impact of these announcements on their IT strategy and define a pragmatic roadmap to leverage the new capabilities of the Microsoft platform. Let's discuss your challenges.
