Microsoft Build 2026: Solara, Majorana 2 & Scout Revealed

Microsoft Bets on an Agent-First World at Build 2026, Unveils Quantum Leap and Autonomous AI for Every Worker

Project Solara redefines what a device can be. Majorana 2 cuts the quantum computing timeline by four years. Scout becomes every Microsoft 365 user’s autonomous assistant. And Anthropic quietly expanded its secret AI vulnerability-hunter to 200 organisations — including Samsung, NATO, and ENISA.

Microsoft used its Build 2026 developer conference this week to declare the end of the app-centric computing era, announcing three technologies — Project Solara, the Majorana 2 quantum chip, and the Scout AI agent — that together map the company’s vision of an operating model where autonomous AI handles tasks across devices, applications, and scientific research. The announcements came on the same day that Anthropic confirmed it had expanded its restricted Claude Mythos cybersecurity model to approximately 200 organisations in 15 countries, including Samsung, NATO, and the EU’s cybersecurity agency ENISA.

Project Solara: Microsoft Builds an Entirely New Device Category

Project Solara is Microsoft’s answer to a question the company has been asking for several years: what does a device look like when it is designed around AI agents rather than apps? The answer, it turns out, runs on Android — not Windows.

Solara is built on Microsoft’s Device Ecosystem Platform (MDEP), which is based on the Android Open Source Project. Rather than requiring users to open menus, select apps, and navigate interfaces, Solara-powered devices are designed to accept natural language instructions and have AI agents carry out the work across services in the background. Microsoft demonstrated prototype concepts at Build, including a wearable badge and a desktop hub — minimal, screen-light devices built for interaction through voice and context rather than touch and type.

“We are moving from a world where you open an app to get something done, to a world where you describe what you need and an agent figures out how to do it — across every service, every device, seamlessly.”

— Microsoft executive, Microsoft Build 2026 keynote, June 2026

The choice of Android over Windows for Solara is a significant strategic signal. It suggests Microsoft sees Windows as the developer and enterprise workhorse — the “control plane,” as the company put it — while Solara-class devices are purpose-built for ambient, always-on AI interaction in contexts where a full PC operating system would be excessive.

Majorana 2: Microsoft Cuts Quantum Computing Timeline by Four Years

Microsoft also unveiled Majorana 2, the second generation of its topological quantum chip, which it described as the most stable quantum hardware the company has ever produced. The chip’s qubits — the fundamental units of quantum computation — operate with 1,000 times greater stability than the previous generation, with a mean operational lifetime of approximately 20 seconds.

The result is a material change to Microsoft’s quantum computing roadmap. The company’s previous projection for delivering a utility-scale quantum computer — one capable of solving real-world problems beyond the reach of classical supercomputers — was 2033. With Majorana 2, Microsoft says that target has moved to 2029, four years sooner.

Majorana 2 was developed partly with the assistance of Microsoft Discovery, the company’s agentic AI platform for scientific research and development. Microsoft said Discovery helped the quantum hardware team automate complex measurements and process decades of research data to accelerate the chip’s design cycle — an example of AI being used to build the next generation of AI infrastructure.

Microsoft Scout: The Always-On Agent That Works While You Sleep

Microsoft Scout was introduced as the company’s first “Autopilot agent” — a new category of autonomous software that operates continuously in the background rather than waiting to be invoked. Unlike Copilot, which responds to prompts, Scout monitors a user’s Microsoft 365 environment — Teams, Outlook, OneDrive, SharePoint — and takes action proactively.

In practice, Scout handles tasks like preparing briefing documents ahead of meetings, scheduling follow-ups, and drafting responses to routine emails — all without a user needing to issue a command. Each Scout agent operates under a dedicated Microsoft Entra identity, meaning every action it takes is attributed, logged, and subject to the same compliance and governance rules as a human employee in a corporate environment.

Scout is built on OpenClaw, an open-source framework that Microsoft described as the foundation for its broader agentic development ecosystem, giving enterprise developers a reference architecture for building their own custom autonomous agents on top of the same infrastructure.

Anthropic Expands Its Secret Cybersecurity AI to 200 Organisations — Including NATO

On the same day as Microsoft Build’s announcements, Anthropic disclosed a significant expansion of Project Glasswing, its initiative to deploy the restricted Claude Mythos Preview AI model for defensive cybersecurity purposes.

Mythos — which Anthropic has deliberately not released to the public because of its ability to rapidly identify and potentially exploit software vulnerabilities — has been expanded to approximately 200 organisations across 15+ countries, up from around 50 partners when the programme launched in April 2026. New participants are drawn from sectors considered critical infrastructure: power grids, water utilities, healthcare, telecommunications, and hardware manufacturing.

Confirmed participants include Samsung, SK Hynix, SK Telecom, Okta, Euroclear, the Intercontinental Exchange (ICE), NATO, and the EU’s cybersecurity agency ENISA, according to reporting by the Financial Times and The Hindu. India was among the newly added countries in the June 2026 expansion.

“For most of these partners, a successful cyberattack on their systems could impact more than 100 million people. Mythos gives defenders a tool that is as capable as anything an attacker could build — but deployed defensively, at scale, before the attack happens.”

— Anthropic, official Project Glasswing statement, June 2, 2026

Since the programme’s April launch, Mythos has helped participating organisations identify more than 10,000 high- or critical-severity security flaws in their codebases and infrastructure, according to Anthropic. All organisations accessing the model must meet strict security requirements set by Anthropic, and the company has updated its disclosure policies to allow partners to share findings with regulators and industry groups under responsible-disclosure guidelines.

What to Watch Next

Apple WWDC 2026 opens on June 8, with industry analysts widely expecting a major overhaul of Siri — internally referred to as “Siri 2.0” — that would put Apple directly in competition with Microsoft’s Copilot and Scout agents for ambient AI on consumer devices. Apple has not confirmed specifics ahead of the event.

Microsoft’s Project Solara prototype devices are in early concept stages. No commercial release date has been announced. The company said it would share Solara’s MDEP specifications with hardware partners and developers in the coming months.

The Majorana 2 chip is not yet available outside Microsoft’s own quantum research labs. The company’s 2029 utility-scale target represents an internal projection and is subject to ongoing engineering milestones.

Microsoft Scout is expected to roll out to Microsoft 365 enterprise subscribers in phases beginning in the second half of 2026. Availability for small and medium businesses has not been confirmed.

Google’s own June 2026 Android feature drop — which includes AI-powered fake call detection, Circle to Search upgrades, and a digital wardrobe feature — is rolling out in parallel this week, setting up a busy fortnight of platform competition ahead of WWDC.