NeuralStackMicrosoft used its Build 2026 developer conference in San Francisco to make a clear statement: it wants to compete more directly with OpenAI and Anthropic, not just distribute their models. The company unveiled five major AI announcements spanning devices, hardware, agents, models, and healthcare. Together, they point to a strategy built around owning more of the AI experience from the device layer to the enterprise workflow.
The most visible announcement was Project Solara, a new family of AI-focused prototype devices. Microsoft said these devices are designed around AI agents that communicate with cloud systems to carry out specific tasks, rather than relying mainly on traditional app-based interactions. The company said the concept could take several forms, including smart speaker-sized devices and badge-like wearables.
Microsoft also introduced the Surface RTX Spark Dev Box, a desktop powered by Nvidia’s RTX Spark chip. Satya Nadella described it as a “dream machine” during his keynote, and Microsoft demonstrated the system running a massive AI model locally. The message was that next-generation PCs may soon handle advanced AI workloads that have largely been confined to data centers or high-end systems.
Another major update was Scout, an AI agent coming to Copilot. Microsoft said Scout will gather emails, messages, and other items needing attention to help users make decisions faster. The company positioned it as a more active assistant that can do work on a user’s behalf, sharpening the competition with other agent-style AI products.
Microsoft also introduced MAI Thinking-1, its first AI reading-focused model, and said it is meant to compete with frontier systems from the industry’s biggest players. The company said the model can deliver performance comparable to some of the most advanced reasoning systems available. It also revealed new transcription and image-generation models as part of the broader push.
In healthcare, Microsoft announced a new collaboration with Mayo Clinic to build advanced AI systems. The companies plan to combine Microsoft’s AI infrastructure and reasoning capabilities with Mayo Clinic’s medical expertise and data. The goal is to create tools that can support clinicians, improve diagnosis, and help medical teams make faster decisions.
Why this matters
Microsoft is no longer presenting itself simply as a platform where other companies’ AI models live. By building its own models, agents, devices, and specialized hardware, it is trying to control more of the stack and reduce dependence on outside labs. That shifts the competitive pressure from software distribution to full-system ownership.
The timing also matters because OpenAI and Anthropic are expanding aggressively across enterprise use cases, where Microsoft already has deep customer relationships through Copilot, Windows, and cloud infrastructure. If Microsoft can make its own tools fast, useful, and cost-effective, it could keep more AI activity inside its ecosystem instead of routing it to competitors. The Build announcements suggest that is now the company’s explicit goal.
Microsoft’s broader message is that enterprise AI is moving beyond chat interfaces and into systems that can act, reason, and operate across devices. The five announcements show a company trying to make that shift look native to Microsoft products rather than borrowed from rival labs. What happens next will depend on how quickly these prototypes, models, and partnerships turn into products customers actually use.