Microsoft is betting the future of computing will revolve around artificial intelligence agents rather than traditional apps and operating systems. At the Microsoft Build 2026 developer conference in San Francisco, CEO Satya Nadella declared that the technology industry is entering a major platform transition where AI agents become the primary way humans interact with computers. This represents a fundamental reimagining of how software works, moving away from users manually opening and switching between applications toward intelligent systems that handle complex tasks autonomously across multiple services and devices.

What Is Project Solara and Why Does It Matter?

Microsoft introduced Project Solara, a chip-to-cloud platform developed jointly with Qualcomm, as the technological foundation for this "agent-first computing" vision. The initiative includes a family of prototype devices ranging from smart speaker-sized gadgets to keycard badge-sized devices, built on chips from Qualcomm and MediaTek. Unlike traditional smartphones or computers, these devices will not run conventional operating systems and apps. Instead, they will host AI agents that communicate with cloud-computing systems to carry out specific tasks, such as documenting a medical visit with a nurse.

"There's a real platform shift. We're moving from building operating systems and devices for apps to agents," said Satya Nadella, CEO at Microsoft.

Satya Nadella, CEO at Microsoft

Nadella emphasized the flexibility this approach offers developers and enterprises. During his keynote address, he explained that with Project Solara, "you, as developers and enterprises, have the flexibility to imagine the form factors that you want and have your agents be ubiquitous". This means developers can design AI experiences for whatever device shape or size makes sense for their use case, without being constrained by traditional app-based interfaces.

Nadella

How Is Microsoft Building Its Own AI Models Independent of OpenAI?

A critical part of Microsoft's strategy involves developing frontier AI models in-house, reducing its dependence on OpenAI, the company it has long invested in and partnered with. During Build 2026, Microsoft's AI unit announced several new models designed to compete directly with offerings from rivals like Anthropic and OpenAI.

• MAI Thinking-1 Reasoning Model: Microsoft released its first reasoning model, which matched the performance of Anthropic's Claude Opus 4.6 launched earlier this year. This demonstrates Microsoft's ability to build advanced reasoning capabilities internally.
• Transcription AI Model: The company unveiled what it claims is the most efficient transcription AI model offered by any major cloud provider, improving on existing solutions.
• Image Generation Model: Microsoft introduced a new image model designed to compete with Google's offerings in visual AI capabilities.

These models underscore Microsoft's broader effort to control more of the end-to-end AI system, known as the "stack," and lock in enterprise customers as competition intensifies. The company is also developing tools to help Windows run OpenClaw, open-source software that can direct groups of AI agents to carry out everyday tasks. Microsoft is focusing on making OpenClaw safe for businesses to use on computers containing sensitive corporate data.

What New Hardware Is Microsoft Showcasing?

Microsoft revealed several new computing devices designed to showcase its AI-first vision. The company introduced the Surface RTX Spark Dev Box, a computer loaded with an Nvidia chip that Nadella called a "dream machine" and said he was on the wait list to buy. This device can run AI models with 120 billion parameters, a rough measure of a model's complexity, which most standard PCs cannot handle.

The Surface RTX Spark Dev Box follows a laptop Microsoft introduced with Nvidia earlier in the week. These new wave of PCs are priced to compete with Apple's premium offerings, though analysts noted it may take time for businesses to adopt the new machines. Microsoft is also competing against rivals to sell cloud-based AI tools for coding and other tasks while nudging customers toward running AI technologies directly on Windows-based laptops and desktops.

How Will AI Agents Actually Work Across Different Devices?

Microsoft outlined a multi-stage vision for how AI will evolve in computing. The company described three distinct stages of AI adoption that show how the technology will gradually reshape user experiences.

• Stage One - AI as Companion: AI acts as an assistant alongside existing apps, similar to today's chatbots and Copilot features that users interact with within traditional software.
• Stage Two - AI Integrated into Apps: AI becomes deeply embedded within applications, making it central to how users experience the software rather than a separate tool.
• Stage Three - Agent-First Computing: AI operates outside individual apps entirely, enabling agents to coordinate actions across platforms, devices, and workflows independently without users manually switching between applications.

Project Solara is specifically designed to support this third stage. One notable feature is "just-in-time UI," which allows AI agents to dynamically generate user interfaces based on the device, screen size, and interaction method, whether voice, touch, or visual inputs. This approach could reduce the complexity developers face when building software for different hardware categories.

Microsoft also stressed that the future will not depend on a single universal assistant. Instead, the company expects a multi-agent ecosystem where specialized AI agents collaborate to complete different tasks. To support this, Solara includes technologies such as an agent dispatcher and an agent task manager that determine which AI system is best suited for a particular action.

What Does This Mean for Enterprise Customers?

Microsoft introduced a new AI agent called Scout within its Copilot software that can carry out tasks such as gathering emails or messages that require decisions by the user to move forward. This demonstrates how the company is bringing agent capabilities directly into tools that enterprise customers already use.

The company is also pursuing healthcare partnerships to showcase AI's potential in enterprise settings. Microsoft announced a deal with the Mayo Clinic to build frontier healthcare AI, drawing on Microsoft's reasoning and compute capabilities combined with Mayo Clinic's clinical expertise and data. The partnership emerged from meetings between Mayo Clinic CEO Gianrico Farrugia and Nadella, with the goal of improving patient outcomes with AI that acts as a team member and helps reach "a diagnosis faster and better".

Microsoft

"You can totally run OpenClaw inside your company now," said Peter Steinberger, the software engineer who created OpenClaw.

Peter Steinberger, Software Engineer

By pairing AI agents with new devices, powerful PCs, and its own models, Microsoft is attempting to control more of the complete AI system and lock in enterprise customers as competition from rivals OpenAI and Anthropic intensifies. This strategy represents a significant shift in how the company approaches the market, moving beyond its traditional reliance on partnerships to building more of its AI capabilities in-house.