Microsoft CEO Satya Nadella announced new AI models, Windows updates, and a quantum computer preview during the Build 2026 conference in San Francisco on June 2, 2026 [1, 2, 3].

These updates signal a shift toward "agentic" AI and quantum computing, potentially altering how developers build software and how users interact with the Windows operating system.

During the event, Nadella introduced the native OpenClaw app for Windows and Unmetered Agentic AI models [1]. The company also provided a preview of the Mojorana 2 quantum computer [1]. While some reports focused on general Copilot features, these specific hardware and software additions represent a push toward more autonomous system capabilities.

On the software side, Microsoft announced two new local AI Windows models [4]. These models are designed to run directly on device hardware, reducing reliance on cloud connectivity for certain AI tasks. This move complements ongoing updates to Copilot features and broader Windows integration.

The keynote address was expected to last 2.5 hours [3]. The event served as the primary venue for the company to showcase its latest innovations to the global developer community.

In a post regarding the event, the Microsoft Blog team said, "We explore, choose tools, dream, create" [5]. The conference focused on the theme of allowing users to be themselves at work while utilizing these new tools to increase productivity.

The San Francisco event highlighted a diversified strategy, moving beyond large language models into specialized agentic AI, and experimental quantum hardware [1, 2].

“"We explore, choose tools, dream, create."”

The introduction of agentic AI and local models suggests Microsoft is attempting to move AI from a reactive chatbot interface to a proactive system capable of executing complex tasks independently. Simultaneously, the preview of the Mojorana 2 quantum computer indicates that Microsoft is positioning itself for the post-classical computing era, aiming to maintain a competitive edge in high-performance computing long before such technology reaches mainstream commercial viability.