Nvidia & Microsoft unveil AI agents for Windows PCs

Nvidia and Microsoft have outlined a joint effort to bring AI agents to Windows PCs, unveiling new hardware and software for local AI use on desktop and laptop systems.

The announcements came at a Microsoft event in San Francisco featuring Nvidia Founder and Chief Executive Officer Jensen Huang and Microsoft Chief Executive Officer Satya Nadella. They described a plan combining changes to Windows with Nvidia chips and systems designed to run AI workloads directly on personal computers.

Microsoft is adapting Windows to support agents that run safely in the background under operating system control. It also announced the general availability of Microsoft Execution Containers, or MXC, which it described as operating-system-level infrastructure to secure, observe and govern those agents.

Pavan Davuluri, Executive Vice President of Windows and Devices at Microsoft, outlined that approach during the event. "We are building these primitives directly into Windows, so agents can be secured, observed and governed," Davuluri said. "With MXC, Microsoft Security, and Agent 365, we're unlocking the full potential of agents on Windows."

Nadella linked the effort to a broader shift in how Windows PCs could be used for AI tasks. "We needed to make the desktop the most secure place for agents to execute," he said.

Huang said the new Windows foundation would shape enterprise computing. "What Satya just said is going to be the foundation of the next generation of IT as we know it," he said. "Just as Windows and DirectX revolutionized how applications were built, MXC is going to revolutionize how agents are built and deployed."

New PC hardware

Alongside the software changes, Nvidia introduced RTX Spark for Windows laptops and compact desktops. It said the design brings its AI software stack to personal computers built for local model development and always-on agents.

Microsoft is also using the design in a new Surface Laptop Ultra, Davuluri said. He said the device is built around Nvidia RTX Spark and includes up to 128GB of unified memory and up to a petaflop of AI compute.

"We built Surface Laptop Ultra around Nvidia RTX Spark. With up to 128 gigs of unified memory and up to a petaflop of AI compute, you can run models on this laptop that simply don't fit on a traditional machine," Davuluri said.

Systems based on RTX Spark are expected from Acer, Asus, Dell, HP, Lenovo, Microsoft, MSI and Gigabyte. The hardware is due in formats ranging from slim laptops to compact desktops designed to run agents continuously.

Nvidia said RTX Spark combines a Blackwell RTX GPU with up to 6,144 cores and an Nvidia Grace CPU with up to 20 cores. It added that the system can deliver one petaflop of FP4 AI performance and as much as 128GB of unified memory, supporting local model use instead of sending data to the cloud.

The specifications also point to broader use beyond AI development. Nvidia said the hardware supports creator workloads, including AV1 and 4:2:2 video encode and decode, as well as gaming features such as DLSS 5, Reflex and G-SYNC.

Desk-side system

The companies also previewed Nvidia DGX Station for Windows, which they described as a desk-side AI supercomputer for the Windows ecosystem. The system is intended to bring larger AI model development to users who have typically relied on Linux-based environments for heavier workloads.

Davuluri described that shift as a practical change for developers. "This unlocks the power to run frontier-class models locally," he said. "Capabilities that once required renting a cluster, now in a desk-side supercomputer."

Nvidia said DGX Station for Windows runs on the GB300 Grace Blackwell Ultra Desktop Superchip and includes 748GB of coherent memory and up to 20 petaFLOPS of FP4 AI compute. It said that is enough to run models at up to a trillion-parameter scale on a local machine.

The move addresses a longstanding split between Windows and Linux in AI development. Nvidia noted that DGX Station previously ran on Linux, leaving many enterprise developers switching between Linux for large AI workloads and Windows for everyday productivity tools and business software.

That issue is especially relevant because many large companies standardize on Windows desktops. By bringing DGX Station to Windows while still allowing Linux AI toolchains through the Windows Subsystem for Linux, Nvidia and Microsoft are trying to keep developers in one main environment for both model work and office applications.

During the discussion, Huang tied the latest push to Nvidia's earlier history with Microsoft's operating system. "If you look at the entire journey of our company, Windows was at the core of it," he said. "If not for Windows there would be no GeForce."

He also cast the shift in personal computing terms. "The personal computer is the ultimate tool, it's my ultimate tool, and for a whole generation of people, it's our ultimate tool," Huang said. "What happens in the era of agents, when the agent is on your computer? It is now your personal assistant."

Nadella used the discussion to underline his view of Huang's long-term approach to AI. "The thing that I really give Jensen all the credit for, quite frankly, is that consistency of vision of what this can be-not today, not tomorrow, but as a secular trend," he said. "And that's what brings us to this moment."