AIToday
Large Language ModelsAI Business & IndustryAI Watch (Impress)Published: Oct 13, 2026, 19:01 JST

NVIDIA, Microsoft unveil MXC and DGX Station for Windows AI agents

NVIDIA, Microsoft unveil MXC and DGX Station for Windows AI agents

3 Key Points

  1. What happened

    At the 'Windows AI and Surface' event in San Francisco, NVIDIA and Microsoft announced plans for hardware and software to run AI agents on Windows PCs. Microsoft will make its 'Microsoft Execution Containers' (MXC) OS-level infrastructure generally available, and NVIDIA previewed 'NVIDIA DGX Station for Windows'.

  2. Why it matters

    MXC is meant to let agents run safely and continuously in the background under OS control, while the DGX Station is positioned to run frontier-class AI models locally — potentially reducing reliance on cloud infrastructure for demanding agent workloads, though adoption will depend on hardware availability.

WHO IT HITSThis matters to enterprise IT teams planning to deploy AI agents on employee Windows PCs and to developers who build agent-based tools, since local execution could tighten data control and cut cloud dependence.

Not sure about something? Ask the AI

Questions and answers are published on this page.

Summaries like this, in your inbox every morning.

Context & Analysis

The announcement builds on Microsoft's effort to turn Windows into a secure platform for AI agents, with MXC as the foundation. By embedding this capability into the OS, Microsoft aims to let agents be protected, monitored and managed, and it says MXC, Microsoft Security and Agent 365 together can maximize agent potential on Windows.

On the hardware side, the Surface Laptop Ultra is presented as built around NVIDIA RTX Spark, with up to 128GB of unified memory and up to 1 petaflop of AI compute, which Microsoft says allows models that previous machines could not handle. NVIDIA also previewed the DGX Station for Windows, bringing GB300 Grace Blackwell-class AI infrastructure to a deskside AI supercomputer that can run frontier-class models locally, with up to 748GB of memory and up to 20 petaflops of FP4 AI compute — enough for models up to 1 trillion parameters.

What ties these pieces together is the push to make AI agents always-on and directly connected to existing Windows applications and infrastructure. The fact that the DGX Station previously ran on Linux and now supports Windows, while still allowing Linux AI tools via WSL, suggests NVIDIA and Microsoft are trying to bridge the two environments for developers. The announcement did not provide dates for general availability of MXC or for broader availability of the DGX Station.

FAQ
What is Microsoft Execution Containers (MXC)?
MXC is an OS-level infrastructure Microsoft will make generally available so AI agents can run securely and continuously in the background under OS control on Windows.
What hardware did NVIDIA and Microsoft mention for local AI?
Microsoft said the Surface Laptop Ultra is built around NVIDIA RTX Spark, with up to 128GB of unified memory and up to 1 petaflop of AI compute. NVIDIA also previewed the DGX Station for Windows with a GB300 Grace Blackwell Ultra desktop superchip, 748GB of memory and up to 20 petaflops of FP4 AI compute.
Can the DGX Station for Windows run Linux AI tools?
Yes, the Linux AI toolchain can still be used via WSL as needed.
AI Watch (Impress)Read Original Article

AI news that matters for your work, delivered every morning.

Pick your industry and the AI tools you use, and get news related to your work every day.

Free · 30 seconds with Google · unsubscribe anytimeWhat is AIToday? →

Ask AI

Ask AI anything about this article. The AI reads this article, earlier AIToday articles, and Wikipedia, and cites its sources. Q&As are published on this page for other readers too.

Questions and answers are published on this page.

Related Articles

Next articleSB Intuitions opens Sarashina service in free beta with 6 AI models