Microsoft and NVIDIA bring always-on agents to Windows: MXC agent containers go GA, RTX Spark laptops ship Oct 16

Why it matters

Key points

At a Windows AI and Surface event in San Francisco on October 7 (early October 8 Beijing time), Microsoft and NVIDIA laid out a joint push to make AI agents run persistently on Windows PCs, with Jensen Huang and Satya Nadella in a fireside chat. NVIDIA's blog summarized the announcements.

On software, according to NVIDIA's account, Microsoft announced general availability of Microsoft Execution Containers (MXC), OS-level infrastructure that lets agents run safely and persistently in the background under operating-system control, alongside Microsoft Security and Agent 365 for observation and governance.

On hardware, NVIDIA introduced RTX Spark: a Blackwell RTX GPU with up to 6,144 cores and an up to 20-core Grace CPU, up to 128GB of unified memory and one petaflop of FP4 compute. Laptop preorders opened the same day with availability on October 16; compact desktops go on sale in November. Acer, ASUS, Dell, HP, Lenovo, Microsoft, MSI and Gigabyte have systems, including Microsoft's new Surface Laptop Ultra. NVIDIA says it can run a 125B-parameter model such as Qwen 3.8 Flash Next locally without sending data to the cloud. The event also previewed DGX Station for Windows, built on the GB300 Grace Blackwell Ultra Desktop Superchip with 748GB of coherent memory and up to 20 petaflops of FP4, which NVIDIA says can run trillion-parameter-scale models locally; no ship date was given.

Verifiable

Primary page: https://blogs.nvidia.com/blog/local-ai-rtx-spark-microsoft-windows-event/

NVIDIA's blog RSS timestamps it October 7, 2026, 18:45 GMT. Compute, memory and "runs model X locally" claims are vendor specifications without third-party testing. MXC's exact capabilities and permission model should be checked against Microsoft's own documentation; this piece relies on NVIDIA's account of the event. DGX Station for Windows has only a preview and sign-up page, with no price or ship date.

Our take

The point of this launch is not the chip specs but that Windows is turning "how an agent runs on your computer" into an operating-system capability. Until now, local agents either ran in sandboxes users built themselves or simply ran with the user's full permissions. If MXC really puts isolation, monitoring and governance into the OS, enterprise IT finally has a reason to let agents live on employee machines, which matters more for adoption than another GPU.

Laptops with 128GB of unified memory also push always-on local mid-size open models toward mainstream PC price bands, a real gain for privacy-sensitive and usage-metered work. It is still a platform and hardware update rather than a jump in model capability, so it goes to Latest. Running locally is not the same as running well: real speed, power draw and price will only be clear from reviews after the October 16 launch.