NVIDIA and Microsoft bring RTX Spark laptops and AI agents to Windows PCs – Unite.AI

NVIDIA opened pre-orders for RTX Spark laptops and Microsoft announced general availability of Microsoft Execution Containers, the operating system-level infrastructure for running AI agents on Windows, at Microsoft’s Windows AI and Surface event in San Francisco on October 7, 2026, according to NVIDIA’s event recap.

The event, held at Dogpatch Studios, featured a chat between NVIDIA founder and CEO Jensen Huang and Microsoft CEO Satya Nadella, hosted by Sriram Krishnan, former White House senior policy advisor on artificial intelligence. Huang traced NVIDIA’s support for Windows and Microsoft over several decades, saying that Windows has been at the center of the company’s journey and that without Windows there would be no GeForce. Nadella credited the consistency of Huang’s vision as a secular trend that has brought companies to this moment.

Microsoft Execution containers reach general availability

The event opened with Nadella before Pavan Davuluri, executive vice president of Windows and Devices at Microsoft, explained Windows updates designed for the agent era. Chief among them: Microsoft announced the general availability of Microsoft Execution Containers (MXC), the operating system-level infrastructure that enables agents to run securely and persistently in the background under the control of the operating system.

Davuluri said that Microsoft is integrating these primitives directly into Windows so that agents can be protected, observed and governed, pointing to MXC, Microsoft Security and Agent 365. Nadella returned to this theme in his talk with Huang, saying that Microsoft needed to make the desktop the most secure place for agents to run. Huang said the approach will form the basis of the next generation of IT: “Just as Windows and DirectX revolutionized the way applications were built, MXC will revolutionize the way agents are built and deployed.”

RTX Spark laptops and compact desktops

RTX Spark puts the full NVIDIA AI stack into Windows laptops and compact desktops. Laptop pre-orders opened on October 7, 2026, with laptops available from October 16, 2026, and compact desktops on sale in November 2026. The systems come from Acer, ASUS, Dell, HP, Lenovo, Microsoft, MSI, and Gigabyte, with designs ranging from slim laptops to compact desktops made for always-on agents. Introducing the Surface Laptop Ultra, Davuluri said Microsoft built it around RTX Spark, with up to 128GB of unified memory and up to a petaflop of AI compute—enough, he said, to run models that don’t scale on a traditional machine.

The RTX Spark superchip combines an NVIDIA Blackwell RTX GPU with up to 6,144 cores and an NVIDIA Grace CPU with up to 20 cores connected at 600GB/s, delivering up to a petaflop of AI FP4 performance with up to 128GB of unified memory. NVIDIA’s RTX Spark product page lists a second laptop configuration that pairs a 5,120-core GPU with an 18-core Grace CPU and up to 64GB of LPDDR5X memory, with laptop power limits of 45-80W, while the compact desktop runs at a 140W TDP in a small chassis designed for 24/7 operation. All configurations run Windows 11. Laptop models shown on the page include ASUS ProArt P16, Dell XPS 16, HP OmniBook Ultra 16, Lenovo Yoga Pro 9n, Microsoft Surface Laptop Ultra, and MSI Prestige N16 Flip AI+.

NVIDIA said RTX Spark runs the entire CUDA platform natively, so developers can move models, tools and workflows without rewriting and run the same NVIDIA AI stack from RTX Spark to DGX Station; the company says the platform runs models of up to 120 billion parameters with 1 million local contexts. NVIDIA also said the systems can run models like Qwen 3.8 Flash Next, a 125 billion model weighing 51 billion n-grams that it says matches the intelligence of many cloud models, without limits and without sending data to the cloud. For creators, NVIDIA lists fifth-generation Tensor Cores with hardware-accelerated NVFP4 and AV1 support and 4:2:2 video encoding and decoding; for gamers, the company says RTX Spark runs AAA titles at 1440p above 100 frames per second with DLSS 5, Reflex and G-SYNC.

NVIDIA DGX station for Windows

The event also previewed the NVIDIA DGX Station for Windows, which NVIDIA describes as the first deskside AI supercomputer to bring GB300-class Grace Blackwell AI infrastructure directly into the Windows ecosystem. “This unlocks the power to run frontier-class models locally,” Davuluri said. “Features that once required renting a cluster are now available in a deskside supercomputer.”

DGX Station for Windows runs on the GB300 Grace Blackwell Ultra desktop superchip, which connects an NVIDIA Blackwell Ultra GPU to a 72-core Grace CPU via NVLink-C2C at 900 GB/s. The system offers up to 748GB of coherent memory and up to 20 petaFLOPS of FP4 compute, which NVIDIA says is enough to locally run models with up to one trillion parameters. The DGX Station for Windows product page lists 252GB of HBM3e GPU memory along with 496GB of LPDDR5X CPU memory, a ConnectX-8 SuperNIC with up to 800Gb/s for connecting multiple systems, an optional RTX PRO 6000 workstation GPU for ray-tracing visualization and simulation, and access to Linux AI toolchains via the Windows Subsystem for Linux.

NVIDIA first announced DGX Station for Windows on May 31, 2026 at GTC Taipei, stating in its launch release that the system will support NVIDIA OpenShell on Windows – an open source runtime that creates an isolated sandbox for each agent using new Windows security and containment primitives – and that it can run hundreds of agents simultaneously. Until now, DGX Station ran on Linux, which meant enterprise developers maintained separate environments for heavy AI workloads and Windows productivity applications; NVIDIA notes that the vast majority of Fortune 500 companies use Windows as the standard.

DGX Station for Windows is expected in Q4 2026 from ASUS, Dell Technologies, GIGABYTE, HP, MSI, and Supermicro, and NVIDIA is tracking notifications for availability.