NVIDIA and Microsoft Bring RTX Spark AI Agents to Windows PCs

Summary

At Microsoft’s Windows AI and Surface event in San Francisco on October 7, 2026, NVIDIA founder and CEO Jensen Huang and Microsoft CEO Satya Nadella detailed co-engineered Windows hardware and software for persistent AI agents. Hosted by former White House AI policy adviser Sriram Krishnan, the discussion followed Microsoft Windows and Devices EVP Pavan Davuluri’s launch of generally available Microsoft Execution Containers, OS infrastructure for securing, observing and governing background agents alongside Microsoft Security and Agent 365.

RTX Spark laptop preorders opened October 7 for availability October 16, while compact desktops designed for continuous operation arrive in November. Acer, ASUS, Dell, Gigabyte, HP, Lenovo, Microsoft and MSI will offer systems, including Surface Laptop Ultra. RTX Spark combines a Blackwell RTX GPU with up to 6,144 cores, an up to 20-core Grace CPU linked at 600 GB/s, 128GB unified memory and one petaFLOP of FP4 AI performance. It runs the 125B-parameter Qwen 3.8 Flash Next with 51B n-gram locally, without metering or cloud data transfers. CUDA compatibility lets developers move workflows from RTX Spark to DGX Station without rewriting. Creator features include fifth-generation Tensor Cores, NVFP4, AV1, 4:2:2 encoding and decoding, DLSS and ray tracing, while DLSS 5, Reflex and G-SYNC support 1440p AAA gaming above 100 frames per second.

NVIDIA also previewed DGX Station for Windows, its first Windows deskside AI supercomputer using the GB300 Grace Blackwell Ultra Desktop Superchip. Its 748GB coherent memory and up to 20 petaFLOPS of FP4 compute support local models reaching one trillion parameters, fine-tuning, inference and always-on multi-agent workloads. It eliminates the split between Windows productivity workflows and Linux AI systems, while preserving Linux toolchains through WSL. Pricing and availability remain undisclosed.