Microsoft and NVIDIA unveiled a new generation of thin-and-light Windows PCs at NVIDIA GTC, built around the RTX Spark platform. These machines target developers, creators, and power users who need local AI capabilities and graphics performance without the traditional desktop footprint. The partnership combines Microsoft's platform leadership with NVIDIA's silicon expertise.

RTX Spark delivers 1 petaflop of AI performance through up to 6144 Blackwell RTX cores, paired with 20 Arm-based efficient cores and up to 128GB of unified memory. This architecture supports intensive creative workflows, gaming, and local AI model execution. The system integrates full NVIDIA AI and RTX graphics technology while maintaining power efficiency through the Microsoft Power and Thermal Framework.

Windows optimizations include workload profile scheduling for better multi-core scaling, expanded GPU-accessible memory limits, and enhanced Prism emulation for x86 applications on Arm. Microsoft also moved core experiences to WinUI3 and improved WSL integration. These changes extract maximum performance from RTX Spark's heterogeneous design.

The partnership targets the emerging agent computing paradigm, with NVIDIA bringing OpenShell to Windows for secure agent containment. OEMs and app developers have optimized software for the Arm architecture, ensuring broad application support at launch. This represents a significant shift toward local AI processing on Windows platforms.