How Windows is being optimized for Nvidia’s RTX Spark laptops

Microsoft and Nvidia recently showcased the first wave of RTX Spark laptops in San Francisco, introducing a new computing platform that combines Arm-based CPUs with Blackwell GPUs and unified memory. According to Microsoft principal program manager Peter Dawoud, significant engineering work went into readying Windows for these chips, which utilize Microsoft's Prism emulator to translate x64 applications to the platform's unique microarchitecture. A primary focus of the RTX Spark platform is local AI performance. The laptops feature a new Get Started setup experience designed to help users install and launch AI agents in just two clicks, with Hermes, Perplexity, and OpenClaw serving as the initial supported agents. Furthermore, Microsoft implemented extensive memory optimizations for systems like the Surface Laptop Ultra, which can support up to 128GB of unified memory, allowing users to run large local models while maintaining system responsiveness and even balancing memory allocation to the GPU using an Nvidia app slider. Beyond AI tasks, Microsoft and Nvidia have heavily optimized the platform for gaming. Because RTX Spark laptops rely on emulation for many titles, Microsoft regularly ships monthly Prism updates to improve compatibility and frame pacing. The platform also includes native support for several major game anti-cheat engines out of the box, with additional engines from Activision, EA, PUBG, and Riot Games slated for support through 2027. Despite these advanced capabilities across AI and gaming, certain Copilot Plus PC features such as Recall are delayed on RTX Spark devices at launch. Adrienne Brewbaker, senior director of Windows marketing, noted that these experiences are expected to become available to Windows Insiders later in the year.