NVIDIA has officially introduced the RTX Spark, a new ARM-based superchip designed to deliver high-performance AI and graphics capabilities to Windows PCs and laptops.

Unveiled at Computex, the RTX Spark combines 20 MediaTek Arm CPU cores with 6,144 Blackwell RTX cores, offering up to 1 petaflop of AI computing power.The chip also includes a powerful NPU that meets Microsoft's Copilot+ requirements of 40 TOPS.

Developed in collaboration with Microsoft over several years, the superchip aims to optimize both efficiency and performance for modern computing needs.It features unified memory ranging from 16GB to 128GB, allowing the GPU to directly access a large memory pool.

Power consumption is highly flexible, ranging from single-digit watts up to 80W, making it suitable for laptops and small desktops while providing performance similar to an RTX 5070 laptop GPU but with significantly lower power draw.

Microsoft has optimized Windows 11's scheduler specifically for the RTX Spark to ensure efficient workload distribution between CPU, GPU, and NPU.NVIDIA positions this as a major step toward transforming PCs into AI agent-centric devices rather than traditional manual input machines.The superchip will debut this fall in new Windows devices from major manufacturers including Microsoft, Dell, HP, ASUS, Lenovo, and MSI.

This development marks NVIDIA's strong push into the PC market, challenging competitors like AMD and Qualcomm in the AI-accelerated computing space.