NVIDIA unveiled the RTX Spark Superchip, merging AI acceleration with RTX graphics in a single processor designed for Windows PCs. The chip targets creators, developers, and gamers with unprecedented performance in compact form factors, delivering what the company calls the slimmest RTX laptops ever built.

Built around 6,144 cores and 20 processing cores, RTX Spark achieves 1 Petaflop of compute performance while maintaining power efficiency for all-day battery life. Native CUDA support means developers can run the full AI stack locally, while 128 GB of unified memory enables prototyping and fine-tuning large models without cloud dependency.

The architecture combines FP4 Tensor Cores for AI workloads with RT Cores for ray tracing, supporting DLSS, NVIDIA Reflex, and G-SYNC for gaming. Creators benefit from 4:2:2 hardware encoding and AV1 encoders for color-accurate video workflows, while NVIDIA Broadcast enhances streaming quality.

RTX Spark represents NVIDIA's push into personal AI agents that operate continuously at users' desks. With over 1,000 accelerated applications, the platform aims to transform PCs from passive tools into active collaborators for coding, content creation, and gaming.