U.S. · ALPHAGBM RESEARCH
At a glanceNVIDIA announced on Oct. 16, 2026, that RTX Spark laptops are becoming available through ASUS, Dell, HP, Lenovo, Microsoft and MSI. According to the company, the systems combine RTX graphics, CUDA and NVIDIA’s full-stack AI platform in slim, premium Windows PCs designed for content creation, gaming and locally run AI agents. RTX Spark is built around a single superchip combining an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU, with configurations offering up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance. NVIDIA said the shared memory is intended to let demanding creative projects, large language models and local AI agents run simultaneously on the device. MediaTek collaborated with NVIDIA on the custom CPU design. The company also said guided setup and installers support agents including GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy. Agent permissions can be managed through the NVIDIA OpenShell secure runtime and Microsoft Execution Containers. NVIDIA PAIR can additionally allow RTX Spark to serve AI inference to compatible Windows, Linux and macOS computers over a home network.
Summary
NVIDIA announced on Oct. 16, 2026, that RTX Spark laptops are becoming available through ASUS, Dell, HP, Lenovo, Microsoft and MSI. According to the company, the systems combine RTX graphics, CUDA and NVIDIA’s full-stack AI platform in slim, premium Windows PCs designed for content creation, gaming and locally run AI agents. RTX Spark is built around a single superchip combining an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU, with configurations offering up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance. NVIDIA said the shared memory is intended to let demanding creative projects, large language models and local AI agents run simultaneously on the device. MediaTek collaborated with NVIDIA on the custom CPU design. The company also said guided setup and installers support agents including GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy. Agent permissions can be managed through the NVIDIA OpenShell secure runtime and Microsoft Execution Containers. NVIDIA PAIR can additionally allow RTX Spark to serve AI inference to compatible Windows, Linux and macOS computers over a home network.