NVIDIA made Rtx Spark laptops available today, and the first systems are rolling out from ASUS, Dell, HP, Lenovo, Microsoft and MSI. The new Windows PC class pairs local AI work with a slim laptop form factor. It also pushes a new burden onto buyers: sort out which partner model matches their memory and agent needs, because the launch is broad before it is simple.
RTX Spark from ASUS to MSI
The platform delivers up to 128GB of unified memory and up to 1 petaflop of FP4 AI performance. That gives users room for larger local models and faster on-device inference without sending every task to a remote service. The tradeoff is obvious: this is a portable PC built for heavier AI work, not a bare-bones laptop.
Jensen Huang said RTX Spark brings NVIDIA AI, RTX graphics and CUDA together with Microsoft Windows in one PC. He also said NVIDIA Blackwell RTX GPU and NVIDIA Grace CPU sit inside the system. MediaTek collaborated with NVIDIA on the custom CPU design. The practical result is a machine aimed at creators, developers and gamers who want local compute rather than a cloud-first workflow.
Dillon Rolnick on Hermes
Dillon Rolnick said Hermes was brought to RTX Spark so people can use open AI models and agents locally on a Windows laptop. That is the real shift for buyers: the laptop is not just running apps, it is set up to host agents on-device without a command line. For people building or testing AI tools, the difference is whether the work stays on the PC or gets pushed elsewhere.
RTX Spark also includes DLSS, OptiX, Reflex, G-SYNC, TensorRT and hardware-accelerated video. NVIDIA says users can get started with agents including GitHub Copilot, Hermes, OpenClaw, Perplexity, Pi and WorkBuddy without requiring a command line. The launch therefore mixes consumer-friendly setup with a deeper stack for AI users, which should help adoption but also raises the first practical question for shoppers: which RTX Spark laptop models are available today from each partner, and at what prices?
