NVIDIA’s RTX Spark laptops went on sale globally on Friday, October 16, from ASUS, Dell, HP, Lenovo, Microsoft and MSI, with prices starting at $2,599. The pitch is a break from the cloud era: these Windows laptops are built to run AI agents and large language models on the device itself, no data centre round-trip required.
The machines are built around the RTX Spark “superchip,” which pairs a Blackwell RTX GPU with an Arm-based Grace CPU designed with MediaTek. The top configuration carries 20 CPU cores, 6,144 CUDA cores and up to 128GB of unified memory shared between CPU and GPU, with up to 1 petaflop of FP4 AI performance; a second configuration steps down to 18 cores and 64GB. The launch line-up includes the ASUS ProArt P14 and P16, Dell XPS 16, HP OmniBook Ultra, Lenovo Yoga Pro 9n, Microsoft Surface Laptop Ultra and MSI Prestige N16 models. NVIDIA’s own model for the platform, Nemotron 3.5 Bolt, is a 74-billion-parameter model tuned for tool calling and multistep agent tasks, and day-one local agents include GitHub Copilot, Perplexity and WorkBuddy.
The platform was unveiled on October 7 at Microsoft’s Windows and Surface event, where NVIDIA chief Jensen Huang appeared alongside Satya Nadella. Microsoft’s Yusuf Mehdi framed the machines as “hybrid intelligence” — AI that runs locally when it makes sense and reaches for the cloud when it does not — a useful corrective to reading the launch as offline-only. The performance claims, including 1440p gaming above 100 frames per second, are NVIDIA’s own figures and await independent benchmarking.
Why it matters: the economics of AI assistance have so far belonged to the cloud, billed by subscription and by token. A laptop that runs a 74B model and a shelf of agents locally moves that value onto the device — private by default, instant, and free at the margin once bought. That pressures the subscription model from below at the very moment assistants are being woven into search and shopping, from agent-driven checkout in Google Search to the first ads being tested inside AI Mode. If the agents live on the laptop, the platforms will have to argue harder for the cloud.
Source: NVIDIA; Dow Jones Newswires
0 Comments