NVIDIA's first laptops built around its own chip went on sale worldwide on Friday, October 16, 2026. Machines from six manufacturers — ASUS, Dell, HP, Lenovo, Microsoft and MSI — are reaching shelves now, with prices starting at $2,599.

RTX Spark is a single "superchip" that puts an NVIDIA Blackwell RTX GPU and an NVIDIA Grace CPU (an Arm design built in collaboration with MediaTek) side by side, with up to 128GB of memory shared between them. NVIDIA claims up to 1 petaflop of AI performance — measured in FP4, a low-precision format used to run AI models quickly.

The shared memory is the whole point. A large language model has to fit in memory to run, and on most laptops the graphics chip only gets a small slice of its own VRAM. RTX Spark lets a model sit in one big memory pool while creative apps run alongside — NVIDIA's pitch is that you can run frontier AI models and personal AI agents locally, without constant calls to cloud servers, with the added privacy benefit that your data never leaves the device.

Beyond local inference, NVIDIA is touting creator workloads: editing 8K video, rendering 3D scenes over 90GB, and gaming at 1440p above 100fps — all company-reported figures, so treat them as marketing until independent testing lands. On the developer side, NVIDIA says its own comparison showed roughly 4.6× coding-agent performance versus an Intel Panther Lake setup, again measured by NVIDIA itself on its own workload suite.

Early independent coverage is more measured. Roundups note real trade-offs: the Arm Grace CPU trails Apple's latest chips on single-core tasks, some x86 apps rely on emulation, and heavy local AI brings sustained heat, audible fan noise, and a reported 20–25% AI performance drop on battery in some tests. In other words, the hardware story is genuinely new — NVIDIA entering the Windows laptop CPU market through the AI door — but your workload may not feel the marketing numbers.

If you're a developer or data scientist paying for heavy cloud inference, a machine that can run serious local models is worth pricing out. For everyone else, the $2,599 entry point and the early caveats make it a "watch the reviews" story rather than a day-one buy.

Sources