Nvidia's RTX Spark laptops are now on sale starting at $2,399, putting a Blackwell-generation GPU and a Grace CPU on a single superchip inside slim Windows machines from six PC makers — and marking the moment local AI moved from a desktop niche to a standard laptop feature.
The rollout was announced by Nvidia on October 16, with a follow-up post on October 17 titled "RTX Spark Ignites Global Celebration as New Windows PCs Arrive" confirming the machines are reaching customers worldwide. Launch coverage from outlets including Make Tech Easier on October 18 highlighted the entry price and the superchip design at the heart of the platform.
The Specs: One Superchip, Shared Memory
At the foundation of every RTX Spark machine is a single superchip combining an NVIDIA Blackwell RTX GPU with an NVIDIA Grace CPU, with up to 128GB of unified memory accessible to both. MediaTek collaborated with Nvidia on the custom CPU design, contributing to power efficiency and connectivity, according to the company's announcement.
The headline number is AI performance: up to 1 petaflop of FP4 compute, which Nvidia says is enough to run frontier-class large language models and AI agents locally. Because the memory is unified, demanding creative projects and local AI models can run simultaneously rather than competing for separate pools of RAM.
The same silicon is pitched at conventional workloads. Nvidia claims RTX Spark machines can render 3D scenes exceeding 90GB and play AAA games at 1440p above 100 frames per second with DLSS 5, including launch titles such as Gears of War: E-Day and NBA 2K27.
Six PC Makers at Launch
The launch window includes systems from ASUS, Dell, HP, Lenovo, Microsoft and MSI, with additional manufacturers to follow, per Nvidia's press release. That is an unusually broad coalition for a new PC category at introduction, and it signals how thoroughly the PC industry has signed on to Nvidia's local-AI framing.
Early demand indicators are suggestive, if anecdotal. Wccftech reported that the higher-end 128GB unified memory configuration of Microsoft's Surface Laptop Ultra — the $2,599 machine announced earlier this month as the flagship RTX Spark PC — had already sold out, with Nvidia's CUDA software ecosystem cited as a driver. The $2,399 entry price undercuts the Surface configuration, which should widen the funnel beyond early adopters.
Local AI Is the Point
Nvidia is explicit that the target customer is not just a gamer who wants ray tracing. The press release centers on running AI agents on-device from day one, citing setup flows for agents including Hermes and OpenClaw, alongside creative applications such as Adobe Photoshop and Premiere that tap the GPU for generative features.
"RTX Spark brings the power of NVIDIA AI, RTX graphics and CUDA together with Microsoft Windows in one PC for creating, gaming and running personal AI agents," said Jensen Huang, Nvidia's founder and CEO, in the announcement. "Together with Microsoft and our PC industry partners, we're bringing creators, developers and gamers a new generation of Windows PC, built for the work they do today and the personal AI agents they'll work with next."
Microsoft, whose Windows division has staked its next act on AI, framed the hardware as half of a two-sided bet. "We're building Windows to be the platform for hybrid intelligence, where AI can run locally when it makes sense and reach the cloud when needed," said Yusuf Mehdi, Microsoft's executive vice president, in the same release.
What It Means for the Market
RTX Spark completes a product line that began with the desktop-class DGX Spark machines launched earlier this year at $4,999, which our AI hardware coverage followed closely. The laptops bring the same architectural idea — unified CPU-GPU memory sized for model weights — to a portable, mass-market form factor, and at hundreds of dollars less than the flagship Surface configuration.
The competitive implications are pointed. Apple's MacBooks have been the default answer for developers running local models, thanks to Apple Silicon's unified memory. Nvidia and its partners are now arguing that Windows machines can do the same job with the added pull of CUDA, the software layer that most AI tooling is built against.
Two caveats belong in any purchase calculus. The performance figures above come from Nvidia's own announcement, and independent reviews of shipping units are only beginning to appear. And the 1-petaflop figure is FP4 precision — ideal for inference, but not comparable to the FP32 numbers historically used to market GPUs.
Still, the arrival is a threshold moment. Local AI on a laptop has moved from demo to shelf, and the question is no longer whether premium Windows PCs will ship with AI-first silicon, but how quickly the capability trickles down to mainstream prices.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
