RTX Spark PCs Are About to Change Windows Forever

The most powerful new Windows laptops in years are about to step into the spotlight. NVIDIA's RTX Spark platform — the chipmaker's first system-on-a-chip built for Windows PCs — is set to headline a Microsoft Windows and Surface event in San Francisco on October 7, 2026, where Microsoft CEO Satya Nadella and NVIDIA CEO Jensen Huang are both scheduled to appear on stage, according to reporting by Neowin. If the teasers are anything to go by, this is the moment NVIDIA goes from supplying chips for other companies' computers to putting its own silicon at the heart of Windows itself.

RTX Spark pairs a twenty-core Grace Arm CPU with Blackwell-generation RTX graphics in a single module, with up to 128GB of unified memory shared between the processor and graphics card instead of sitting in separate pools. NVIDIA says the top configuration delivers roughly one petaflop of AI performance at FP4 precision — enough to run models with roughly 120 billion parameters and a million-token context window entirely on the device, per Guru3D's rundown of the platform's announced specifications. In practice, that means workloads that used to need a data-center connection could run on a laptop sitting on your desk.

Why this event matters more than a normal product launch

Microsoft is framing the show around how local AI will shape the next chapter of the PC. A teaser published this week narrowed the theme further, promising RTX Spark and "new experiences designed to unlock more possibilities for developers and builders," as The Verge noted in its coverage of the event preview. The framing is a shift in emphasis for a company that has spent the past two years pushing Copilot and cloud services: the pitch now is that serious AI work should happen on the device, privately and without a subscription meter running.

The timeline to watch has been building for months. NVIDIA first unveiled the RTX Spark concept at Computex in June 2026, then in early September 2026 Reuters reported that Lenovo and Acer would be among the first manufacturers shipping RTX Spark machines in October, a story carried by KTWB. Microsoft added its own hardware to the picture at Build 2026, when it unveiled the Surface RTX Spark Dev Box with serious on-device AI compute and that full 128-gigabyte memory pool. Now the October event looks like the date all the threads get tied together: pricing, configurations, shipping dates, and the Windows features built to take advantage of the hardware.

Six laptops are already lined up

NVIDIA has pointed to six initial notebook designs: the ASUS ProArt P16, Dell XPS 16, HP OmniBook Ultra 16, Lenovo Yoga Pro 9n, Microsoft's own Surface Laptop Ultra, and the MSI Prestige N16 Flip AI+. Acer and Gigabyte systems are expected to follow later, according to Guru3D. These are creator and professional machines rather than gaming laptops — a deliberate positioning choice, since NVIDIA is selling the platform first as a workstation for local AI, 3D work, and development rather than as a rival to traditional gaming notebooks.

The physical specs NVIDIA has quoted are striking for hardware this powerful. The company says RTX Spark laptops can be around 14 millimeters thick and weigh roughly three pounds, depending on the manufacturer, with premium tandem-OLED displays part of the reference platform. Two chip configurations are planned: the flagship twenty-core version with 6,144 CUDA cores and 24GB to 128GB of memory, plus a lighter eighteen-core, 5,120-CUDA-core version with 24GB to 32GB aimed at laptops only, per Android Headlines' report on the IFA announcements.

For buyers, the closest comparison is what happened when Apple moved its laptops to a unified-memory architecture: suddenly memory that used to be split between the CPU and GPU could be shared, and machines that looked modest on paper punched far above their weight for creative work. NVIDIA is betting that the full unified memory pool on a Windows machine will do the same thing for local AI agents, huge context windows, and models that previously needed a cloud subscription or a server rack.

The price question and the fine print

Nobody has confirmed retail pricing yet, and the San Francisco show should finally fill in that blank. Analyst estimates reported by Tech Insider put the fully loaded configuration near three thousand dollars, with more modest builds starting below two thousand — but those are estimates, not price tags, and the real numbers will decide whether RTX Spark machines land as developer workstations or as premium machines normal buyers can justify.

There is also a fair counterpoint to the hype. Windows on Arm has historically been a mixed bag, with inconsistent software compatibility and trouble from anti-cheat systems in some games. NVIDIA claims its machines will handle modern games well, but independent benchmarks do not exist yet, and the company's own marketing is the only evidence so far. Until reviewers get units, the safest read is that these are AI workstations first and gaming machines second — which matches exactly how NVIDIA and Microsoft are positioning them.

Still, the direction of travel is hard to ignore. Two years ago the conversation about AI PCs was dominated by Qualcomm's Copilot+ machines and modest on-device models. Now the most valuable chip company on the planet is shipping its own processor into Windows laptops, with the head of Windows sharing a stage with the head of NVIDIA to explain why. For anyone shopping for a laptop this fall, the advice is simple: wait for the October event before deciding. The next generation of the Windows PC is about to get its price tags — and if you care about the latest gadgets, you will want to compare these machines against Google's new October laptop lineup before opening your wallet.