Nvidia spent the last decade dominating the chips that train and run artificial intelligence inside data centers. On May 31 it told the world it now wants the same role inside your laptop. At Computex 2026, the company unveiled the RTX Spark superchip — its first processor built for Windows PCs — and confirmed that machines using it will ship from Microsoft, Dell, HP, Asus, Lenovo and MSI.
For a company whose last quarter brought in a record $81 billion in revenue, the PC market is not where the money is today. But it is one of the few large compute markets Nvidia does not yet own, and the move signals where the company believes the next wave of AI demand is heading: onto the device, not just in the cloud.
What the Chip Actually Is
The RTX Spark is, in Nvidia’s framing, two of its flagship chip types fused into one package. It pairs a Blackwell graphics processing unit — the same architecture powering Nvidia’s data-center accelerators — with a new Arm-based “Grace” central processing unit, plus 128 gigabytes of unified memory shared between the two.
That unified-memory design is the part worth understanding. In a conventional PC, the CPU and GPU keep separate pools of memory and shuttle data between them. By giving both a single large pool, Nvidia is targeting AI workloads that need to hold big models in memory locally — the kind of “agentic” software that runs assistants and copilots directly on the machine rather than calling a server. Nvidia is pitching Windows on this platform as something closer to an AI operating system, with models running on-device.
The first laptops are expected to be thin-and-light designs, some as slim as 14 millimeters, aimed at creators, AI developers and gamers rather than the mainstream office buyer — at least to start.
Why Nvidia Wants This Market
Three forces are pulling Nvidia into PCs.
First, the Arm transition. Windows on Arm has spent years as a niche, but the combination of better battery life and on-device AI has made Arm-based laptops a genuine growth category. Nvidia’s own Arm CPU lets it capture the whole chip, not just the graphics slice.
Second, the on-device AI thesis. If the industry is right that a meaningful share of AI inference moves from data centers to local hardware over the next few years, the company that supplies that local silicon captures a new and recurring market. Nvidia would rather be early than defend its data-center lead alone.
Third, competition. Intel and AMD have both leaned hard into “AI PC” chips, and Qualcomm’s Arm-based laptop processors opened the category. The semiconductor tape has been volatile lately — even amid the AI boom, the sector saw a sharp selloff on bubble fears in May — and Nvidia entering PCs reshuffles the competitive map for everyone holding chip stocks.
The Market Read
The announcement landed during one of the strongest stretches the semiconductor sector has ever had. The PHLX Semiconductor Index is up more than 65% on the year, and the Nasdaq closed at a record into the holiday weekend, helped by Nvidia’s news. Memory names rode the same wave, with Micron crossing a $1 trillion valuation days earlier on AI demand.
There are reasons for restraint, too. The PC market is large but low-margin compared with data-center accelerators, and the first RTX Spark machines are aimed at a premium slice of buyers, not the volume center of the market. Whether on-device AI becomes a mass-market reason to upgrade — or stays a feature for developers and enthusiasts — is the open question. Reports describe strong industry interest, but shipping silicon and selling laptops are different milestones.
What to Watch Next
The near-term signposts are concrete: launch timing and pricing for the first RTX Spark laptops from the named PC makers, early reviews of on-device AI performance, and how Intel, AMD and Qualcomm respond. Longer term, the test is demand — whether buyers treat local AI as a reason to replace aging machines. That is the bet Nvidia is making by stepping outside the data center for the first time.
Related reading: