How much will you pay to be on the frontlines of an agentic AI world? I’ve wondered this since Sunday night, when Nvidia unveiled its first-ever consumer processor for PCs at Computex 2026. With 20 CPU cores and 6,144 CUDA graphics cores packed into a single chip, the flagship RTX Spark is no slouch—and the devices sporting one won’t be, either.
All of Nvidia’s device partners, which includes Dell, Asus, MSI, HP, Lenovo, and Microsoft—promise the thinnest, lightest machines for AI-development laptops and desktops. (Frankly, a mini PC with so much power seems more a flex than a practical move, but I respect the engineering.) In particular, Microsoft devoted time to explaining its design choices during a Windows & Devices Showcase presentation at its Build conference, which is also happening this week.
Nothing sounds cheap—but no one’s talking price specifics yet. At least, not officially.
When I posed the question directly to Microsoft and Nvidia representatives at Build 2026, they largely deflected in their answers. For the Surface Laptop Ultra and the Surface RTX Spark Dev Box, Andrew Hill, corporate vice president of Surface, said, “We’re not ready to talk about pricing yet until we get closer to availability,” but did confirm a fall release.
Alaina Yee / Foundry
As for Nvidia, one representative echoed the same sentiment, saying I should expect pricing information closer to the launch period. (The word I heard several times? That it would be “months” ahead of launch.) But they did give at least some insight into the coyness: Everyone’s waiting to see how memory and storage prices shake out closer to fall.
Meanwhile, over in Taiwan, other vendors have been more loose-lipped. When my colleague Adam Patrick Murray asked the same question at Computex 2026, one source told him devices with the RTX Spark N1X chip (the same one shown on stage by Nvidia CEO Jensen Huang) will start at $2,500. PCs with a stepped-down N1 processor will start around $2,000.
Analysts have apparently heard similar numbers, with chatter on X/Twitter quoting a Morgan Stanley analyst as believing N1X models will start at $2,900, and the N1 models at $1,800.
For ordinary folks, those numbers will hurt—but arguably, these devices straddling the line between enterprise and consumer may not seem expensive to those seeking easier, less expensive development. (Few people get free, unlimited AI tokens.) We’ll have to see if that translates to a trickle-down effect for the rest of us, both in terms of performance boosts…and higher hardware costs.
