Summary created by Smart Answers AI
In summary:
- Nvidia launched RTX Spark, a new Windows on Arm processor featuring a 20-core Grace CPU and 6,144 Blackwell RTX cores, challenging Qualcomm in the laptop market.
- PCWorld questions whether these laptops can effectively support always-on agentic AI due to battery life constraints that conflict with constant connectivity requirements.
- Major manufacturers like Asus, Dell, HP, and Lenovo will ship RTX Spark laptops this fall, targeting demanding tasks like 12K video editing and AI generation.
Nvidia has a vision to put AI productivity into the hands of virtually everyone on the planet. On Sunday night it advanced that mission with the debut of the Nvidia RTX Spark, a challenge to Qualcomm as the second Windows on Arm processor for consumer and business PCs.
In conjunction with Mediatek, which designed the N1 and N1X CPU at the heart of the RTX Spark platform, Nvidia is “reinventing the personal computer, for creating, for gaming, for agents,” Nvidia chief executive Jensen Huang (or an AI facsimile?) told attendees at Nvidia’s GTC conference at the opening of Computex 2026 in Taipei.
PC companies, ever quick to jump on a new trend, are on board with Nvidia’s vision. Asus, Dell, HP, Lenovo, and MSI showed RTX Spark laptops that will ship in the fall. Acer and Gigabyte will follow that wave. Microsoft, which has pioneered Windows on Arm development with Qualcomm, will release the Surface Laptop Ultra based upon the RTX Spark platform inside.
From one perspective, Nvidia’s RTX Spark processor is a powerful, yet familiar update to the PC. It’s loaded with a 20-core Arm “Grace” CPU, developed by Mediatek, connected to 128GB of unified memory via a 600 GB/s NVLink connection. The kicker is the integrated GPU, which contains a petaflop of AI performance, which Microsoft revealed contains 6,144 Blackwell RTX cores, about the equivalent of an RTX 5070 CPU. We’ve known about the N1X for some time — and now it’s reached fruition.
Nvidia
Will consumers accept Nvidia’s framing, which positions that powerful GPU as a gaming device? Huang appeared with two laptops playing Microsoft’s Forza franchise and the 007: First Light game. Or will RTX Spark machines be celebrated as AI productivity enablers?
Nvidia’s PC business is now a fraction of its total revenue, and in fact most of Huang’s speech covered the buildout of large data centers using Nvidia’s systems and infrastructure. Nonetheless, with RTX Spark, Huang showed he hasn’t lost touch with the PC line that launched Nvidia decades ago.
Huang said AI will be fundamental to the PC experience: “That is the modern application: an [AI] agent.”
“What becomes of our personal computer in a world of agents, agents running native, connected to models, local or in the cloud, our personal AI sandboxed for security, running continuously, getting work done?” Huang asked. “The chips and the OS must evolve.”
What we know of RTX Spark PCs
Unlike rival CPU makers, which prioritize the speed, core count, number of threads, and other deeply technical characteristics of their CPU cores, Huang downplayed the N1X, barely mentioning it by name. Instead, he focused on the RTX Spark platform itself, and how the RTX Spark platform fit within the overall Nvidia AI ecosystem.
An early leak indicated that the N1X would consist of 20 cores, confirmed by Microsoft, made up of Cortex-X925 “extreme ” cores and Cortex-A725 cores, with an 18-core variant as well. (In an email, Arm itself confirmed the N1X configuration.)
Nvidia
The lower-end N1 could be made from 12 (8+4) cores, as well as a 10 (7+3) configuration. The key will be the number of CUDA cores, which will affect the GPU and AI capabilities: 6,144 or 5,120 in the N1X, and 2,560 or 2,048 CUDA cores within the N1, VideoCardz reported. At the low end, that’s about the number in a GeForce RTX 2050 GPU.
The one metric that we didn’t hear: TOPS, which had been a staple of AI hardware discussions for several years. (Microsoft officials confirmed that the RTX Spark platform at the heart of the Surface Laptop Ultra has enough TOPS to qualify it as a Copilot+ PC, but didn’t disclose the actual figure.)
“RTX Spark lets creators, AI developers and gamers render ultralarge 90GB+ 3D scenes, edit 12K 4:2:2 video, generate 4K AI videos, run 120B-parameter LLMs with up to 1 million tokens context using agents locally, and play AAA games at 1440p and over 100 frames per second,” Nvidia said.
Nvidia and its partners said that RTX Spark laptops would be available in 14-inch and 16-inch sizes, as thin as 14mm and weighing three pounds on up. Nvidia stated that the laptops would be made from a “precision-machined aluminum chassis” with color-accurate OLED displays.
Those laptops, “purpose-built for personal agents,” will be complemented by several mini PCs, also with an RTX Spark chip inside, Huang said, showing them off next to the third-party laptops. Huang also showed several DGX Station desktops, which will be out of reach for most consumers with loadouts of 768GB of memory for running trillion-parameter AI models. For every successive generation of CPU, Nvidia will have a chip for laptops, one for desktops, and one for workstations, he said.
Nvidia
Huang said he envisions the PC evolving much the same way as a smartphone does, which is now barely a phone and more of a pocket computer.
“Here’s my theory,” Huang said. “I could totally imagine that someday there’s actually an AI supercomputer in your house, and it’s running all of your agents, it’s running all of your assistants, and they’re doing all kinds of things for you all the time.
“You have to have it in your house, just like you have a home theater in your house,” Huang said. “You want assistant AI agent computers running in your house, and these in time becomes a lot more like R2D2 to you. It becomes more like C3PO to you than it feels like a PC to you.”
Huang said he would have a “conversation” with Microsoft chief executive Satya Nadella talking about their agentic vision for PCs later in CES.
Adobe also offered its support, pledging to add a new video pipeline in Adobe Premiere specifically tapping into the RTX Spark unified memory. Adobe’s Substance 3D Painter and Stager will run natively on RTX Spark, the company said. Adobe’s next-generation Photoshop engine “will be optimized for GPU-accelerated compositing, enabling live filters, high dynamic range and modern natural brushing,” it added, with an “AI-native pipeline.”
Gaming laptops? Sure. But agentic PCs?
What we haven’t heard is whether the concept of an agentic PC actually can live in a laptop world. By definition, an agentic PC is always active, always connected, sending snippets of code around the Internet to perform the tasks you ask of it.
Nvidia
But laptops, by the traditional definition, migrate. They prioritize performance, yes, but also battery life. Gaming laptops with large GPUs typically last only a couple of hours at best before running out of juice. I’d like my agents to keep working while I’m on the road… but they can’t when the laptop is hibernating.
Is that what we hope for in the agentic AI era, laptops with constantly spinning fans, trapped in a backpack as their owners rush to find a nearby power cable? Or can agents actually run with a laptop closed, quietly getting work done with all-day battery life?
Neither Microsoft nor laptop makers have told us. It’s very easy to buttonhole the RTX Spark platform as just another gaming laptop, first and foremost. But if these laptops are going to be productivity devices, we’ll need to hear how they fit into the laptop paradigm. Something’s missing.
I’m absolutely willing to believe that mini PCs with Nvidia’s RTX Spark platform inside can be a success story for AI aficionados working at their desk. I’m much more skeptical about how RTX Spark laptops can succeed as on-the-go, always-connected productivity devices. Nvidia, Microsoft, and hardware vendors will need to convince me that this new platform is in fact better than anything we already have on the market…or just another gaming PC dressed up in a business suit.