Nvidia and Microsoft recently unveiled the much-awaited N1X and N1 SoCs, now called RTX Spark. These Arm-based chips are meant to reinvent Windows PCs for the AI era, with Nvidia claiming that laptops powered by RTX Spark are purpose-built for agentic AI. Offering 1 petaflop of compute, RTX 5070–level GPU performance, and tons of unified memory, these laptops are set to change what Windows on Arm laptops will look like at the premium end of the market. Nvidia is trying to do for Windows what Apple did for macOS, bringing its hardware leadership, software dominance, and partner ecosystem to the personal AI market. To be clear, these machines, especially the maxed-out memory models, will be too expensive for the average user, but adoption by the high-end creator market could slowly influence other segments. Nvidia could see itself competing head-to-head with AMD and Qualcomm, as each of them tries to grab the attention of serious creators and AI professionals.

Nvidia sets its sights on the local AI pie

One company to rule them all

After becoming the predominant hardware vendor powering much of the world's AI data centers, Nvidia wants to capture the local AI market. Rumors surrounding the N1X and N1 chips have been circulating since 2023, convincing consumers that Nvidia was readying its Arm SoC pipeline. We now know these chips as RTX Spark, and the laptops powered by this new platform are packing some serious power. The top-end N1X chip features 6,144 Blackwell cores, 20 Arm-based CPU cores designed in collaboration with MediaTek, and up to 128GB of unified memory with 300GB/s bandwidth. The CPU, GPU, and memory will be connected with Nvidia's NVLink-C2C interconnect, enabling ultra-fast performance. These workhorses will offer local AI enthusiasts up to 1 petaflop of FP4 compute in a 45–80W package. Nvidia claims users can run 120B models with up to a million token context, something most local AI setups can only dream of.

Microsoft is fully onboard the plan, customizing Windows so it can leverage all the power behind RTX Spark. It has created workload profile scheduling (WPS) to optimize the hardware for the OS, while Microsoft Power and Thermal Framework (MPTF) will ensure the chip delivers the goods without draining the battery or raising temps too much. Besides, Microsoft's Prism emulator will handle non-native Arm applications. Nvidia further said that Adobe will fully support RTX Spark, with users being able to use Adobe's MCP server for local agent workloads. Besides Microsoft's Surface Laptop Ultra, other vendors, such as Asus, Dell, Lenovo, HP, and MSI, are also set to launch RTX Spark-powered models later this year.

Local AI deployment often runs into memory constraints, which is where unified memory comes into the picture. Apple Silicon has shown what a shared memory pool can do for massive LLMs, and Nvidia is combining that capability with equally impressive compute to transform AI workloads on Windows PCs. Nvidia also shared details about a DGX workstation with 768GB of memory and an MSI mini PC, showcasing that the technology isn't limited to laptops. After all, RTX Spark is essentially Nvidia's GB10 chip (behind the DGX Spark desktop workstation) repackaged for laptops and mini PCs. Nvidia hasn't confirmed pricing yet, but RTX Spark machines, especially the N1X models, will reportedly start from around $2,899.

AMD isn't worried yet, but RTX Spark adoption could change everything

Arm could dominate x86

After the RTX Spark announcement, AMD executives confidently declared that they're "excited that Nvidia has joined the game." For context, AMD launched its RTX Spark competitor way back in early 2025. Its Strix Halo platform, specifically the Ryzen AI Max+ 395 chip, features 16 Zen 5 cores and 32 threads, a 40-CU integrated GPU powered by RDNA 3.5, and up to 128GB of unified memory. Laptops and mini PCs with Strix Halo chips have been around for a while now, so it's not hard to see AMD's confidence — it was technically the first mover in the local AI hardware space. It also launched its DGX Spark rival, the Ryzen AI Halo developer mini PC, at $3,999, which is $700 cheaper than Nvidia's GB10 machine, thanks to a recent price hike by Team Green.

You could also argue that Nvidia's RTX Spark machines aren't directly competing with AMD, since the latter isn't present in the Windows on Arm world at all. Its Strix Halo, as well as the upcoming Gorgon Halo laptops, are x86 devices. That said, if Nvidia gains traction with its high-end RTX Spark laptops, it could increase the pressure on Team Red. Nvidia's superior software stack, powered by technologies like CUDA, TensorRT, DLSS, OptiX, and Reflex, could easily sway the tide in its favor as creators and professionals flock to RTX Spark devices. The pricing will probably be more than AMD's Gorgon Halo laptops, but relatively affordable RTX Spark devices could later join the fray, further eating into the x86 market.

What will eventually matter in the local AI hardware race is superior performance combined with reliable software and a competitive price. Buyers in this segment won't be as price-conscious as the regular consumer, but they'll still expect some level of price parity when choosing between AMD and Nvidia models. AMD's ROCm software stack is no longer a major roadblock, but it still trails Nvidia's CUDA in key areas. New models and software packages arrive for CUDA first, followed by ROCm, and usually with some rough edges. Tinkerers will be okay with making AMD hardware work for them, but Nvidia's superior endpoint experience might be preferred by the majority of consumers. If RTX Spark laptops are within the same price segment as that of AMD, broadly speaking, it could be enough to pave the way for Nvidia's growth in the local AI space.

Qualcomm just got a Windows on Arm competitor

Snapdragon X, meet RTX

Qualcomm's existing Snapdragon X Elite (and X2 Elite) aren't the only Windows on Arm devices anymore. Nvidia's Arm chips have made the Windows on Arm conversation much more exciting. As the Snapdragon machines don't quite extend beyond the $2,000 mark, even for the maxed-out models, comparing them to the top-end RTX Spark laptops might seem odd, but you need to remember that Nvidia's Arm laptops with the cheaper N1 chip will reportedly start from around $1,800. At that price, users can expect a 12-core CPU, 2,560 CUDA cores (RTX 5050 equivalent), and 16GB of unified memory from an RTX Spark N1 machine. A Snapdragon X2 Elite Extreme laptop, such as the Asus Zenbook A16 (2026), offers 18 cores, 48GB unified memory, and an integrated Adreno GPU.

When it comes to local AI, the Snapdragon X2 Elite Extreme packs the world's fastest NPU to date, offering 80 TOPS of AI compute. While it's hard to compare NPUs and GPUs on the same metric, Qualcomm might offer competitive performance in agentic AI workloads, thanks to its superior CPU power. Nvidia's offerings will outclass Qualcomm's in terms of local AI content creation due to the dedicated GPU. The bigger advantage that Nvidia might have is its mature software stack. Windows on Arm has struggled in some respects on the Snapdragon devices, partly due to less-than-universal driver and app support. Nvidia brings its tried-and-tested architecture and acceleration technologies that creators, professionals, and gamers are already familiar with, giving the company a clear ecosystem advantage.

Of course, this narrative hinges on local AI workloads becoming way more popular than they are right now. However, with AI agents set to transform potentially every single personal and professional workload, it's not that far-fetched. We're set to enter a market where Intel, AMD, Qualcomm, and Nvidia could compete in the same price segment, for the same consumer, and in the same workloads. If we weren't in the middle of the worst PC hardware crisis, I would have termed this year as the best yet to buy a Windows laptop.

Nvidia might just end up dominating yet another consumer market

Nvidia has remained the leader, by far, in gaming as well as enterprise GPUs for years now. With its new RTX Spark SoC, it is set to transform high-end consumer laptops in ways we hadn't predicted. Thanks to the rapid development of local and agentic AI, consumers are looking for powerful, tailored devices that can run massive LLMs without compromises. The RTX Spark N1X and N1 machines will cater to this very need while competing with AMD, Intel, and Qualcomm laptops, making for an exciting few years ahead.