Nvidia RTX Spark May Get a Lower-End GPU Configuration With 5,120 CUDA Cores
Nvidia’s ARM-based RTX Spark chip is already shaping up to be one of the most interesting new platforms for AI PCs, developer hardware, and next-generation laptops. After officially introducing RTX Spark on May 31 alongside the first devices expected to use it, Nvidia confirmed that the chip would arrive in multiple configurations. Now, a newly released native GeForce graphics driver appears to reveal another version of the chip.
The latest clue comes from Nvidia’s official driver package version 616.00 for Microsoft’s RTX Spark Dev Box. Inside the package, an INF file named nv_surface_woa.inf reportedly includes references to two different graphics configurations.
The first configuration matches what Nvidia previously announced: a Blackwell-based GPU with 6,144 CUDA cores. The second configuration appears to be a slightly scaled-down version with 5,120 CUDA cores.
That suggests Nvidia may be preparing at least two RTX Spark variants, potentially targeting different price points, performance levels, or device categories. A 5,120 CUDA core version could allow manufacturers to offer more affordable RTX Spark laptops or compact desktop systems while still delivering strong GPU and AI acceleration.
At the moment, the driver does not reveal whether this lower-end RTX Spark version will differ in other areas. Details such as CPU core count, memory capacity, clock speeds, power limits, and AI performance remain unknown.
Nvidia has already confirmed that RTX Spark can feature up to a 20-core Grace CPU, up to 1 petaflop of AI performance, and up to 128 GB of unified memory. It is possible that the 5,120 CUDA core model keeps the same CPU and memory options while reducing only GPU resources. If that is the case, Nvidia may be taking an approach similar to other chipmakers that offer multiple versions of the same processor family with different GPU core counts.
The same driver package also includes references to “NVIDIA Desktop Device” and “Nvidia NPU.” The desktop device mention may be related to Microsoft’s RTX Spark Dev Box, though the driver files do not provide enough detail to confirm exactly what hardware it refers to. The Nvidia NPU reference is also notable, as neural processing units are becoming increasingly important in AI-focused PCs and Windows on ARM devices.
RTX Spark has already appeared in benchmark testing inside a Surface Laptop Ultra test system equipped with a 20-core CPU. Early Cinebench results suggest the chip can compete with high-end processors such as AMD’s Ryzen AI Max+ 395 and Intel’s Core Ultra X9 388H. However, it reportedly falls behind Qualcomm’s Snapdragon X2 Elite and Apple’s M5 Max in both single-core and multi-core performance.
Even so, RTX Spark is not only about traditional CPU benchmarks. Nvidia’s biggest advantage is likely to come from its combination of ARM-based CPU cores, Blackwell graphics, CUDA support, unified memory, and AI acceleration. For developers, creators, and AI users, native Nvidia GPU support on Windows ARM hardware could be a major selling point.
If the 5,120 CUDA core configuration launches, it may help Nvidia expand RTX Spark into a wider range of devices. High-end models could use the full 6,144 CUDA core GPU, while more efficient or lower-cost systems could use the reduced version. This would give laptop makers more flexibility and could make RTX Spark hardware more accessible to users who want strong AI and graphics performance without paying for the top-tier configuration.
For now, Nvidia has not officially announced this second RTX Spark variant. Still, the driver references strongly suggest that the company is preparing more than one GPU configuration for its ARM-based platform. More details are likely to emerge as RTX Spark laptops, dev kits, and desktop systems move closer to release.
