Nvidia's first N1X benchmark shows single-core performance on par with Apple M3 Max

Nvidia’s N1X chip just appeared in its first public benchmark — and the numbers suggest it could reshape the Windows on Arm market.

Leaker @ExoticSpice101 posted Cinebench 2026 results from a Surface Ultra engineering sample on Friday. The machine scored 540 in single-core performance and 5771 in multi-core. According to the leaker, the single-core figure lands “comparable to Apple’s M3 Max” — the high-end chip Apple ships in its top-tier MacBook Pros.

The caveats arrive right on schedule. The N1X is still an engineering sample, running on pre-production firmware and drivers — all of which can shift by launch. The Cinebench 2026 benchmark itself is new, designed by Maxon to stress modern CPU architectures on rendering and multi-threaded workloads.

Still, the data gives the clearest picture yet of what Nvidia is building for Windows on Arm.

The N1X sits at the heart of Nvidia and Microsoft’s RTX Spark platform, first announced on May 31. The architecture combines Arm-based CPU cores with Nvidia’s Blackwell GPU technology — up to 20 Arm CPU cores, 6144 Blackwell RTX cores, and as much as 128 GB of unified memory in a single package. The platform targets developers and anyone running large AI models locally rather than through the cloud.

Microsoft’s first RTX Spark device is the Surface Laptop Ultra, which the company says will ship with the N1X chip and Blackwell RTX graphics, plus CUDA ecosystem support and that same 128 GB unified memory configuration. The company has pegged its release for later this year.

What this means for the broader Windows on Arm push is straightforward: the ecosystem finally has a credible high-end competitor to Apple Silicon. Qualcomm’s Snapdragon X Elite chips have already brought Arm laptops into the mainstream conversation, but Nvidia’s entry adds GPU firepower that Qualcomm — despite its own Adreno efforts — hasn’t matched at the high end. A laptop that can run CUDA workloads locally, with Blackwell ray tracing cores and unified memory, changes the calculus for developers who have stayed on x86 machines.

Whether the shipping hardware can maintain that M3 Max-level single-core speed is the open question. Engineering samples often punch above their final weight class, and Cinebench is only one benchmark. But for a first showing, Nvidia has the industry’s attention.