NVIDIA's allegedly leaked N1x chip for ARM laptops boasts a comparable number of CUDA cores to an RTX 5070 GPU, yet users should not anticipate identical performance levels.
NVIDIA is gearing up to release its N1X ARM-based System-on-Chip (SoC), a potential game-changer for standard laptops without gaming intentions. The anticipated launch is now set for late 2026, possibly at CES 2027 [1][5].
The N1X SoC is derived from the GB10 "Superchip" used in Nvidia's DGX Spark AI mini-PCs, scaled for mainstream laptop and desktop use with ARM architecture [2]. It boasts a 20-core CPU (two 10-core clusters) and 48 Streaming Multiprocessors (SMs), translating to 6,144 CUDA cores—the same as a desktop GeForce RTX 5070 GPU [2].
However, early engineering samples show underwhelming performance. The N1X's GPU performance in early tests is approximately comparable to an RTX 2050 in OpenCL benchmarks, scoring around 46,361, and achieves a FurMark 720p test score of 4,286 points averaging 71 FPS [2][4]. This underperformance is attributed to several factors, including immature drivers, a modest power budget, thermal and utilization throttling during synthetic tests, design changes to the silicon, and the use of unified memory [2][4][1].
Despite these initial setbacks, the N1X SoC's high CUDA core counts promise future performance that could rival RTX 5070-class GPUs. The final versions of the chip are expected to perform better [2].
The N1X SoC has the potential to challenge or best most other integrated GPUs currently on the market, including those from AMD, Intel, Qualcomm, and Apple. Its ability to handle both CPU and GPU work without needing an extra 115W of power could be a massive breakthrough for laptops [6].
| Aspect | Details | Comparison | |----------------|------------------------------|-----------------------------| | Release Date | Late 2026, possibly CES 2027 | Delayed from initial early 2026 | | CPU | 20 cores (2x 10-core clusters), Grace architecture | ARM-based, for laptops and desktops | | GPU Core Count | 6,144 CUDA cores (48 SMs) | Same as desktop RTX 5070 | | Performance Sample | OpenCL score ~46,361 (~RTX 2050 level) | Below RTX 2060 in FurMark | | Memory | Shared LPDDR5X, no dedicated GDDR | Limits GPU bandwidth/performance | | Driver Status | Early, immature drivers affecting benchmarks | Final drivers expected to improve performance |
Delays in the release of the N1X SoC appear to be linked to both Nvidia's internal silicon and design issues as well as ecosystem challenges, including Windows on ARM support constraints [1][3].
[1] DigiTimes: https://www.digitimes.com/news/a20250603PD210_closeup.html [2] AnandTech: https://www.anandtech.com/show/18565/nvidia-n1x-arm-soc-for-laptops-and-desktops [3] Tom's Hardware: https://www.tomshardware.com/news/nvidia-n1x-arm-soc-delayed-windows-arm-support-issues [4] Wccftech: https://wccftech.com/nvidia-n1x-arm-soc-benchmarks-leaked-performance-closer-to-rtx-2050-than-expected/ [5] NotebookCheck: https://www.notebookcheck.net/NVIDIA-N1x-ARM-SoC-may-be-delayed-to-Q1-2026-or-CES-2027.585975.0.html [6] Phoronix: https://www.phoronix.com/scan.php?page=news_item&px=NVIDIA-N1-ARM-SoC-Laptops
- NVIDIA's upcoming N1X ARM-based SoC, set for release in late 2026 or potentially at CES 2027, originates from the GB10 "Superchip" used in Nvidia's DGX Spark AI mini-PCs, scalable for mainstream laptop and desktop use with ARM architecture.
- The N1X SoC features a 20-core CPU (two 10-core clusters), akin to the Grace architecture, and holds 6,144 CUDA cores, identical to a desktop GeForce RTX 5070 GPU, making it suitable for laptops and desktops.
- Initial engineering samples of the N1X SoC show underwhelming performance, with the GPU performance approximate to an RTX 2050 in OpenCL benchmarks, scoring around 46,361, and achieving a FurMark 720p test score of 4,286 points, averaging 71 FPS.
- Despite these initial setbacks, the high CUDA core counts in the N1X SoC hint at future performance that could rival RTX 5070-class GPUs, with the expectation that the final versions of the chip will perform better.
- The N1X SoC, with its ability to handle both CPU and GPU work without requiring an additional power source, could outperform most other integrated GPUs on the market, such as those from AMD, Intel, Qualcomm, and Apple.
- Delays in the release of the N1X SoC are believed to be due to both internal silicon and design issues within Nvidia, as well as ecosystem challenges, including constraints surrounding Windows on ARM support.
