GB10 (Grace Blackwell)

GB10 is NVIDIA's Grace Blackwell superchip for desktop AI boxes — the CUDA-native alternative to Strix Halo. It combines a 20-core Arm CPU with a Blackwell GPU and 128GB of coherent unified memory, and ships in the NVIDIA DGX Spark and ASUS Ascent GX10. Its headline '1 PFLOP FP4' is a sparse figure; dense compute is roughly half (~500 TFLOPS, about RTX 5070-class), and at 273 GB/s memory bandwidth is the real ceiling. A 200GbE ConnectX-7 NIC lets two units cluster to run 405B-class models locally.

You buy GB10 for CUDA compatibility and the NVIDIA software stack, not for bandwidth — at 273 GB/s it trails Apple's Ultra chips on token throughput. Choose it if your toolchain assumes NVIDIA; choose Strix Halo or Apple if it doesn't.