Arm vs x86 server
Arm servers run CPUs built on the Arm instruction set, such as NVIDIA Grace. x86 servers run Intel Xeon or AMD EPYC. Each needs its own builds.
Arm and x86 are instruction set architectures (ISAs), the machine instructions a CPU understands, and software must be built for the one it runs on. Intel's Xeon and AMD's EPYC are x86. Arm licenses its instruction set and its Neoverse core designs to other chip makers. NVIDIA's Grace uses Neoverse cores, while its newer Vera runs custom NVIDIA cores on the same instruction set.
Grace has 72 Arm Neoverse V2 cores, and a GB200 NVL72 rack pairs 36 Grace CPUs with 72 Blackwell GPUs. Vera has 88 Olympus cores that run Arm v9.2 software, including existing Arm containers and binaries. NVIDIA uses Vera as the host CPU in Vera Rubin NVL72 racks and in the HGX Rubin NVL8 platform. On the x86 side, the DGX B200 hosts eight Blackwell GPUs with two Intel Xeon Platinum 8570 CPUs, 112 cores in total. Software ships as separate builds for each, aarch64 (64-bit Arm) for Arm and x86-64 for Intel and AMD.
Choose Arm when the GPU platform is built around it, as Grace and Vera are in NVL72 racks, and every container has an aarch64 build. Choose x86 when any required application, driver or agent ships only as an x86-64 binary.
Sources
Source | Publisher |
|---|---|
NVIDIA | |
NVIDIA Vera CPU Delivers High Performance, Bandwidth, and Efficiency for AI Factories | NVIDIA |
NVIDIA | |
Arm |
- Publisher
NVIDIA
- Publisher
NVIDIA
- Publisher
NVIDIA
- Publisher
Arm
Last verified October 7, 2026.
- Arm vs x86
- aarch64 vs x86-64
- ARM64 server
- Arm server CPU
- NVIDIA Grace CPU
- NVIDIA Vera CPU
- Neoverse V2
- AWS Graviton
- Ampere AmpereOne
- Arm vs x86 for AI servers