Arm has introduced a CPU evaluation methodology designed to measure performance on AI-agent workloads. The methodology was developed with support from Fujitsu, Meta, and OpenAI.
Arm says activity from AI agents—persistent virtual users—is increasing exponentially, creating a need to assess compute, memory, and storage performance under high concurrency.
Alongside the methodology, Arm released independently verified results showing its Arm AGI CPU delivers up to twice the performance per rack of leading x86 server processors on these specific workloads.