Global Finance News
07 Nov 2026, 11:18
NVIDIA Says Vera CPU Delivers Faster AI Agent Performance With Higher Energy Efficiency
NVIDIA (NASDAQ: NVDA) said its Vera CPU can improve both AI agent performance and energy efficiency, targeting data centers increasingly constrained by power availability as agentic AI workloads expand.
According to NVIDIA, Vera enables 1.6 times more AI agents per watt than the fastest x86 CPU while delivering industry-leading agent speed. Compared with the most energy-efficient x86 processor tested, Vera delivered roughly twice the agent speed while maintaining strong power efficiency.
The company highlighted early performance results from several partners. Perplexity reported agentic coding workloads running 1.5 times faster on Vera than traditional CPUs, while DeepInfra achieved up to 2.2 times the speed of its best x86 processor in agent orchestration workloads.
Other tests showed ClickHouse scaling nearly linearly across 176 cores, while Daytona brought 1,000 AI-agent sandboxes online in 10.8 seconds compared with 88.4 seconds using x86 infrastructure. Modal reported up to twice the sandbox throughput per core.
The results reinforce NVIDIA’s push beyond GPUs into CPUs and full-stack AI infrastructure, as data center operators increasingly seek to maximize AI workloads per watt while supporting growing numbers of autonomous agents.
NVIDIA (NASDAQ: NVDA) said its Vera CPU can improve both AI agent performance and energy efficiency, targeting data centers increasingly constrained by power availability as agentic AI workloads expand.
According to NVIDIA, Vera enables 1.6 times more AI agents per watt than the fastest x86 CPU while delivering industry-leading agent speed. Compared with the most energy-efficient x86 processor tested, Vera delivered roughly twice the agent speed while maintaining strong power efficiency.
The company highlighted early performance results from several partners. Perplexity reported agentic coding workloads running 1.5 times faster on Vera than traditional CPUs, while DeepInfra achieved up to 2.2 times the speed of its best x86 processor in agent orchestration workloads.
Other tests showed ClickHouse scaling nearly linearly across 176 cores, while Daytona brought 1,000 AI-agent sandboxes online in 10.8 seconds compared with 88.4 seconds using x86 infrastructure. Modal reported up to twice the sandbox throughput per core.
The results reinforce NVIDIA’s push beyond GPUs into CPUs and full-stack AI infrastructure, as data center operators increasingly seek to maximize AI workloads per watt while supporting growing numbers of autonomous agents.
