Vera Delivers the Best of Both Worlds: Highest Throughput and Fastest Agents
AI agents do more than generate answers. They search for information, query databases, write and test code and take action. Each of those steps relies on a CPU, with agents moving between GPU reasoning and CPU execution throughout a task. As businesses put more agents to work, AI factories need CPUs that can execute those […]
In a data center comparison, NVIDIA's Vera CPU outperformed both the fastest x86 CPU and the most energy-efficient x86 CPU. According to NVIDIA's findings, Vera delivers industry-leading agent speed while enabling 1.6 times more agents per watt compared to the fastest x86 CPU. This means that AI infrastructure providers can support a larger number of agents without compromising on speed or power efficiency.
Early adopters have already recognized the exceptional performance of Vera in production-style tests. Perplexity, for instance, found that Vera ran agentic coding tasks 1.5 times faster than traditional CPUs and plans to incorporate Vera into an upcoming production system. DeepInfra achieved up to 2.2 times the speed of the best x86 processor on agentic orchestration, confirming Vera's superior performance.
ClickHouse, too, scaled near-linearly across 176 cores, topping the ClickBench hardware leaderboard without any Vera-specific tuning.
DuckLabs achieved about 1.5 times the median TPC-H SF1000 performance of an Intel Xeon 6 baseline, while Daytona brought 1,000 agent sandboxes online in just 10.8 seconds and achieved 12.7 completed jobs per second with 2,000 concurrent sandboxes using Vera. This compares to 88.4 seconds with x86 hardware.
Modal also demonstrated the potential of Vera, achieving up to 2 times sandbox throughput per core compared to comparable x86 hardware, with gains that increase under heavy load. These results highlight the versatility and efficiency of NVIDIA's Vera CPU, making it an ideal choice for AI factories seeking to maximize agent performance while optimizing power consumption.
Written by urgent.news from NVIDIA Blog's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
