NVIDIA's Vera CPU pairs agent speed with density: 1.6x more agents per watt than the fastest x86
Synopsis
NVIDIA has introduced Vera, a CPU aimed at AI agents, seeking to meet both requirements in one chip: fast execution of agent steps and high concurrent agent counts. Previously AI factories had to choose between the fastest x86 processor and the most energy-efficient one to balance speed and capacity. In a 100-megawatt data center comparison, NVIDIA reports Vera supports 1.6x more agents per watt than the fastest x86 CPU and delivers 2x faster agent speed than the most energy-efficient x86 CPU; early adopters including Perplexity and DeepInfra report speedups in production-style tests, though Perplexity's deployment is still planned.
Interpretation
NVIDIA puts agent workload speed and concurrent capacity on a single CPU, so AI infrastructure providers no longer have to mix fast and dense x86 parts for different agent tasks. Operators previously had to pick between the fastest x86 processor and the most energy-efficient x86 processor to balance per-agent latency against how many agents could run at once. In a 100-megawatt data center comparison, NVIDIA reports that a single Vera CPU performs better than both the fastest x86 CPU and the most energy-efficient x86 CPU.
Agents move back and forth between GPU reasoning and CPU execution during a task, and steps such as searching, querying databases, and writing and testing code land on the CPU. Vera is NVIDIA's CPU designed for these agent steps, positioned to cover a range of agent workloads with one chip. NVIDIA gives two comparisons: versus the fastest x86 CPU, Vera enables 1.6x more agents per watt; versus the most energy-efficient x86 CPU, Vera delivers 2x faster agent speed.
Early adopters report speedups on Vera in production-style tests spanning agentic coding, orchestration, database queries and sandbox startup. These tests compare Vera with x86 platforms on the same workloads rather than reporting chip specifications alone. Perplexity reports agentic coding tasks 1.5x faster and plans to deploy Vera in an upcoming production system; DeepInfra reports up to 2.2x the speed of the best x86 processor on agentic orchestration; Daytona reports 1,000 agent sandboxes online in 10.8 seconds versus 88.4 seconds on x86.
Perspective
The report covers NVIDIA's published 100-megawatt data center comparison and the production-style test results reported by Perplexity, DeepInfra, ClickHouse, DuckLabs, Daytona and Modal, for engineers and operators evaluating agent infrastructure.
NVIDIA's core comparison is its own estimate; Perplexity's Vera deployment is still a planned production system, and the other partner figures are their own reported test results.