Modal Adopts Nvidia Vera CPUs for AI-Agent Workloads
The AI cloud company says Vera handled up to twice as many agent tasks in early internal tests, though the result has not been independently validated.
Modal plans to add Nvidia’s Vera CPUs to the infrastructure behind its AI-agent workloads.
What happened
The AI cloud company says its early internal tests showed Vera handling up to twice as many tasks as comparable x86 servers in some container-heavy workloads. The result is company-reported and has not been independently validated. Modal will retain Intel and AMD processors and route different workloads to the most suitable architecture.
Why it matters
Agents use CPUs heavily for software execution, data processing and parallel tasks that do not belong on a GPU. That creates a new server market for processors designed around agent workloads rather than traditional cloud applications.
The bigger picture
Nvidia is expanding from AI accelerators into data-centre CPUs. Modal’s adoption is an early test of whether Vera can win meaningful workloads from established x86 suppliers despite the engineering work required to support Arm-based infrastructure.