2x throughput is the only figure in the Tom's Hardware document, and it comes from Modal's internal tests of Nvidia Vera CPUs against Intel and AMD chips in AI agentic workflows. The number does not prove that Vera beats x86 in every server workload, nor that the gap survives outside Modal's fleet.

Die Brief reads this as a workload-specific signal, not a CPU architecture verdict. Modal keeps x86 in a heterogeneous fleet, which means the buyer is constrained by routing, not by a clean replacement. For the fab, the useful fact is that Arm-based Vera is being placed beside x86 in production AI work, not replacing it.

The document leaves out the benchmark workload, test methodology, throughput units, specific Intel and AMD models, deployment date, power draw, and server count. Without those, the 2x figure is a deployment observation, not a reproducible spec.

The 2x figure is a workload-specific deployment signal, not a general x86 replacement claim. Modal's mixed fleet means buyers should treat Vera as a routing option, not a default CPU swap.

The document does not state the benchmark workload, test methodology, throughput units, specific Intel and AMD models, deployment date, power draw, or server count.

The document supports a narrow Arm advantage in Modal's agentic workload, but lacks methodology, units, and chip models. The heterogeneous fleet keeps x86 relevant and limits the result to a routing decision.