European Leaders Turn to NVIDIA Vera CPUs to Build Local Infrastructure

Nebius and Nscale are among the partners building around NVIDIA’s Vera CPUs, and the bet is pretty straightforward: agentic computing needs faster CPUs, not just bigger models. Agents don’t only answer prompts. They pull information, run tools, execute code and check what came back, which means the CPU side has to keep up if the whole system is going to feel responsive.

Vera is being used to support that layer of the stack. NVIDIA says its design helps agents respond faster and lets data centers make better use of compute capacity, even when thousands of tasks are running at once. That matters for cloud providers trying to grow services without turning every extra request into a slowdown.

The European angle runs through the infrastructure partners. AI cloud providers Nebius, Nscale and Verda are building with Vera, while Aleria is part of the broader sovereign AI ecosystem focused on keeping data and operations under customer control. Server makers Bull, Dell Technologies, HPE, Lenovo and Supermicro are also building Vera-based systems, so the chip isn’t just for one style of deployment. It can sit in standalone CPU servers or work alongside NVIDIA GPUs in accelerated systems.

Early testing is also giving Vera something more concrete than a marketing pitch. ClickHouse, Cloudera, Duck Labs, Elastic, Polars, Redpanda, SAP and Starburst all reported strong results on analytics, extract-transform-and-load and semantic search workloads. NVIDIA points to per-core performance, LPDDR5X memory bandwidth and its Scalable Coherency Fabric as the reasons those workloads move better.

Then there’s the agent sandbox crowd, where the whole business model depends on quick, isolated startup cycles. Baseten-Blaxel, Daytona, E2B, Modal, Prime Intellect and Vercel are building on Vera for the infrastructure that lets agents execute code, run tools and test results. In that corner of the market, speed isn’t a nice-to-have. It’s the product.