NVIDIA Preferred Partner status, paired with a build-partner network spanning NVIDIA, ARM, Supermicro, and Compal, takes AI infrastructure company Verda's first-availability window into three regions in parallel.
HELSINKI, Finland (June 2nd, 2026) European AI infrastructure company Verda will be among the very first to deploy NVIDIA Vera Rubin NVL72 and HGX™ Rubin NVL8 at scale – with the earliest deployments in H2 2026 and the expected capacity exceeding 100,000 GPUs throughout 2027. Verda has been among the first to deploy every NVIDIA generation since Ampere, and Rubin extends that first-availability window beyond Europe, into the US and APAC.
The Rubin platform is purpose-built for agentic reasoning, low-latency long-context inference, and massive-scale mixture-of-experts training, with NVIDIA reporting up to 10x lower cost per inference token and 4x fewer GPUs to train MoE models versus Blackwell.
First on every generation, now across three regions
Verda has had Ampere, Hopper, and Blackwell systems running production workloads at scale ahead of most of the European market. With Rubin, that first-availability window opens across three regions in parallel – Europe, the US, and APAC – instead of just one. The development builds on Verda’s position as a NVIDIA Preferred Partner, combined with a partner network spanning industry leaders across the value chain, including ARM, Supermicro, and Compal.
“Pioneering teams shouldn't have to choose between latest-generation hardware and being close to their workloads. The partner network we've built is what removes that tradeoff. Our customers get the latest-generation hardware, in the region they need, on the timeline they need it,” says Jorge Santos, COO of Verda.
Why being early matters: the AI Labs feedback loop
Rubin is purpose-built for the workloads Verda already specializes in: large-scale mixture-of-experts training and serving, agentic inference on long contexts, and real-time generative video. The teams behind Verda AI Lab run on the same infrastructure as Verda’s customers, and the architecture-level expertise they build through early hardware access flows back into the platform and the support offered to customers.
“Being early on every NVIDIA generation since Ampere is how Verda AI Lab has built deep architecture expertise, and that translates directly into customer outcomes. We expect Rubin to compound that advantage,” Ruben Bryon, Founder and CEO of Verda, said.
Security and regulatory compliance are foundational to Verda, attested by its confidential computing offering and SOC 2 Type II audit. As the company expands its global reach, Verda will enable customers to leverage the same full-stack AI infrastructure with more extensive regional availability while maintaining a uniform rigorous compliance program across all sites.
For additional information:
Jani Hellström
Head of Comms, Verda
[email protected]