© 2026 NervNow™. All rights reserved.
ByNN Desk
Verda to Deploy NVIDIA’s Next-Gen Rubin Platform in H2 2026 Across Europe, US and APAC
The Helsinki-based AI infrastructure company expects deployments to exceed 100,000 GPUs through 2027 and is opening its first-availability window across three regions for the first time, on the back of NVIDIA Preferred Partner status and tie-ups with ARM, Supermicro and Compal.
The Helsinki-based AI infrastructure company Verda expects deployments to exceed 100,000 GPUs through 2027 and is opening its first-availability window across three regions for the first time, on the back of NVIDIA Preferred Partner status and tie-ups with ARM, Supermicro and Compal.
Helsinki-based AI infrastructure company Verda will be among the first to deploy NVIDIA’s next-generation Vera Rubin (VR200 NVL72) and Rubin (R200) systems at scale, with rollouts beginning in the second half of 2026 across Europe, the US and APAC. The company expects total deployed capacity to exceed 100,000 GPUs through 2027.
It is the first time Verda’s early-access window opens in three regions in parallel. The company has been an early deployer of every NVIDIA generation since Ampere, including Hopper and Blackwell, but until now those rollouts were primarily concentrated in Europe. The shift rests on Verda’s NVIDIA Preferred Partner status and a build-partner network that includes ARM, Supermicro and Compal.
Rubin is NVIDIA’s next-generation AI platform, purpose-built for agentic reasoning, low-latency inference on long contexts and large-scale mixture-of-experts (MoE) training. NVIDIA says the platform delivers up to 10x lower cost per inference token and requires 4x fewer GPUs to train MoE models than its current Blackwell generation, claims the company has not independently verified.
Pioneering teams shouldn’t have to choose between latest-generation hardware and being close to their workloads. Our customers get the latest-generation hardware, in the region they need, on the timeline they need it.
Jorge Santos, COO of Verda
The early access matters for more than customer rollouts. The teams behind Verda AI Lab use the same infrastructure as paying customers, and architecture-level work they do on new hardware feeds back into the platform and support. Founder and CEO Ruben Bryon said being early on every NVIDIA generation since Ampere is how Verda AI Lab has built its expertise, and that the company expects Rubin to extend that advantage.
ALSO READ: Verda Raises $117M to Meet Surging Demand for AI Infrastructure
The company is also leaning on its compliance posture as it expands geographically. Verda has a SOC 2 Type II audit and offers confidential computing, and says it will maintain the same compliance program across all sites as it adds regions.
The bigger picture: the next 18 months of AI infrastructure spend will be defined by who can stand up Rubin capacity first and where. For customers running mixture-of-experts training and long-context inference, the location of early supply matters as much as the supply itself, since latency, sovereignty and proximity to engineering teams all bear on real-world workloads. Verda’s bet is that being early in three regions at once, rather than one, turns a hardware advantage into a footprint advantage.
LATEST MOVES IN AI
New York’s Daloopa Raises $47 Million to Fix the Trust Problem in AI Finance
Syn Ventures Leads $4.8M Seed in AI Security Startup CodeIntegrity
Kopa.ai Raises €2M to Free E-Commerce Teams Buried in Busywork
