Supermicro and Verda Deploy Liquid-Cooled NVIDIA Blackwell AI Cloud with Waste Heat Reuse

Supermicro and Verda have partnered to deploy a next-generation, sustainable AI cloud infrastructure powered by NVIDIA Blackwell and Blackwell Ultra architectures. Verda, a leading European AI cloud provider, is leveraging Supermicro's end-to-end rack-scale solutions—including NVIDIA GB300 NVL72, HGX B300, HGX B200, and RTX PRO 6000 Blackwell Server Edition systems—to deliver high-performance, energy-efficient compute for large language model training, multimodal AI, robotics, and enterprise applications. The deployment enables Verda to offer self-service instances, serverless containers, and managed inference endpoints to frontier model developers, AI-native scaleups, and regulated enterprises across Europe, the US, and Asia.

Sustainability is a core pillar of this collaboration. Verda’s data centers are powered entirely by 100% renewable energy, and excess heat from the liquid-cooled racks is being repurposed to heat up to 15,000 local homes via partnerships with utilities. Supermicro’s pre-tested, validated systems and rack-scale integration capabilities accelerate deployment timelines while reducing operational risk. The highly energy-efficient architecture also lowers total cost of ownership (TCO), aligning with Verda’s long-term operational efficiency goals. This full-stack AI cloud marks a significant step toward environmentally responsible, scalable AI infrastructure for the next decade.