Lambda has deployed the first hydrogen-powered NVIDIA GB300 NVL72 systems at ECL’s Mountain View data center, marking a milestone in sustainable AI infrastructure. The Supermicro-built, 142 kW systems run on hydrogen fuel cells at ECL’s off-grid, zero-emission campus, where Lambda has doubled its footprint. The units, weighing 4,000 pounds each, were integrated into production in just two hours, underscoring the operational viability of high-density, liquid-cooled AI systems in hydrogen-powered environments.
The deployment pairs Supermicro’s direct-to-chip liquid cooling technology with ECL’s hydrogen power architecture, which recycles water byproducts from energy generation into cooling loops. This design allows Lambda to meet the extreme density and efficiency demands of NVIDIA GB300 NVL72 systems while aligning with sustainability requirements for AI factories. Lambda now occupies 100% of ECL’s MV1 facility, signaling confidence in hydrogen as a foundation for future gigawatt-scale AI clusters.
Executives from Lambda, Supermicro, and ECL positioned the launch as a model for next-generation AI infrastructure. “As we move toward gigawatt AI factories, diversified power is becoming essential infrastructure,” said Ken Patchett, VP Data Center Infrastructure, Lambda. “These NVIDIA GB300 NVL72 systems represent the building blocks for training and deploying tomorrow’s foundation models. Hydrogen-based energy ensures we can power that superintelligence-class compute responsibly.”
• First hydrogen-powered, production-grade NVIDIA GB300 NVL72 systems now online
• Each system delivers 142 kW compute, cooled via direct-to-chip liquid loops
• Integration achieved in under two hours at ECL’s Mountain View MV1 campus
• Lambda doubled footprint to full campus occupancy (from 50% to 100%)
• Systems weigh 4,000 lbs, highlighting power density and cooling challenges
• Partnership involves Lambda (deployment), ECL (hydrogen data centers), and Supermicro (system build)
🌐 Analysis: This collaboration illustrates the convergence of AI compute density and clean energy infrastructure. NVIDIA’s GB300 NVL72 pushes data centers into unprecedented thermal and power requirements, forcing operators to pursue diversified energy sources. Hydrogen-based off-grid campuses, like ECL’s MV1, provide a real-world test case for zero-emission, high-performance AI factories. Competitors such as CoreWeave and IREN are scaling AI capacity primarily with grid-connected renewables, but Lambda’s hydrogen-first approach demonstrates a parallel path for sustainable scaling at gigawatt levels.
🌐 We’re tracking the latest developments in AI infrastructure. Follow our ongoing coverage at: https://convergedigest.com/category/ai-infrastructure/
