Lambda Taps Supermicro Servers, Nvidia Blackwell Chips to Build AI Factories

Super Micro Computer Inc. said Monday that cloud-infrastructure provider Lambda has begun deploying a wide range of Supermicro GPU-optimized servers powered by Nvidia’s new Blackwell architecture, expanding what the companies call "AI factories" designed for large-scale training and inference workloads. The roll-out, which started in June at Cologix’s COL4 Scalelogix data center in Columbus, Ohio, gives customers in the US Midwest access to enterprise-grade compute aimed at speeding development of generative-AI models. Lambda plans to offer similar capacity to AI labs, corporations and hyperscalers worldwide. Lambda is using Supermicro systems such as the SYS-A21GE-NBRT with Nvidia HGX B200 and the company’s AI Supercluster racks featuring GB200 and GB300 accelerators. The liquid-cooled servers are intended to curb power and cooling costs while providing what Lambda describes as gigawatt-scale computing on demand.