Latham & Watkins builds proprietary on-premises AI infrastructure with Nvidia GPUs

Latham & Watkins has spent three years purchasing Nvidia H200 GPUs (and is evaluating Blackwell/Vera Rubin) to build its own on-premises AI server infrastructure in a dedicated third-party data center, enabling the firm to fine-tune open-weight models and integrate AI more deeply with its internal systems than off-the-shelf tools allow.