Beating AI News Flash: Latham & Watkins, the second-largest law firm in the United States, has begun purchasing its own NVIDIA GPUs to run AI models internally.
The firm has already procured H200 GPUs and placed the servers in a data center accessible only to its own employees. Engineers are fine-tuning NVIDIA's Nemotron 3 open-weight model.
This way, the most sensitive client materials do not need to be handed over to external cloud providers. Commercial models such as ChatGPT, Claude, and Gemini continue to be used, and lawyers can choose different models based on the task.
This is the first publicly disclosed case of a major law firm purchasing its own AI hardware and fine-tuning models. Latham has been preparing for this system for three years, also leaving itself a path that does not depend on a single supplier such as OpenAI or Anthropic.

