Latham & Watkins has purchased Nvidia GPU servers and begun developing its own in-house artificial intelligence systems, giving the law firm greater control over sensitive client data while reducing its dependence on external AI and cloud providers.
The move represents an unusually significant infrastructure investment for the legal industry. Latham has purchased several servers equipped with multiple GPUs and is using the hardware to run and customize open-weight AI models internally. The Financial Times reported that Latham is the first major law firm publicly known to establish this type of in-house AI infrastructure.
Latham’s machine learning and software engineers are currently fine-tuning Nvidia’s Nemotron 3 open-weight models for the firm’s requirements. Unlike closed models offered through providers such as OpenAI and Anthropic, open-weight models can be downloaded, customized, and operated on infrastructure controlled by the user.
One of the primary motivations is client confidentiality. Running AI workloads internally allows Latham to process particularly sensitive information without necessarily sending that data to an external cloud provider.
The firm leases space in a secure data center facility accessible only by Latham personnel, where its GPU infrastructure is housed. While operating its own systems also means assuming responsibility for cybersecurity, maintenance, and hardware operations, the approach provides greater control over where sensitive information is processed.
Latham is not abandoning commercial AI platforms. Instead, the firm is building an architecture that lets lawyers and technology teams choose between internally operated models and third-party AI services depending on the task.
That flexibility could become increasingly important as AI usage grows and providers adjust their pricing, access policies, and product offerings. Latham’s internal infrastructure provides an alternative if the economics or terms of major commercial AI services change.
The strategy also represents a significant expansion of Latham’s internal technology capabilities. The firm now employs more than 900 technology specialists, including machine learning engineers, AI engineers, software professionals, innovation lawyers, and lawyers with coding expertise.
Latham generated approximately $8.3 billion in revenue last year, giving it the financial scale to make infrastructure investments that could be difficult for smaller firms to replicate. The firm has not disclosed how much it has invested in the initiative.
Owning and operating GPU servers can require substantial capital spending and specialist personnel, but it also gives Latham direct control over its computing infrastructure and the ability to tailor AI systems to its specific legal workflows.
The move illustrates how generative AI adoption in the legal industry is beginning to progress beyond subscriptions to third-party software. Large professional-services firms are increasingly evaluating whether AI models, engineering talent, and computing infrastructure should become core internal capabilities rather than services purchased entirely from technology vendors.
