Latham & Watkins Buys Nvidia Servers to Build In-House Legal AI Stack
Per the Financial Times, Latham & Watkins, the second-largest US law firm with $8.3B in 2025 revenue, is buying its own Nvidia servers and building GPU racks, planning to spend possibly hundreds of millions of dollars on an internal legal AI system: fine-tuning Nvidia's open-weight Nemotron models with data accessible only to firm staff, bypassing paid tokens from OpenAI/Anthropic. The story spread widely on X and is seen as a landmark case of a data-intensive vertical industry building a sovereign AI stack.
Confirmed
- Latham & Watkins is purchasing Nvidia servers to build internal AI infrastructure (FT report, echoed across multiple posts)
- The approach: fine-tuning Nvidia's open-weight Nemotron models, with data restricted to internal staff
- Budget described as "possibly hundreds of millions of dollars"
- Firm revenue of $8.3B in 2025
Why it matters
- Commentators highlight the formula: open weights + proprietary data + local compute = sovereign enterprise AI stack. The firm holds decades of contracts, negotiation histories, client context and legal reasoning it won't hand to external API providers
- @ivanbezdomny and others argue large firms with troves of proprietary data are post-training their own models before OpenAI/Anthropic penetrate their verticals; some call it the signal Big AI vendors fear most
- Law going first could push other regulated industries to follow, possibly triggering a corporate GPU-hoarding wave
- edsimo notes Nvidia wins either way: frontier labs buy GPUs, and open-model adoption drives even more GPU sales
2026-09-14 ~ 2026-09-15 · 8 related posts
Primary sources
- Second-largest US law firm may spend hundreds of millions fine-tuning open-weight Nemotron on its own GPUs — ivan_bezdomny ·
- No.2 US law firm Latham & Watkins buys Nvidia servers to fine-tune open models in-house — ai ·
- Law firm Latham & Watkins buys Nvidia servers to run AI in-house as enterprises stock up on GPUs — AccBalanced ·
- [source] Second-largest US law firm may spend hundreds of millions fine-tuning open-weight Nemotron on its own GPUs — ivan_bezdomny · 2026-09-14
- [source] No.2 US law firm Latham & Watkins buys Nvidia servers to fine-tune open models in-house — ai · 2026-09-14
- Latham & Watkins buys Nvidia servers for in-house AI; post-training on proprietary data called a no-brainer — AccBalanced · 2026-09-14
- US's No.2 law firm Latham & Watkins builds in-house AI stack with Nvidia servers — ayushtweetshere · 2026-09-14
- [source] Law firm Latham & Watkins buys Nvidia servers to run AI in-house as enterprises stock up on GPUs — AccBalanced · 2026-09-15
- $8.3B Law Firm Latham & Watkins Builds In-House AI Stack, Signaling Local Model Bifurcation — ayushtweetshere · 2026-09-15
- US's #2 law firm Latham & Watkins is building its own in-house Nvidia AI stack — MikeBirdTech · 2026-09-15
1 near-duplicate retellings: ayushtweetshere