Thinking Machines Lab Signs Massive Nvidia Deal for 1GW+ of Vera Rubin Chips

Futuristic AI data center with rows of glowing Nvidia GPU server racks

Thinking Machines Lab, a one-year-old AI startup, has signed a multiyear deal with Nvidia to deploy more than 1 gigawatt worth of next-generation Vera Rubin chips. Nvidia is also making a “significant” investment in the startup as part of the partnership.

The Scale of the Deal

The agreement calls for more than 1 gigawatt of computing power using Nvidia’s upcoming Vera Rubin architecture — a staggering amount for any company, let alone a startup that’s barely a year old. To put that in perspective, 1 gigawatt is enough power to run approximately 750,000 homes. In the context of AI computing, this represents one of the largest single commitments of GPU resources ever made.

Who Is Thinking Machines Lab?

Founded just over a year ago, Thinking Machines Lab is focused on training frontier AI models. The startup has moved quickly from founding to securing one of the largest GPU deals in the industry. While details about the company’s specific research agenda remain limited, the scale of the Nvidia deal suggests ambitious plans for training very large AI models.

Nvidia’s Investment Strategy

The deal signals Nvidia’s broader strategy of backing promising AI companies with both hardware access and capital. By investing in Thinking Machines Lab while simultaneously signing a massive chip supply agreement, Nvidia ensures demand for its next-generation hardware while also gaining exposure to potential breakthroughs in AI research. It’s a pattern Nvidia has repeated with several AI startups.

Vera Rubin: The Next Generation

The Vera Rubin architecture represents Nvidia’s next-generation GPU platform, succeeding the current Blackwell chips. Named after the astronomer who discovered evidence of dark matter, the Vera Rubin chips are expected to deliver significant performance improvements for AI training workloads. The fact that companies are already signing deals for these chips before they’re available shows the intensity of the AI hardware arms race.

The Bottom Line

A one-year-old startup securing a 1+ gigawatt GPU deal from Nvidia would have been unthinkable even two years ago. The AI arms race has reached a scale where startups are making infrastructure commitments that would have been reserved for nation-states or the largest tech companies. Whether Thinking Machines Lab can deliver results that justify this massive investment remains to be seen — but Nvidia is clearly willing to bet big on finding out.