- Updated: March 11, 2026
- 2 min read
Thinking Machines Lab Secures Gigawatt‑Scale AI Compute Deal with Nvidia
Thinking Machines Lab (TML), the AI research hub founded by former OpenAI and DeepMind veterans, has announced a multi‑year partnership with Nvidia that will bring a gigawatt‑scale compute platform to its new research campus. The deal includes the deployment of Nvidia’s next‑generation Vera Rubin supercomputers and access to the Tinker API, positioning TML as one of the world’s most powerful AI‑focused compute clusters.
According to the announcement, the partnership will deliver more than 1 GW of AI‑optimized compute power, enough to train the most demanding foundation models at a fraction of current costs. Nvidia will also make a strategic equity investment in TML, underscoring its commitment to fostering cutting‑edge AI research outside of the traditional cloud providers.
The new infrastructure will be housed in TML’s flagship campus, slated to open later this year. The facility will combine custom‑built hardware, Nvidia’s DGX‑H100 systems, and a high‑speed networking fabric to enable researchers to run experiments that were previously infeasible due to compute constraints.
Industry analysts see the deal as a signal that the AI landscape is shifting toward specialized, on‑premise compute clusters that can deliver petaflop‑scale performance with lower latency. By securing a dedicated gigawatt of Nvidia‑powered hardware, TML aims to accelerate breakthroughs in generative AI, reinforcement learning, and large‑scale model alignment.
Read the original story on TechCrunch for more details.
Related reads on ubos.tech:
- AI Infrastructure: Building the Future of Compute
- Nvidia Partnerships in 2026: What They Mean for the Industry
- Inside Thinking Machines Lab: Vision, Mission, and Projects
