Overview
- Deployment of the next‑generation Vera Rubin systems is targeted to begin early next year to train Thinking Machines’ frontier models and support its platforms.
- The companies will co‑design training and serving infrastructure optimized for Nvidia architectures and plan to broaden access to frontier and open models for enterprises and research institutions.
- Financial terms were not disclosed, and Reuters cited industry executives who estimate a one‑gigawatt build could cost on the order of tens of billions of dollars.
- Thinking Machines, founded in 2025 by former OpenAI CTO Mira Murati, previously raised about $2 billion at a roughly $10–12 billion valuation, with Nvidia participating.
- The pact adds to Nvidia’s pattern of pairing chip supply with ownership stakes, following reported multibillion‑dollar investments in OpenAI and Anthropic.