Google Announces Multi-Billion Dollar Infrastructure Deal with Thinking Machines Lab

Mira Murati's Thinking Machines Lab, founded in February 2025 after she left OpenAI, has signed a single-digit billion-dollar deal with Google Cloud for AI infrastructure powered by Nvidia's GB300 chips, making it among the first customers to access these systems offering 2X speed improvements over prior GPUs. The non-exclusive agreement supports the startup's reinforcement learning workloads needed for Tinker, its frontier AI model automation tool launched in October. The deal represents Thinking Machines Lab's first cloud services partnership since raising a $2 billion seed round at a $12 billion valuation, and follows similar massive infrastructure agreements by competitors like Anthropic with Google and Amazon.

Google Cloud's Multi-Billion Dollar Deal with Thinking Machines Lab

• ▪The Thinking Machines Lab deal with Google Cloud is not exclusive
• ▪The Thinking Machines Lab deal with Google Cloud includes infrastructure services to support model training and deployment
• ▪The Thinking Machines Lab deal with Google Cloud includes access to AI systems built on Nvidia's GB300 chips
• ▪Google Cloud's GB300-powered systems offer a 2X improvement in training and serving speed compared to prior-generation GPUs
• ▪Thinking Machines Lab is among the first Google Cloud customers to access GB300-powered systems
• ▪Google Cloud can support Thinking Machines Lab's reinforcement learning workloads
• ▪Thinking Machines Lab signed a multi-billion-dollar agreement with Google Cloud to expand its use of Google Cloud's AI infrastructure

Competitive Landscape of Cloud Infrastructure for AI Companies

• ▪Thinking Machines Lab partnered with Nvidia in a deal that included an investment from Nvidia earlier this year
• ▪The Google Cloud deal is the first time Thinking Machines Lab has struck a deal with a cloud services provider
• ▪Anthropic signed an agreement with Amazon to secure up to 5 gigawatts of capacity for training and deploying Claude

Thinking Machines Lab's Development and Product Strategy

• ▪Myle Ott is a founding researcher at Thinking Machines Lab
• ▪Thinking Machines Lab launched its first product in October
• ▪Mira Murati founded Thinking Machines Lab in February 2025
• ▪Thinking Machines Lab raised a $2 billion seed round at a $12 billion valuation

Perspective of Google Cloud

• ▪Google Cloud sees providing early access to Nvidia GB300-powered systems as a way to attract high-profile AI startups like Thinking Machines Lab
• ▪Google Cloud views bundling storage, Kubernetes engine, and Spanner database services with AI infrastructure as a competitive strategy against rival cloud providers

Perspective of Thinking Machines Lab

• ▪Thinking Machines Lab chose a non-exclusive agreement with Google Cloud to maintain flexibility in its infrastructure partnerships
• ▪Thinking Machines Lab requires cloud infrastructure specifically optimized for reinforcement learning workloads to support Tinker's architecture

Perspective of Nvidia

• ▪Nvidia's earlier investment in Thinking Machines Lab is complemented by the startup's adoption of GB300-powered infrastructure through Google Cloud
• ▪Nvidia's GB300 chips are being adopted by major AI startups like Thinking Machines Lab for next-generation model training infrastructure