Google Invests Billions in Thinking Machines Lab Partnership
▼ Summary
– Thinking Machines Lab, founded by former OpenAI executive Mira Murati, has signed a multi-billion-dollar deal to expand its use of Google Cloud’s AI infrastructure, including systems with Nvidia’s latest GB300 chips.
– The agreement, valued in the single-digit billions, provides access to Google’s latest AI systems and infrastructure services for model training and deployment.
– This is Thinking Machines’ first deal with a cloud services provider, following an earlier partnership and investment from Nvidia, and it is not exclusive to Google.
– The company, which launched its first product called Tinker in October, uses reinforcement learning, a computationally expensive training approach that Google’s infrastructure will support.
– Google is actively securing cloud deals with AI developers like Thinking Machines and Anthropic to integrate its services and lock in fast-growing AI labs early.
In a major move for the AI infrastructure race, Google Cloud has secured a multi-billion dollar partnership with Thinking Machines Lab, the secretive startup founded by former OpenAI executive Mira Murati. The agreement grants the lab extensive access to Google’s latest AI systems, which are powered by Nvidia’s new GB300 chips, alongside the cloud services needed for advanced model training and deployment. This deal, valued in the single-digit billions, marks a significant commitment from both parties as competition for hosting frontier AI labs intensifies.
This partnership represents a strategic win for Google as it bundles its cloud offerings with other services to attract leading AI developers. The company has been actively pursuing such agreements, evidenced by a separate deal with Anthropic earlier this month for access to multiple gigawatts of capacity on Google’s custom tensor processing units (TPUs). However, the cloud market remains fiercely contested. Just this week, Anthropic also signed a new agreement with Amazon Web Services for up to 5 gigawatts of capacity to support its Claude models, highlighting the high-stakes battle for compute resources.
For Thinking Machines, this is its first publicly disclosed contract with a major cloud provider, though the arrangement is not exclusive. The lab had previously partnered with Nvidia, which included an investment from the chipmaker. Securing this level of infrastructure from Google so early in its development signals the lab’s ambitious scaling plans and Google’s strategy to lock in relationships with fast-growing AI research entities.
Founded by Murati in February 2025 after she left her role as OpenAI’s chief technologist, Thinking Machines quickly raised a $2 billion seed round at a $12 billion valuation. The company launched its first product, an automation tool for creating custom AI models called Tinker, last October. The scale of the Google Cloud deal offers a rare glimpse into the lab’s technical direction. Google noted its infrastructure will support the startup’s reinforcement learning workloads, a computationally intensive training approach that is foundational to Tinker’s architecture and has driven major breakthroughs at labs like DeepMind and OpenAI.
As one of the first customers for Google’s GB300-powered systems, Thinking Machines will leverage hardware that Google claims offers a twofold improvement in training and serving speed over previous GPU generations. “Google Cloud got us running at record speed with the reliability we demand,” said Myle Ott, a founding researcher at Thinking Machines. This access to cutting-edge compute is becoming a critical differentiator as AI models grow more complex and expensive to develop.
(Source: TechCrunch)
