EXCLUSIVE: Google is deepening ties with Thinking Machines Lab through new multibillion-dollar deal

Former OpenAI executive Thinking Machines Lab has signed a new multibillion-dollar agreement to expand its use of Google Cloud’s AI infrastructure, including systems powered by Nvidia’s latest GPUs, TechCrunch has learned exclusively.

The deal is worth billions, according to a source familiar with the matter, and includes access to Google’s latest AI systems built on Nvidia’s new GB300 chips, along with infrastructure services to support model training and deployment.

Google is actively working on a number of cloud deals with AI developers as it aims to bundle its cloud offerings with other services such as storage, the Kubernetes engine, and Spanner, its database product. Earlier this month, Anthropic signed an agreement with Google and Broadcom for multi-gigawatt capacity of tensor processing units (TPUs) (Google’s specially designed AI chips for machine learning workloads).

But the competition is fierce. And just this week, Anthropic also signed a new agreement with Amazon to secure up to 5 gigawatts of capacity to train and deploy Cloud.

Earlier this year, Thinking Machines partnered with Nvidia in a deal that included an investment from the chip maker. But this is the first time the lab has struck a deal with a cloud service provider. The deal isn’t exclusive, so Thinking Machines may use multiple cloud providers over time, but it’s still a sign that Google is looking to secure its fast-growing frontier labs early.

Moratti left her job as CTO at OpenAI and founded Thinking Machines in February 2025. The company, which shortly thereafter raised a $2 billion seed round at a $12 billion valuation, has remained top secret, but it launched its first product in October. Dubbed Tinker, it’s a tool that automates the creation of custom frontier AI models.

Wednesday’s deal provided some insight into what Thinking Machines is developing. In a press release, Google noted that it can support the startup’s reinforcement learning workloads, which the Tinker architecture is based on. Reinforcement learning is a training approach that has fueled recent breakthroughs in labs, including DeepMind and OpenAI, and the size of the Google Cloud deal reflects how computationally expensive this work can get.

Thinking Machines is one of the first Google Cloud customers to gain access to its 300GB systems, which deliver a 2x improvement in training and service speed compared to previous generation GPUs, according to Google.

“Google Cloud has got us up and running at record speed with the reliability we demand,” Mile Ott, founding researcher at Thinking Machines, said in a statement.