Per a Tuesday (March 10) blog post, the agreement includes plans for Thinking Machines Lab to deploy at least 1 gigawatt of Nvidia chips for AI training and inference. The companies will also collaborate on designing artificial intelligence training and serving systems built on Nvidia technology, according to reporting by The Wall Street Journal.
The size and structure of Nvidia’s investment have not been made public. The partnership provides the young AI lab with access to the computing infrastructure required to develop and operate large-scale AI models, and deepens its existing use of Nvidia hardware.
Nvidia’s technology “is the foundation on which the entire field is built,” Thinking Machines Co-founder and CEO Murati said in the announcement. “This partnership accelerates our capacity to build AI that people can shape and make their own, as it shapes human potential in turn.”
“Thinking Machines has brought together a world-class team to advance the frontier of AI,” Nvidia CEO Jensen Huang said in the announcement.
Murati founded Thinking Machines after she left OpenAI, where she had served as chief technology officer and helped oversee the development of several of the company’s major AI systems. She launched the new research lab with a group of former colleagues as the industry sees the emergence of new independent AI labs competing to build frontier models.
Murati has said the company’s larger goal is to build AI systems that work alongside humans rather than operating independently.
The company has also been recruiting aggressively as it builds out its research team. According to The Information, the company has already seen some turnover within its leadership. Co-founder Andrew Tulloch left the startup in October and is returning to Meta, where he worked for 13 years before leaving in 2023 to join OpenAI.
Despite the high-profile founding team and strong investor backing, Thinking Machines Lab has revealed little publicly regarding its long-term strategy or commercialization plans. So far, the company has released only one product: a training API called Tinker, unveiled in October 2025, to help researchers and organizations train and fine-tune AI models.
For all PYMNTS AI coverage, subscribe to the daily AI Newsletter.