(Reuters) -Anthropic said on Thursday its Claude models would be trained using up to one million of Google’s artificial intelligence chips, worth tens of billions of dollars, as it aims to improve its generative AI offerings in a fast-moving landscape.
The deal underscores the immense computing demands of AI firms for training, deployment and continuous inference in GenAI.
To secure infrastructure swiftly, developers have been rushing into multi-billion-dollar agreements.
Alphabet’s Google, also a backer of Anthropic, will provide additional cloud computing services to the startup behind the Claude chatbot.
For Google, the deal comes as it is expanding external availability of its in-house tensor processing units, or TPUs, which were historically reserved for internal use. It rents the TPUs through Google Cloud.
Anthropic said it chose the TPUs due to their price-performance and efficiency and its existing experience in training and serving its models with the processors.
Reuters exclusively reported earlier in October that Anthropic is projecting to more than double and potentially nearly triple its annualized revenue run rate next year, fueled by the rapid adoption of its enterprise products.
(Reporting by Juby Babu in Mexico City and Zaheer Kachwala in Bengaluru; Editing by Shilpi Majumdar, Anil D’Silva and Alan Barona)
Comments