SAN FRANCISCO, Sept 28 (Reuters) - Cerebras Systems plans to supply AI chips and hardware with capacity to consume roughly 100 megawatts to startup Gimlet Labs to help run its cloud computing business, the companies said on Monday.
The deal between Gimlet and Cerebras is part of rising interest from AI companies in procuring hardware that is capable of speedy AI calculations known as inference, the process in which a chatbot such as Anthropic's Claude generates a response to a user question.
Performing inference operations speedily can help in areas such as cybersecurity, voice applications and financial analysis, Gimlet CEO Zain Asgar said in an interview with Reuters. For this hardware, Gimlet plans to target selling to companies and startups that are building their entire products around AI in some way, Asgar said.
Gimlet plans to use the Cerebras hardware to run so-called frontier models — the cutting-edge AI models — that require enormous amounts of computing power to run speedily.
Cerebras plans to supply its CS-4 systems, which it unveiled over the summer, over one to two years, its CEO, Andrew Feldman, said. Gimlet plans to make the CS-4 hardware available in its cloud in 2027, the company said.
The deal is "further validation from Cerebras ... of how easy it is to deploy, even in heterogeneous environments," Feldman said, referring to installing the hardware alongside chips and systems from other companies.
Cerebras and Gimlet declined to disclose the financial terms of the deal. Gimlet will be responsible for maintaining and operating the systems once delivered by Cerebras.
OpenAI and companies such as Nvidia have recognized the importance of the ability to perform speedy inference. Earlier this year, Cerebras signed a deal to supply OpenAI with chips. And last year, Nvidia signed a licensing deal with Groq for the company's chips and hardware.
(Reporting by Max A. Cherney in San Francisco; Editing by Cynthia Osterman)
