As AI chipmaker Cerebras prepares for an eventual IPO, it appears the company has landed a key cloud computing customer. oracle.
On a conference call with analysts Tuesday after Oracle’s quarterly results, of Two CEOs of software vendors indicated that their infrastructure includes Cerebras chips in addition to the market leader’s graphics processing units (GPUs). Nvidia and rival advanced micro device.
“We are building an infrastructure that is flexible, fungible, and capable of supporting the smallest to largest workloads,” Magwirk said. “We continually deliver the latest accelerators, from the latest Nvidia and AMD options to new designs from companies like Cerebras and Positron,” said another AI hardware startup.
Cerebras offers cloud services powered by large-scale WSE-3 chips. The company filed paperwork for an IPO in 2024, but withdrew the application last October. Days later, the company announced a $1.1 billion funding round at a valuation of $8.1 billion, and CEO Andrew Feldman said Cerebra still intended to go public.
For prospective investors, one of the most obvious concerns in Cerebras’ original prospectus was its reliance on a single customer based in the Middle East. Microsoft-backed G42 is headquartered in Abu Dhabi, United Arab Emirates and accounted for 87% of Cerebras’ revenue in the first half of 2024.
Bolstering its customer roster with a name like Oracle could be a huge boon for Cerebras, and it comes on the heels of another major announcement earlier this year. In January, Cerebras announced a $10 billion investment from OpenAI, which relies on companies like Oracle for cloud services. The following month, OpenAI announced that it would be collaborating with Cerebras on a research preview of Codex-Spark, a turnkey AI model for software development, for ChatGPT Pro customers.
Oracle did not immediately respond to a request for comment, and its price list does not list the Cerebras option. Cerebras had no immediate comment.
Oracle’s earnings release came after the company reported better-than-expected results, raised its outlook for fiscal year 2027 and announced that its remaining performance obligations more than quadrupled from a year ago to $553 billion.
“Overall, we believe that the investments we are making now in data centers, computing power and customer relationships will become even more valuable over time,” Magouik said, citing Cerebras and other chipmakers.
As a startup trying to compete with the world’s most valuable companies, Cerebras is playing in a market that appears to have an insatiable demand for computing power as AI model developers scale up to quickly respond to user needs.
Nvidia is using its huge cash pile to expand into new product areas. In December, the company acquired major assets from AI chip startup Groq for about $20 billion. Nvidia plans to announce a new architecture using Groq at the GTC developer conference in California next week, the Wall Street Journal reported.
Magouik said on the conference call that GTC plans to make several “important announcements.” He also said that in addition to strategically located data centers, innovative technology is also needed to quickly respond to incoming requests.
“This is the type of hardware that is being deployed, and that’s why we’re seeing so much innovation happening around these AI accelerators,” he said. “If you look at what Groq, Cerebras, and Positron are doing, all different types of customers are saying, not only how can we reduce the cost of inference, but how can we significantly reduce the latency of inference?”
Spotlight: OpenAI unveils first AI model running on Cerebras chips
