Cerebras Systems NASDAQ: CBRS and OpenAI executives used a Paris technology event to outline the companies’ expanding infrastructure partnership, emphasizing faster AI inference, enterprise adoption of agents and a new European data center build-out.
Andrew Feldman, CEO of Cerebras, said the companies’ collaboration began after OpenAI identified fast inference as a key requirement as AI models became more widely useful in workplace applications. Feldman said OpenAI CEO Sam Altman contacted him in the summer of 2025 to discuss the need for faster inference, leading to what Feldman described as “one of the largest deals in Silicon Valley history.” He said the agreement was “north of $20 billion” over several years for compute capacity.
Sachin Katti, OpenAI’s Head of Industrial Compute, said the company increasingly views latency as a critical product feature as AI becomes part of daily work. He compared the shift to the evolution of internet search, where quality came first but speed later became central to user growth and revenue.
“Latency is a very critical product ingredient for us going forward,” Katti said. He added that OpenAI’s Phi-6 model will be available on Cerebras and said it is “the only frontier model” expected to run at 750 tokens per second. Katti described that speed as “probably an order of magnitude faster than anything else that’s out there.”
Executives Say Speed Will Drive AI Usage
Feldman said faster AI responses are essential if AI tools are to become embedded in enterprise workflows. He argued that there is no meaningful market for “slow search” or “dial-up internet,” and said the same expectations will apply to AI systems used throughout the workday.
“If you give people fast tools, they use them more often, they enjoy using them, and they use them on harder and more interesting problems,” Feldman said.
Katti said OpenAI is seeing broader use of Codex internally, beyond software engineering. He said Codex has become “the default user interface” at OpenAI, with employees in legal, go-to-market, finance and other functions using it for increasingly complex tasks. He said OpenAI employees even use Codex to interact with browsers because of its computer-use capabilities.
Katti said tasks with measurable outputs are especially well suited for agents, because the systems can iterate toward better results. As an example of how far usage has spread internally, he said OpenAI’s human resources department built an agent for human reorganizations, calling reorgs “very complex topics.”
Productivity, Not Token Counts, Seen as Key Metric
The executives also addressed how enterprises should measure AI adoption. Katti said OpenAI is already seeing company-level productivity gains, pointing to the pace of model releases. He said OpenAI is now releasing a new model every month and attributed the faster pace in part to Codex.
“Previously, AI research was human limited, fundamentally,” Katti said. “We are increasingly getting to the point where recursion begins to become real, where AI is going to help, if not do, the AI research itself.”
Feldman cautioned against using token consumption alone as a measure of AI maturity. He said enterprises should instead focus on business metrics and productivity outcomes.
“I don’t think you should count your tokens as a measure of how AI forward you are,” Feldman said. “I think we’re building AIs to do work. You should count the productivity of the work.”
Infrastructure Bottlenecks Remain a Major Focus
Katti said rising use of agentic AI is increasing demand across the technology stack, including CPUs, GPUs, networking, storage and memory. He said OpenAI is “hunting for supply wherever we can get it” and also facing the challenge of finding data centers to house the infrastructure.
He said there is no “silver bullet” for resolving those bottlenecks, but that software optimization and efficiency are becoming more important as AI scales.
“We’ve been in this phase in AI where we are going quickly to new products and new models, it’s all been about time to market,” Katti said. “We are now getting to the point where AI is scaling, efficiency becomes important, too.”
Cerebras Announces European Data Center Expansion
Feldman said Europe is a key market because of strong demand for advanced AI and more token capacity. He announced that Cerebras is building 200 megawatts of data center capacity in Europe, including sites in Lyon, France, Norway and Finland.
Feldman said the 200 megawatts of capacity would be completed by the end of next year, with some delivered this year. He said much of the capacity is intended to meet OpenAI’s needs and that Cerebras is deploying “billions of dollars of capital” in data center development.
“We anticipate many more big scale deployments and big data centers here,” Feldman said.
The executives also tied the infrastructure build-out to the growing discussion around sovereign AI. Feldman said AI infrastructure is increasingly viewed as a “critical national resource,” while Katti called data centers “the factories of our age” and “intelligence factories.”
Next 12 Months Expected to Bring Faster Change
Looking ahead, both executives said they expect the pace of AI development to continue accelerating. Feldman noted that 12 months earlier Cerebras was still private and had “$25 billion less in sales,” adding that the market had advanced faster than expected.
Katti said “12 months is an eternity in AI” and that he could not predict what will happen even over the next three months. Still, he said the “one constant” is likely to be an accelerating pace of change, with model capabilities continuing to improve quickly.
“The bigger question will be how quickly can these capabilities be adopted for the real world, for enterprise usage, for whatever consumer usage,” Katti said.
About Cerebras Systems NASDAQ: CBRS
Cerebras Systems is a technology company focused on building artificial intelligence infrastructure, including hardware and software designed to accelerate deep learning and large-scale AI workloads. The company is best known for its wafer-scale processor architecture, which is intended to provide high-performance compute for training and inference applications.
In addition to its AI chips, Cerebras offers systems and related software tools that support researchers and enterprises working with machine learning models.
This instant news alert was generated by narrative science technology and financial data from MarketBeat in order to provide readers with the fastest reporting and unbiased coverage. Please send any questions or comments about this story to contact@marketbeat.com.
Before you consider Cerebras Systems, you'll want to hear this.
MarketBeat keeps track of Wall Street's top-rated and best performing research analysts and the stocks they recommend to their clients on a daily basis. MarketBeat has identified the five stocks that top analysts are quietly whispering to their clients to buy now before the broader market catches on... and Cerebras Systems wasn't on the list.
While Cerebras Systems currently has a Moderate Buy rating among analysts, top-rated analysts believe these five stocks are better buys.
View The Five Stocks Here
Learn the basics of options trading and how to use them to boost returns and manage risk with this free report from MarketBeat. Click the link below to get your free copy.
Get This Free Report
