CBRS stock is trading at 90.2 times trailing-12-month sales, a valuation built on OpenAI-linked demand and a $24.6 billion backlog. That leaves investors judging Cerebras Systems on whether promised compute turns into delivered capacity and booked revenue, not on the size of the order book alone.
90.2 times trailing-12-month sales is a rich multiple for any chipmaker, and Cerebras’ 2025 results help explain why the market is still assigning it such a premium. Revenue rose 75.7% to $510 million, while cloud and other services revenue climbed 93.6% to $151.6 million, showing that the company’s growth has already started shifting beyond its core hardware line.
OpenAI’s January deal
750 megawatts of high-speed compute capacity is the clearest sign of why OpenAI sits at the center of the Cerebras story. In January 2026, OpenAI agreed to add that capacity to its platform in a deal valued at over $10 billion through 2028, and the newer commitments reportedly doubled the earlier agreement. For Cerebras, that makes the backlog less like a static sales figure and more like a delivery schedule tied to a single demanding customer.
About $1 billion in loan support from OpenAI also gives Cerebras room to scale manufacturing and build data centers for AI workloads. The deal included warrants that OpenAI can convert into a minority stake in Cerebras Systems, adding another layer to the relationship without changing the immediate task: turning capacity promises into revenue that can be recognized on time.
Backlog versus sales multiple
$24.6 billion in remaining performance obligations at the end of 2025 gives Cerebras a large base of contracted demand to work through. later reported that OpenAI could spend more than $20 billion over three years on servers powered by Cerebras chips, which is close enough to the backlog scale to show how concentrated the growth story has become around one customer relationship.
One practical way to read the numbers is straightforward: if a large share of the $24.6 billion backlog is tied to OpenAI, then execution risk sits in whether Cerebras can convert compute commitments into shipped capacity and recognized sales without delay. The market is paying for that conversion now, at 90.2 times sales, before the revenue has fully landed.
Amazon Bedrock in AWS
Amazon has also partnered with Cerebras to make fast AI inference services available through Amazon Bedrock in AWS data centers, giving the company another route to commercialization. Inference has become the more important test case for AI infrastructure, because customers want models to respond quickly while handling reasoning-heavy work at scale, and that makes actual throughput more important than headline demand.
For CBRS stock, the near-term question is simple: how much of that backlog turns into booked business on schedule, and how much depends on the pace of manufacturing, data-center buildout, and customer rollout. The valuation already assumes a lot of that conversion is coming; the filings and the OpenAI deal now have to show it.
