Cerebras Raises 2026 Revenue Forecast to $960 Million as OpenAI Deal Fuels 70% Sales Jump

admin · November 12, 2026 · 6 min read

Illustration of a giant glowing wafer-scale AI computer chip inside a modern data center, for Cerebras Systems

Cerebras Raises 2026 Revenue Forecast to $960 Million as OpenAI Deal Fuels 70% Sales Jump

Cerebras Systems raised its 2026 revenue forecast on Thursday after third-quarter sales jumped 70 percent, powered by a fast-growing computing deal with OpenAI, and its shares surged in extended trading.

The Sunnyvale, California-based AI chipmaker said it now expects 2026 adjusted revenue of $940 million to $960 million, up from its previous forecast of $880 million to $890 million. Reuters reported that the shares rose nearly 18 percent in extended trading after the results were released following the close of the U.S. market on Thursday, November 12.

The report makes Cerebras one of the clearest winners yet of the AI infrastructure boom — a smaller challenger growing rapidly by selling computing capacity to the world’s most demanding AI labs, in a market long dominated by Nvidia.

A quarter driven by OpenAI

Cerebras posted third-quarter adjusted core revenue of $230.9 million, up 70 percent from a year earlier and ahead of analysts’ average estimate of $214.7 million, according to data compiled by LSEG. On a GAAP basis, revenue was $210.8 million, up 55 percent.

The standout figure was in cloud and services. Adjusted core cloud and services revenue — the line that reflects the OpenAI agreement — more than tripled to $144.9 million in the quarter. Reuters said the company is scaling chip volumes for a $20 billion multi-year agreement with OpenAI, and that relationship is a big reason it reported $26.1 billion of remaining performance obligations as of September 30. That backlog figure represents work customers have committed to buy, though Cerebras can only count it as revenue once it delivers the computing capacity.

Profitability remains a work in progress. Cerebras reported adjusted earnings of 13 cents a share, where analysts had expected a loss of 13 cents, while its GAAP net loss was $161.2 million. The net loss widened to 67 cents per share from 41 cents in the same quarter last year.

Background: the wafer-scale challenger

Cerebras builds its AI systems around a wafer-scale engine — a processor made from a single, entire silicon wafer rather than many small chips cut from one. The design is meant to keep more of the chip’s computing power and memory in one place, reducing the time data spends moving between components, which matters for training and running very large AI models.

The company has long pitched itself as an alternative to Nvidia’s graphics processors for AI workloads, but its commercial breakthrough has come through cloud computing rather than chip sales alone. By building data centers filled with its own systems and renting out the capacity, Cerebras has turned itself into a supplier of AI computing power, with OpenAI as its anchor customer.

That strategy is also what makes the OpenAI deal so important to investors. The agreement, viewed as key to justifying Cerebras’ valuation, underpins the $26.1 billion backlog the company reported on Thursday. The question now is whether Cerebras can build out capacity fast enough to convert that committed work into revenue at the pace its raised forecast implies.

Why this matters

For the AI industry, Cerebras’ quarter is another sign that demand for computing capacity is still outrunning supply. OpenAI, Google, Anthropic and other labs are racing to secure the chips and data-center capacity needed to train bigger models and run AI agents, and suppliers that can deliver are being rewarded with rapid growth.

For investors, the results land in a week already full of AI news. Google launched its new Gemini 4 Hydrogen model on Thursday, and Nvidia — the dominant AI chipmaker — reports earnings on Tuesday, November 17. Anthropic withdrew IPO pricing details from banks earlier this month after a filing valuing the company at nearly $70 billion circulated, a reminder of how much money is now chasing the AI buildout. Read more of our reporting in the AI section.

Cerebras is still far smaller than Nvidia, and its losses are widening even as sales grow. But Thursday’s report shows a company converting a single giant customer relationship into a rapidly growing business — and Wall Street, at least for one evening, was willing to pay up for it.

Latest developments

Cerebras released its third-quarter results after the U.S. market closed on Thursday, November 12, and its earnings call was still under way as the first reports of the numbers circulated. The company raised its full-year adjusted revenue forecast to $940 million to $960 million and reported $26.1 billion in remaining performance obligations as of September 30.

Frequently Asked Questions

What did Cerebras report for the third quarter?

Cerebras reported adjusted core revenue of $230.9 million, up 70 percent from a year earlier and above analysts’ average estimate of $214.7 million. GAAP revenue was $210.8 million, up 55 percent, and adjusted earnings were 13 cents a share, against an expected loss of 13 cents.

Why is the OpenAI deal so important to Cerebras?

Reuters said Cerebras is scaling chip volumes for a $20 billion multi-year agreement with OpenAI. The deal drove cloud and services revenue to more than triple in the quarter and underpins $26.1 billion in remaining performance obligations — work customers have committed to buy — as of September 30.

How did Cerebras shares react?

Reuters reported that the shares rose nearly 18 percent in extended trading on Thursday evening, after the company raised its 2026 revenue forecast to $940 million to $960 million from $880 million to $890 million.

What is a wafer-scale engine?

Cerebras builds its AI processors from a single, entire silicon wafer instead of cutting a wafer into many small chips. The design keeps more computing power and memory in one place, which the company says speeds up training and running very large AI models.

Final paragraph

Cerebras’ raised forecast and 70 percent sales jump show how much of the AI boom is now flowing to suppliers beyond Nvidia — at least to those with a major customer locked in. The company’s challenge is execution: turning a $26.1 billion backlog into delivered computing capacity, and doing it while losses widen. Its next reports will show whether Thursday’s surge was the start of a durable growth story or a single quarter powered by one deal.

Sources: