Cerebras Lifts 2026 Outlook, Sees Revenue Tripling in 2027

Cerebras Systems posted record third-quarter revenue and raised its full-year outlook in its first earnings season as a listed company, reporting revenue of around $211 million for the quarter ended 30 September, up 55% from a year earlier, and a core measure up 70% to around $231 million. The Sunnyvale, California-based chip designer said on 12 November 2026 that its AI cloud business drove the gain while hardware sales fell, and chief executive Andrew Feldman said he expects revenue to triple in 2027. Cerebras listed on Nasdaq under the ticker CBRS this year.

Context — why the AI inference shift matters now

The quarter matters because it is the first set of numbers Cerebras has published as a public company, and it arrived alongside beats and raised guidance from two adjacent names in the AI supply chain. Applied Materials beat on its fourth quarter and guided above forecasts as chip equipment demand held. Cisco beat and lifted its fiscal 2027 view, saying AI revenue from hyperscalers is set to more than double.

Cerebras's own comparable is the year-earlier quarter, when hardware revenue was about $97 million. That line has now fallen to around $69 million, while cloud and other services revenue more than tripled to around $142 million. The mix has inverted in twelve months.

The catalyst is structural rather than cyclical. Cerebras sells wafer-scale chips — a single processor roughly the size of a dinner plate, cut from an entire silicon wafer rather than diced into many smaller chips — and says the design is especially fast at inference, the work of running trained models to answer queries. It still sells those systems outright, but increasingly rents out their computing power through its own cloud service.

The macro backdrop is a market rewarding AI infrastructure capacity over AI training narratives. In the live data at 22:19 UTC today, Nvidia traded at $166.19, down 1.60%, while Cisco traded at $102.01, up 3.93%.

Data — what the numbers show

The quarter's revenue lines moved in opposite directions. Cloud and other services revenue more than tripled to around $142 million. Hardware revenue fell to around $69 million, from about $97 million a year earlier.

Metric	Q3 FY2026	Year-earlier quarter
Total revenue	~$211m	~$136m implied
Core revenue	~$231m	—
Cloud and other services	~$142m	—
Hardware	~$69m	~$97m
GAAP net loss	~$161m (67 cents a share)	—
Core net income	~$30m	—

Profitability splits sharply. Cerebras reported a GAAP net loss of around $161 million, or 67 cents a share, but posted core net income of around $30 million. The gap between the two measures will be scrutinised by investors still pricing a recent IPO.

Guidance frames the rest. For the fourth quarter, Cerebras expects core revenue of $310 million to $330 million, core gross margin of 47% to 49% and an operating margin of around breakeven. For the full year, it expects core revenue of $940 million to $960 million.

The balance sheet and pipeline are the headline support. Cerebras reported a backlog of around $26 billion and about $7.8 billion of cash and short-term investments. New deals include an Oracle cloud integration due in 2027 and two hyperscaler wins.

The peer comparison sits in the tape. Nvidia traded at $166.19 within a range of $148.80 to $167.14, while Cisco traded at $102.01 within $99.41 to $103.18.

Analysis — what it means for markets and sectors

Cerebras is turning from a wafer-scale chip seller into an AI cloud provider, and it now has to deliver on both. The second-order read is that demand for inference is spreading beyond Nvidia to specialist designers, which supports the argument that AI infrastructure spending is shifting from training models to running them.

Sector exposure follows the backlog. AI cloud capacity providers and data centre power names may draw support from the scale of the $26 billion pipeline and the associated capacity build-out, because converting that backlog into revenue requires physical infrastructure, not just silicon.

The limitation is concentration. Cerebras relies on a small number of large customers, including OpenAI. To turn the backlog into revenue it also has to deliver about 700 megawatts of data centre capacity by the end of 2027 — a construction and power-procurement problem as much as a chip problem.

The counter-argument is the GAAP line. A $161 million net loss alongside $30 million of core net income means the reported earnings and the adjusted earnings tell different stories, and investors pricing a recent IPO tend to trade that gap violently in the first few quarters.

Positioning is split accordingly. Growth money is underwriting the cloud transition and the 2027 triple, while the flow into Cisco, up 3.93% today, suggests some investors prefer the diversified networking exposure to the single-name inference bet.

Outlook — what to watch next

The next hard catalyst is fourth-quarter results, against guidance of $310 million to $330 million in core revenue, 47% to 49% core gross margin and operating margin around breakeven. Whether Cerebras hits the top or bottom of that band sets the base for the 2027 triple.

The Oracle cloud integration is due in 2027 and is the first dated milestone on the pipeline. Progress on the roughly 700 megawatts of data centre capacity by the end of 2027 is the operational gate between backlog and revenue.

The two hyperscaler wins have no disclosed terms, so the market has no visibility on their value or timing.

On the tape, Nvidia's intraday range of $148.80 to $167.14 is the level to watch for sentiment on the incumbent, while Cisco's move through $102.01 frames how the market is pricing AI networking against AI compute.

Frequently Asked Questions

What does Cerebras's core revenue measure exclude?

Cerebras reports a preferred "core" measure alongside GAAP results. In the third quarter, core revenue was around $231 million against GAAP revenue of around $211 million, and core net income was around $30 million against a GAAP net loss of around $161 million. The company did not disclose in its release which specific items reconcile the two figures, so the exact composition of the adjustment is not public.

Why did Cerebras hardware revenue fall while total revenue rose?

Hardware revenue fell to around $69 million from about $97 million a year earlier, even as total revenue rose 55%. The offset came from cloud and other services, which more than tripled to around $142 million. That reflects a deliberate shift: Cerebras still sells wafer-scale systems outright, but increasingly rents their computing power through its own cloud service, converting one-time hardware sales into recurring service revenue.

What has to happen for Cerebras to hit its 2027 target?

Chief executive Andrew Feldman said he expects revenue to triple in 2027. Two conditions sit behind that: converting the roughly $26 billion backlog, and delivering about 700 megawatts of data centre capacity by the end of 2027. The company also depends on a small number of large customers, including OpenAI, so customer concentration is a live risk to the target.

Bottom Line

Cerebras has the backlog and the cash; the 2027 triple now rests on data centre delivery and customer concentration.

Disclaimer: This article is for informational purposes only and does not constitute investment advice. CFD trading carries high risk of capital loss.