CBRS · Cerebras Systems

The thesis

Cerebras Systems is modestly more likely than not to beat the S&P 500 over the next three months. On 12 November 2026 it guided fourth-quarter core revenue (its adjusted revenue measure) to $310–330 million, against a $271.1 million consensus cited by Benzinga. On the same day it named Microsoft as a new cloud customer. The fourth-quarter report, expected in late February 2027, shows whether that revenue arrives at the guided 47%–49% core gross margin.

The rating is Medium because the guidance surprise is large but the doubts are concrete. Five of the six firms that published targets on 13 November 2026 cut them. CNBC reported on 12 November 2026 that management is weighing financing options to meet demand. Generally accepted accounting principles (GAAP) gross margin was 20.0% in the September 2026 quarter, down from 40.6% a year earlier. Most of the backlog rests on one customer, OpenAI.

The shares closed at $108.72 on 13 November 2026, up from $97.75 on 12 November 2026. That is still well under the $185 IPO price of May 2026 and the 50-day average of $151.02. The case rests on delivery against the backlog, not on the post-report gain.

Bullish and bearish views on Cerebras Systems Inc.

The bull case on Cerebras Systems Inc. (CBRS), in brief: fourth-quarter guide well above consensus. The bear case: one customer carries the backlog.

Bull case

• Fourth-quarter guide well above consensus. On 12 November 2026 the company guided fourth-quarter core revenue to $310–330 million, against a $271.1 million estimate. CNBC reported that the guided 47%–49% core gross margin compared with 42.1% expected.
• Full-year outlook raised a second time. The 2026 core revenue range moved to $940–960 million from $880–890 million on 12 November 2026, against an $887.0 million estimate. The company had already raised it from $855–865 million in August.
• Microsoft joins as a cloud customer. CEO Andrew Feldman named Microsoft as the top-three hyperscaler in the 12 November 2026 release, CNBC reported. The deal was signed in October, and Feldman said its revenue begins to appear in the first quarter of 2027.
• Cloud revenue is scaling fast. Third-quarter GAAP cloud and services revenue was $141.9 million, up 265% from $38.9 million a year earlier, per the 12 November 2026 release. Core revenue grew 70% to $230.9 million.
• Backlog and cash are large. The 10-Q filed 12 November 2026 shows backlog of $26,100 million at 30 September 2026, cash and short-term investments of $7,361 million and debt of $829 million. Oracle plans to run Cerebras inference in Oracle Cloud Infrastructure from 2027.

Bear case

• One customer carries the backlog. MarketScreener reported on 13 August 2026 that growth and valuation rest on a multiyear $20 billion OpenAI agreement. The 12 November 2026 release flags OpenAI's Master Relationship Agreement among its risk factors.
• GAAP results missed and margins fell. Third-quarter GAAP revenue of $210.8 million missed the $214.7 million estimate cited by Benzinga. The net loss was $161 million in the quarter to 30 September 2026, and operating margin was −106.3%, per the 10-Q.
• Analysts cut targets after the report. On 13 November 2026 Citi cut its target to $250 from $320, Mizuho to $240 from $300 and UBS to $155 from $250, per StockAnalysis and Benzinga; Needham cut to $200. Wedbush raised to $205 from $190 and Barclays kept $280.
• The 2027 goal sits under estimates. CNBC reported on 12 November 2026 that tripling revenue in 2027 implies about $2.85 billion, against $2.92 billion expected by LSEG analysts. Feldman also said the company is evaluating financing options.
• Share supply is rising. Diluted share count rose 342.5% from a year earlier in the quarter to 30 September 2026, per the 10-Q. Under the S-1 terms reported by Seeking Alpha, the main lock-up frees roughly 171 million shares around the third-quarter report; a common market view is that such releases weigh on newly listed shares.
• Insiders sold into the decline. SEC Form 4 filings show the chief operating officer selling 396,000 shares at an average $204.04 on 29 September 2026 and the chief technology officer selling 60,000 at $115.79 on 27 October 2026, both under Rule 10b5-1 trading plans set in advance.

Price setup

The shares closed at $97.75 on 12 November 2026, before the report, and at $108.72 on 13 November 2026. CNBC reported them up about 7% late that morning, so the gain widened into the close. Adjusted daily closes show a 56.9% fall over three months to 13 November 2026.

The 50-day average stood at $151.02 on 13 November 2026, far above the price. The $185 IPO price and the $204–209 range where executives sold in late September 2026 mark where the shares traded before the decline. The 13 November 2026 close sits just under the $115.79 average of the chief technology officer's sale on 27 October 2026.

Two tests matter. The first is whether the shares hold above the $97.75 pre-report close as the lock-up supply arrives. The second is whether they recover toward the falling 50-day average before the February report. Wide swings are common across the market: the standard deviation of 63-session returns across liquid US stocks was 27% in 2026.

Catalysts ahead

• 17 November 2026: Cerebras at the TomorrowX Summit. A company investor event where management can update data-center delivery timing, the Microsoft ramp and financing plans behind the 2027 tripling goal.
• 9 December 2026: Cerebras at the Barclays Global Technology Conference. Management's first public check on fourth-quarter progress against the $310–330 million core revenue guide.
• ~25 February 2027 (est.): Fourth-quarter and full-year 2026 report. Shows whether core revenue reached $310–330 million at a 47%–49% core gross margin, and sets the first dollar outlook for 2027, including Microsoft revenue.

What would change the view

A daily close below $85, well under the $97.75 close of 12 November 2026 that preceded the raised guidance, which would show the market rejecting the outlook. Fourth-quarter 2026 core revenue below the $310 million low end of the guide would also break the case.

The risks this note tracks, each with what would confirm it:

• OpenAI demand or payment timing slips, cutting into revenue that rests on a multiyear $20 billion agreement. Confirmed by: a reduced or rescheduled OpenAI commitment, or backlog falling below $26,100 million at the fourth-quarter 2026 report.
• Data-center capacity arrives late, pushing backlog conversion and the 2027 tripling goal out. Confirmed by: management delaying any part of the capacity contracted for delivery by the end of 2027, or delaying CS-4 full production.
• Pre-IPO and insider share sales add supply as the main lock-up ends. Confirmed by: new Form 4 filings showing large executive or director open-market sales, or a secondary offering by existing holders.
• Spending outruns funding and forces an equity raise that dilutes shareholders. Confirmed by: an equity offering or a large new debt facility, or cash and short-term investments falling well below $7,361 million at the fourth-quarter report.
• Hyperscalers' in-house chips take inference work that would otherwise go to Cerebras. Confirmed by: Microsoft or Oracle scaling back or delaying its Cerebras deployment, or Microsoft revenue failing to start in the first quarter of 2027.

What it trades with

Cerebras trades with the AI infrastructure group: GPU cloud providers, AI chipmakers and data-center suppliers. Benzinga reported on 11 November 2026 that fabless chip stocks rose after CoreWeave's third-quarter results. Its 13 November 2026 market report led with Nvidia's chief executive exit, a sentiment event for the whole AI chip group.

The company-specific driver is OpenAI's demand and how fast Cerebras brings data-center capacity online. The Microsoft deal adds a second large buyer, but Microsoft is also one of OpenAI's main cloud providers, so the two exposures can move together. That makes the shares more sensitive to OpenAI's funding and compute plans than to Nvidia's product cycle.