Cerebras Systems shares tumbled 16% in extended trading on Wednesday after reporting second-quarter revenue of $180 million that trailed Wall Street forecasts, even as the company raised its full-year guidance to $880 million and pointed to soaring artificial intelligence demand for its specialized wafer-scale chips.
The stock market reaction laid bare the strict math governing richly valued artificial intelligence stocks as shares plummeted in extended trading following the earnings release. While Cerebras posted an adjusted loss of five cents per share, which easily beat Wall Street consensus forecasts of a 17-cent loss, its top-line revenue figure fell short of expectations according to financial estimates. Analysts surveyed by LSEG had expected $194 million in revenue.
Core Revenue Metrics and Accounting Adjustments
Evaluating the chipmaker’s financial health requires navigating a nonstandard accounting picture. The reported second-quarter sales figure of $180 million represents total revenue, but the company also highlights a core revenue figure that includes pass-through revenue and adjustments. Cerebras factors in pass-through revenue that carries no margin, and adds back the amortization of warrants issued to customers, which is subtracted under standard accounting practices.
Through this core metric, the company reported $210 million for the quarter, marking a 50% increase from the same period a year earlier and surpassing analyst expectations. Overall, Cerebras recorded a net loss of $450.5 million, a sharp reversal from a profit of $309.5 million in the prior year. The vast majority of that net loss stemmed from stock-compensation costs totaling $386.6 million.
Soaring AI Demand Versus Margin Pressures
Chief Executive Andrew Feldman offered an upbeat assessment of market conditions during post-earnings interviews and conference calls. Demand for specialized inference hardware is through the roof
as companies willingly pay premium prices for rapid processing capabilities.
“Gross margins are are in a good spot, and growing, because fast inference is priced at a premium,”
Andrew Feldman, CEO of Cerebras Systems
Despite strong customer interest, gross margins experienced near-term friction. The company’s adjusted gross margin for the quarter ended June 30 dropped by about 5 percentage points to 40.6%, down from 46.5% in the preceding three months due to higher costs incurred to rent computing capacity previously deployed at other customer sites.
Wafer-Scale Architecture and the Nvidia Rivalry
At the center of Cerebras’ hardware strategy is its flagship wafer-scale engine, a single processor roughly the size of a standard dinner plate that integrates compute cores and memory onto the same silicon slab without requiring a separate interconnect. This design aims to provide low-latency performance for real-time interactions, challenging Nvidia in specific AI tasks.
Feldman argued that the company’s architecture provides a distinct insulation against rising component expenses. While market-leading processors rely on expensive high-bandwidth memory, on-chip memory placement helps shield the firm from surging costs that have driven Nvidia’s component prices upward.
“Nvidia’s prices have gone through the roof because of HBM prices,” Feldman said, referring to the high-bandwidth memory included with AI processors. “This is a battleground, and if they can’t deliver or they’re having significant component price increases, of course that helps.”
Andrew Feldman, CEO of Cerebras Systems
Furthermore, manufacturing efficiency benefits from supply chain positioning. Because Cerebras utilizes Taiwan Semiconductor Manufacturing Company’s 5-nanometer process, it faces less manufacturing pressure than competitors relying on more constrained 2-nanometer and 3-nanometer nodes.
Raised Full-Year Guidance and Future Outlook
Looking beyond the immediate quarterly shortfall, executive leadership expressed confidence in future scaling, raising its full-year revenue outlook to a range of $880 million to $890 million, up from its previous projection of $855 million to $865 million. The company also boosted its annual adjusted gross margin outlook to 41%-43%.
Cloud services emerged as a major growth driver, with cloud revenue roughly quadrupling to $126 million compared to the same period last year even as physical hardware sales declined from $70.3 million to $54.1 million. The firm’s total remaining performance obligations held steady at $25.4 billion, anchored by a multiyear agreement to provide AI compute infrastructure to OpenAI and recent collaborative ventures with Advanced Micro Devices following its public debut on the Nasdaq in May.
“We will manufacture more efficiently. We’ll get better pricing on componentry. We’ll amortize our manufacturing organization over more units,” Feldman said. “All of those point up and to the right.”
Andrew Feldman, CEO of Cerebras Systems
With production scaling planned to increase manufacturing capacity by more than tenfold throughout the year and core revenue expected to triple by the end of the next fiscal year according to forward projections, investors will watch whether expanded factory output translates into closed hardware sales when the chipmaker reports its third-quarter results later in the year.
Related reading
