Cerebras Systems Shares Drop 16% After Q2 Revenue Misses Expectations

Cerebras Systems saw its shares fall approximately 16% in after-hours trading after the AI chipmaker reported second-quarter revenue that fell short of Wall Street expectations, despite delivering stronger adjusted earnings and raising its full-year outlook. The sell-off comes as investors scrutinize one of the most prominent newly public AI hardware companies for signs it can sustain momentum against entrenched competitors. (Sources: Reuters, SiliconANGLE, Cerebras investor relations)

Headline Revenue Falls Short

Cerebras reported quarterly revenue of $180 million, missing the Street's consensus estimate of roughly $194 million. The shortfall raised questions among investors about whether the company can translate surging AI demand into sustainable, profitable growth. The gap between headline and expected revenue was amplified by accounting treatment tied to customer warrants and pass-through revenue, items that do not reflect underlying chip demand but do shape the GAAP figures investors see first.

Core Revenue Exceeds Expectations

Beneath the headline figure, Cerebras' core revenue reached approximately $210 million, representing nearly 50% year-on-year growth and surpassing Wall Street's expectation of $201 million. The company adjusts its core revenue metric to account for pass-through revenue and the accounting effects of customer warrants, resulting in a notable divergence between the headline and core numbers. This adjusted figure is the metric management encourages investors to track, as it strips out non-cash and passthrough items to reflect direct hardware and services demand.

Net Loss Driven by Stock-Based Compensation

Cerebras recorded a net loss of $450.5 million for the quarter, a sharp swing from the $309.5 million profit posted in the same period a year earlier. However, approximately $386.6 million of the quarterly loss was attributable to stock-based compensation, which is why investors have been paying closer attention to the company's underlying operating performance. Elevated stock-based compensation charges are common among recently IPO'd technology companies, where equity grants vesting post-listing create large non-cash expenses that distort GAAP earnings without affecting cash flow.

Raised Full-Year Guidance

Despite the top-line miss, Cerebras raised its 2026 core revenue forecast to a range of $880 million to $890 million, up from its prior guidance of $855 million to $865 million. The company also projected third-quarter core revenue of $214 million to $216 million, signaling continued confidence from management regarding demand. The guidance increase stands in contrast to the share price decline, highlighting a disconnect between management's demand outlook and the market's immediate reaction to the headline miss.

Wafer-Scale Technology vs. Nvidia GPUs

Cerebras positions its wafer-scale chips as an alternative to Nvidia's GPUs, with a particular focus on low-latency AI inference. The company's approach integrates computing cores and memory onto a single large silicon wafer, enabling certain AI workloads to execute at high speeds. Nvidia commands an estimated 80% or more of the AI chip market, meaning any competitor's gains represent share taken from an entrenched incumbent with deep software ecosystem lock-in through its CUDA platform. Cerebras and others must compete not only on hardware performance but also on developer adoption.

Cloud Business Gains Traction

The company's cloud and services segment has grown rapidly, with cloud revenue reaching approximately $126 million — a several-fold increase from the prior year. This shift toward recurring AI infrastructure and inference services may grow in importance as hardware sales experience fluctuations. A growing cloud business also positions Cerebras alongside providers like CoreWeave and the hyperscalers' own internal AI cloud efforts, where selling inference capacity as a service creates recurring relationships less exposed to lumpy chip procurement cycles.

$25 Billion Backlog Holds Steady

Cerebras' order backlog remained at approximately $25 billion during the quarter, unchanged rather than expanding. Investors are now closely monitoring how quickly the company can convert that backlog into recognized revenue and margins. A flat backlog in a period of surging AI demand suggests that either order inflow has moderated or that fulfillment is roughly matching new commitments, making backlog conversion rate a critical metric to watch in coming quarters.

Key Customer Relationships

OpenAI ranks among Cerebras' largest customers, supported by a major multi-year agreement for AI inference capacity. Cerebras has also formed a partnership with AMD, pairing its chips with AMD's rack-scale architecture to accelerate inference workloads. Concentration in a small number of large customers carries both opportunity and risk: deep relationships with AI leaders validate the technology, but revenue can be lumpy and dependent on individual customer deployment timelines.

Scaling Profitability Remains the Challenge

The post-earnings sell-off reflects investor demands for evidence beyond rapid revenue growth. Cerebras faces the task of demonstrating that it can scale production, improve margins, and convert its substantial backlog into profitable revenue — all while competing in a market where Nvidia holds a dominant position. With full-year guidance raised and core revenue growing nearly 50% year-on-year, the company's next test will be whether third-quarter results can close the gap between headline and core metrics while demonstrating margin progress on the path toward sustainable profitability.