Should You Hold on to Cerebras Stock Despite Its 25% One-Month Dip?

Cerebras shares fell 25.3% over the past month as rising capacity costs and customer concentration pressured the outlook, despite Q2 2026 core revenue rising 103% year over year to $209.9 million and remaining performance obligations reaching $25.4 billion as of June 30, 2026. Core gross margin fell 590 bps sequentially to 40.6%; management expects Q3 gross margin of 38%-40% and core operating margin of negative 25% to negative 23%. The company forecasts Q3 core revenue of $214-$216 million and expects manufacturing capacity to increase more than tenfold in 2026, but competition, losses and valuation remain risks.

Analysis

The key issue is whether Cerebras can turn demand into attractive owned-capacity economics—not whether inference demand exists. Renting systems back may protect near-term service levels, but it transfers value to capacity providers and makes reported growth less informative until utilization and gross margin improve. If customer deployments slip while infrastructure commitments remain fixed, operating leverage runs in reverse: losses and cash needs can expand even with rising revenue.

Treat the $25.4B RPO as a visibility claim, not equivalent to funded, near-term revenue. Verify duration, cancellation terms, customer concentration, and the portion dependent on OpenAI before capitalizing it into valuation. AWS availability in 2027 is a later diversification catalyst, not a near-term hedge against concentration. Flex, Sanmina, and TSMC may see incremental activity from expansion, but supplier revenue or margin benefit is not established by the article; capacity execution and Cerebras’ ability to finance deployments matter more.

Over the next 1–3 months, the test is whether Q3 is actually the margin trough and whether delivery constraints impede revenue conversion. Over 6–18 months, the bull case requires owned systems to displace rented capacity, manufacturing expansion to ship on schedule, and non-OpenAI demand to become material. A sustained recovery in gross margin alongside revenue growth would invalidate the cautious view; another margin decline, delayed deployments, or weaker RPO conversion would undermine the growth premium. The drawdown alone does not establish cheapness, and consensus target upside is not an independent valuation anchor.

Key Decisions for Investors

• Avoid chasing CBRS solely on the selloff; keep it on a catalyst watchlist until the next results clarify gross-margin progression, owned-versus-rented capacity mix, deployment timing, and cash consumption.
• For a relative-value expression, consider a small, defined-risk short CBRS / long NVDA or AMD basket only if CBRS continues to lag while its margin outlook weakens. The pair can lose if Cerebras posts a credible margin inflection or a major customer-diversification milestone; size for high single-name volatility.
• Do not treat the RPO headline as a standalone buy signal. Seek disclosure on contract duration, cancellation rights, customer concentration, and conversion schedule; materially slower conversion or renewed dependence on rented systems would strengthen the downside case.
• Monitor FLEX, Sanmina, and TSMC as possible secondary beneficiaries, but require evidence of incremental orders or guidance impact before trading them on Cerebras expansion news.