Cerebras Systems Inc. IPO at $42 per share surged to $55 then settled near $48. One customer – G42 – accounts for 83% of revenue. The post-IPO dip is not a value opportunity given concentration and lack of profitability.
Cerebras Systems Inc. (CBRS) went public at $42 per share, giving it a market capitalization near $8.7 billion on a non-diluted basis. The stock initially surged to $55 before settling near $48. That first-week pullback looks like a standard IPO pop-and-retrace. The pattern masks real risk for anyone buying the dip on wafer-scale AI compute alone.
The simple read is that Cerebras owns a structural moat in AI hardware – the WSE-3 is the largest commercial processor ever built, and the company claims superior performance for training large language models. New investors pile in expecting an Nvidia-lite trajectory. The better market read focuses on concentration, valuation, and competitive timeline.
Customer G42 accounted for 83% of revenue in the most recent period. Losing that relationship or seeing G42 slow purchases would crush the top line. Most AI chip companies diversify across hyperscalers. Cerebras relies on one Middle Eastern AI player. That is not a fix in one quarter. CEO Andrew Feldman and the board hold a supermajority of voting control, which limits shareholder recourse. Insider selling post-IPO could add pressure.
The raise came after years of delays, and the $55 intraday high suggests demand was real but not overwhelming. A 30% first-day pop for an AI chip stock in 2025 is modest. The subsequent drift lower signals that marginal buyers are not stepping in at those levels. The typical IPO aftermarket dip often finds a floor. Cerebras has structural issues that make this dip different.
Cerebras is not profitable, reporting GAAP net losses. The market cap of $8.7 billion prices in aggressive revenue growth and eventual margin expansion. For context, that multiple far exceeds what traditional semiconductor peers trade at when they have comparable revenue scale and no profit. The stock is pricing in success, not survival.
Nvidia, AMD, and Intel are the direct competitive threat. Cloud providers building custom silicon – Amazon, Google, Microsoft – are a second, slower-moving risk. If any hyperscaler deploys a chip that matches WSE-3 efficiency at a lower cost, Cerebras loses its only real argument: architectural superiority. The wafer-scale approach has high fixed costs and long sales cycles.
What would reduce the risk: a second customer signing a multi-year deal, or a major hyperscaler licensing the wafer-scale design. What would make it worse: G42 revenue share creeping above 90%, or a competitor matching WSE-3 benchmarks. Until one of those signals materializes, the dip does not look cheap. It looks like an IPO finding its level.
The next catalyst is the first earnings call as a public company. Investors will hear about G42 renewal terms and new customer pipeline. A buyer at $48 is still paying for a near-$9 billion enterprise with one customer, no operating profit, and a market that is shifting toward custom ASICs. The Nvidia franchise is not cracking. AMD is gaining. Intel is still in the fight. In this environment, Cerebras must not only execute on product but also expand its customer base before the next funding round or insider lockup expiry.
The post-IPO dip offered a lower entry price. It did not solve the concentration problem. Until the customer base widens, the risk-reward favors staying on the sidelines.
This article first appeared on AlphaScala's market analysis and stock market analysis.
Prepared with AlphaScala research tooling and grounded in primary market data: live prices, fundamentals, SEC filings, hedge-fund holdings, and insider activity. Each story is checked against AlphaScala publishing rules before release. Educational coverage, not personalized advice.