UPDATE: Cerebras Stock Drops 10% After 68% IPO Pop

Cerebras Systems' stock fell roughly 10% on Friday, May 15, in its first full day of regular trading — a sharp pullback from the 68% pop the AI chipmaker rang up on Thursday's debut. CBRS opened around $311 and slid through the session, with shares trading near $279 in the afternoon as Wall Street re-examined the numbers underneath the most-hyped tech IPO of 2026.

Thursday's debut had been a rocket. Cerebras priced shares at $185, opened far above that, and closed at $331.07, putting its market cap near $95 billion and making it the largest U.S. tech IPO since Uber in 2019. Twenty-four hours later, the same investors are doing harder math.

What Triggered the Pullback

Two numbers are doing most of the work in analyst notes today: 200 and 80.

At Thursday's close, Cerebras was trading at roughly 200 times trailing sales. For comparison, Nvidia — the company Cerebras is positioned against — trades around 27 times sales. Even by the elevated standards of the 2026 AI hardware market, a 200x multiple requires nearly flawless execution and continued demand growth that few companies have ever sustained.

The other number is the customer concentration. About $20 billion of Cerebras' reported $24+ billion order backlog comes from a single source: a multi-year cloud infrastructure agreement with OpenAI. That's more than 80% of the book riding on one customer, in an industry where buyers like OpenAI have shown they're willing to spread risk across multiple chip suppliers, not consolidate it.

Neither concern is new. Both were in the S-1. But IPO-day momentum tends to push that kind of analysis to the side, and Friday is when it usually returns.

Context: A Hot IPO Window

Cerebras went public into a market that has been unusually receptive to AI infrastructure stories. Nebius reported 684% year-over-year revenue growth in Q1 2026. CoreWeave continues to extend its lead. Anthropic just committed $200 billion to Google's cloud and TPU stack over five years. Capital is flowing into anything that promises to power the next wave of model training and inference.

That backdrop is part of what made Cerebras' debut explosive. The company's wafer-scale engine — a chip the size of a dinner plate, with 900,000 cores on a single piece of silicon — is one of the few credible architectural alternatives to the GPU-cluster approach Nvidia and AMD have ridden to dominance. For investors looking for a non-Nvidia way to bet on AI compute, Cerebras was the cleanest pure play available in the public market.

But "clean pure play" cuts both ways. It also means there's nowhere to hide if the underlying assumptions wobble.

Why the OpenAI Dependency Matters

The OpenAI deal that anchors Cerebras' backlog is unusually large and unusually long. It runs across multiple years and ties Cerebras' near-term revenue trajectory closely to OpenAI's compute roadmap. If OpenAI's demand pattern shifts — toward different architectures, toward in-house chips, toward another supplier — Cerebras feels it immediately and there's no second-largest customer of comparable size to absorb the slack.

That's not a hypothetical concern. OpenAI publicly maintains relationships with multiple chip vendors precisely so it isn't dependent on any single one. The $20 billion Cerebras order is a meaningful share of OpenAI's compute spend but not all of it, and the order is structured with delivery obligations that Cerebras still needs to actually meet.

Manufacturing wafer-scale engines at the volumes implied by a $20 billion contract is an industrial challenge no company has ever attempted at this scale. Yield, packaging, and supply chain execution risks all become real once production ramps.

How Other AI Chip Stocks Are Responding

The Cerebras pullback is happening in the context of a broader cooling in AI-adjacent stocks Friday. Ford Motor Co. — which has been bid up partly on the assumption that automakers benefit from the energy demand around AI data centers — is having its worst day since 2024, off more than 8%. Bond yields and oil prices are putting pressure on the broader tech tape.

Nvidia and AMD are also down on the day, though by smaller amounts than Cerebras. The sector-wide move suggests Friday's CBRS slide is partly Cerebras-specific (the valuation and concentration story) and partly a market-wide reset on AI infrastructure exuberance.

What Investors and Analysts Are Watching

The first earnings report after IPO will be the meaningful test. Cerebras needs to show that the OpenAI backlog is converting to actual recognized revenue on schedule, and that sales pipeline outside OpenAI is growing fast enough to make customer concentration a story about today rather than the future.

Analysts will also be watching for signs that Cerebras can capture meaningful inference workloads, not just training. Inference is where the long-term volume sits — every query to a deployed model runs on inference hardware — and it's a market currently dominated by Nvidia GPUs and increasingly by hyperscaler-designed chips like Google's TPUs and AWS Trainium.

Lockup expirations are also on the calendar. Insider selling typically begins six months after an IPO, and a successful early performance can encourage early investors to take profits, adding selling pressure.

What's Next

The next several weeks will tell whether Friday's drop was a healthy reset or the beginning of a longer correction. Bull-case scenarios depend on Cerebras announcing additional large customer wins outside OpenAI — particularly in the enterprise inference market — and on demonstrating production execution at scale.

Bear cases focus on the math: at any multiple meaningfully above current sector averages, Cerebras has to grow into expectations that other companies have struggled to meet. A reset to a more conventional valuation, even with strong execution, would still mean meaningful losses from current levels.

The Bottom Line

Cerebras is a real company with real technology and a real customer relationship that anchors a multi-billion-dollar revenue trajectory. None of that changed between Thursday's close and Friday's open. What changed is that the public market took a second look at the multiple and the concentration, and decided the price wanted to come in. Whether Friday's 10% drop is the start of a longer reset or just normal post-IPO volatility will depend on execution that hasn't happened yet — and on whether the AI infrastructure boom continues funding companies at multiples no one would have entertained two years ago.