Cerebras Systems: The Contrarian Entry Was at $180. Chasing at $220 Is a Different Story

Cathie Wood bought Cerebras Systems the same week the AI chip startup cratered after its first earnings report. That makes for a tidy contrarian headline. But the more useful question isn't whether she timed it well - it's whether the risk/reward for investors looking at the stock today is still there.

The short answer: the buying zone was the $160-$185 pocket that followed the earnings flush. Cerebras has since bounced back to roughly $220. The stock remains an interesting long-term AI infrastructure bet, but the valuation gap that makes GARP setups compelling has narrowed. Don't chase.

What Happened

Cerebras priced its IPO on May 14 at $185 - well above its initial range - and opened at $350 before settling at $311. The stock had just gone public, and the market was treating it like the next Nvidia. Ark Invest bought 105,616 shares on that first day for roughly $32.8 million, averaging about $311 per share, then added another 226,430 shares across the next few trading days.

Then came the first earnings report. On June 23, Cerebras reported fiscal Q1 revenue of $193.4 million, up 94% year-over-year, beating consensus of $180.8 million. Hardware revenue was $110.6 million and cloud services $82.8 million. The GAAP net loss narrowed to $14 million from $23.9 million a year earlier.

On the surface, that's a solid debut quarter. Revenue nearly doubled. Losses narrowed. The company had just announced a multiyear deal with OpenAI valued at more than $20 billion for 750 megawatts of compute, plus a partnership with Amazon Web Services to deploy Cerebras inference hardware across AWS data centers. It entered the quarter with $3.3 billion in liquid capital.

But the guidance destroyed the stock. Cerebras forecast core gross margins - the profit left after cost of goods sold, excluding stock-based compensation and other adjustments - of 36% to 38% for Q2, down sharply from the 47% it just posted. Full-year core margins were guided to 38% to 41%. The market heard one thing: scaling is margin-dilutive, and fast.

The stock plunged more than 12% in premarket trading the next day and eventually sank to a 52-week low of $160.81, a drop of roughly 48% from the IPO-day close. From the $350 open, the decline was about 54%.

That's when Ark struck again. On June 25, it purchased 111,989 shares for $20.4 million - an average of $182 per share. That was the contrarian entry, near the exhaustion zone, after the selloff had done its work.

The Valuation Gate

Here's where the GARP test matters. Growth-at-a-reasonable-price requires the valuation to compress meaningfully relative to the growth rate. That compression happened at the $160-$180 level. It has been partially reversed.

At today's price near $220, Cerebras has a market cap of approximately $50 billion. Trailing price-to-sales is 82x. Using the company's own full-year FY2026 guidance of $855-$865 million, the forward P/S is roughly 58x. For comparison, Nvidia - the actual AI chip company with a $5 trillion market cap, dominant market position, and massive free cash flow - trades at about 20x sales. AMD sits at roughly 21x sales.

Cerebras is growing revenue at an extraordinary rate. Annual revenue has climbed from $24.6 million in 2022 to $78.7 million in 2023, $290.3 million in 2024, and approximately $510 million in 2025. That's roughly a 20x increase in four years. But a 58x forward sales multiple demands that this acceleration continues and that margins eventually expand, not contract.

The margin guidance is the problem. The TTM gross margin sits around 40%, and the company expects further compression to the 36%-38% range. Operating margin is -22% and free cash flow is negative $359 million over the trailing twelve months, driven by $416.5 million in capital expenditures. Cerebras is spending aggressively on data centers to support its cloud inference service, and that spending is eating into profitability.

The revenue trajectory is real. The margins are not.

The Moat Check

Cerebras's competitive advantage rests on its wafer-scale engine - a single processor the size of a dinner plate, fabricated on an entire silicon wafer rather than cut into dozens of smaller dies. The architecture eliminates the interconnect bottleneck that limits traditional GPU clusters, enabling up to 1,000 tokens per second for inference workloads. That speed matters for applications where latency is the difference between a usable product and a frustrating one.

The $20 billion OpenAI deal and the AWS partnership are validation, not guarantees. OpenAI has also invested directly in Cerebras with a $1 billion working capital loan. AWS is integrating Cerebras chips alongside its own Trainium processors in a disaggregated inference strategy.

But the moat needs stress-testing. Nvidia doesn't need to win inference to keep its valuation intact - it just needs to stay relevant. Google's TPUs, AWS's Trainium, custom silicon from hyperscalers, and Groq's LPUs all compete in the same space. Cerebras needs to prove its wafer-scale approach scales profitably, not just technically. The margin guidance suggests that proof is still pending.

Where the Stock Is Now

The technical picture shows a recovery in progress. The stock is trading above its 50-day moving average of roughly $211, and the RSI at 54 sits in neutral territory. The MACD line remains negative at -3.3, suggesting the rebound hasn't yet established full momentum. Over the past 20 days, the stock is up roughly 14.6%.

AInvest's aggregate signal labels Cerebras a Buy, with a composite analysis rating of 2.2. That consensus view reflects the growth trajectory and the strategic positioning, not the current valuation math.

Next earnings come on August 12. That date matters because the market will want to see whether Q2 margins held up against the 36%-38% guidance and whether revenue growth stays above the 88% year-over-year rate the company projected.

The Investor Implication

Cerebras is not a falling knife. The revenue growth is real, the customer relationships are genuine, the technology is differentiated, and the balance sheet - with $1.7 billion in cash against $2.2 billion in debt - is manageable. The company isn't going broke.

But the stock is also not dirt cheap. At 58x forward sales, negative operating margins, and declining gross margins, the valuation still demands years of flawless execution. The contrarian opportunity was the $160-$185 zone where ARK bought after earnings. The 37% recovery from the low to the current $220 level has compressed the upside.

For investors already holding Cerebras from the $180 zone or below, the position is defensible. The long-term AI infrastructure story is intact, and the company has time to prove itself.

For those looking to enter, I don't think chasing at $220 offers attractive risk/reward. The setup remains constructive on a multi-year horizon, but the better entry is likely on a pullback toward the $170-$190 range or after the August 12 earnings report clarifies whether margin pressure is stabilizing. If the stock drifts back to those levels, the risk/reward becomes compelling again.

If Cerebras pushes toward $300 without meaningful margin improvement or profitability inflection, I would reassess the valuation entirely. A 70x+ forward sales multiple on a loss-making company with declining margins isn't GARP - it's a narrative trade. Don't mistake the two.