2026-08-14

Cerebras (CBRS) Stock Sinks 14% on Earnings - So Why Did Wall Street Raise Its Price Targets? Inside the Headline Miss vs. 103% Core Revenue Growth

Written & reviewed by the TradeSmrt editorial team · 2026-08-14

In this article

What Happened
Why Wall Street Didn't Flinch at the Selloff
What to Take Away From This
FAQ
Sources

What Happened

Shares of AI chipmaker Cerebras Systems (NASDAQ: CBRS) whipsawed after the company reported second-quarter earnings following Wednesday's close on August 12. The stock dropped nearly 14% in after-hours trading immediately following the release, and the selling continued into Thursday's regular session, extending a double-digit decline from the prior close of $262.06. At one point in Thursday's premarket, the drop exceeded 17%. On the surface, the disappointment made sense: second-quarter revenue came in at $180 million, missing the $194 million analysts had modeled by roughly 7%. A revenue miss - especially from one of the more closely watched names in AI chips - triggered an immediate sell-first reaction.

But look one layer deeper into the report and the picture changes considerably. On the profitability side, Cerebras actually beat expectations: adjusted loss per share came in at just 5 cents, far better than the 17-cent loss Wall Street had projected. And the metric the company itself treats as the real measure of its business, "core revenue," rose 103% year over year to $209.9 million. Within that, core cloud revenue - arguably the most direct read on actual AI inference demand - jumped 287% to $127.7 million. Backed by that strength, the company raised its full-year core revenue guidance to $880 million-$890 million, up from a prior range of $855 million-$865 million. It's unusual for a company to raise its annual outlook in the same report where it misses its headline revenue number.

Making sense of that gap requires understanding how Cerebras's revenue is structured. The GAAP total revenue the company reports and the "core revenue" figure it discloses separately to show underlying business performance follow different recognition rules. This quarter's miss traced mainly to a 23% year-over-year decline in hardware sales, which fell to $54.1 million - largely a function of certain large hardware-supply contracts having their revenue recognition pushed into future quarters rather than any drop in underlying demand. Cloud services revenue, which most directly reflects how much customers are actually using Cerebras's chips right now, nearly quadrupled instead. In short, the metric that shows real customer usage surged, while the metric that shows what got booked as revenue this particular quarter slipped because of contract-timing mechanics.

Why Wall Street Didn't Flinch at the Selloff

Analysts appear to have parsed that distinction correctly. Morgan Stanley kept its Overweight rating and raised its price target from $273 to $279. Wedbush maintained Outperform and lifted its target from $280 to $290. UBS went further, setting a $330 target and calling the selloff a "compelling buying opportunity" outright. There was some caution too - Mizuho kept its Outperform rating but trimmed its target from $310 to $300, and Citi maintained Buy while cutting its target from $340 to $320. What's notable is that even the firms that lowered targets kept their bullish ratings intact. No major shop downgraded the stock to Sell.

That restraint reflects Cerebras's position as a company that has approached the AI chip market from a genuinely different angle from the start. Founded in 2015 by Andrew Feldman, Cerebras took the opposite approach from Nvidia's strategy of networking many parallel GPUs together. Instead, it built the Wafer-Scale Engine (WSE), a single chip carved from - and using - an entire silicon wafer rather than being diced into smaller pieces, measuring 46,225 square millimeters, roughly the size of a dinner plate. That design lets the chip process massive amounts of data without the bottlenecks that come from moving data between separate chips, an advantage that's proven especially valuable for AI inference workloads where response speed matters. The company went public on the Nasdaq on May 14, closing its debut day with a market capitalization of $95 billion and making Feldman a billionaire with an estimated net worth of $3.4 billion. This report was only the company's second as a public company, a useful backdrop given that the market is still in the early stages of learning how to read Cerebras's quarterly disclosures.

The reaction elsewhere in the chip sector the same day is telling. While Cerebras sold off, established names like Intel and AMD actually traded higher, suggesting the market read this as a company-specific revenue-timing issue rather than a signal of broader weakness in AI chip demand. The operational details Cerebras disclosed support that read: the company said it has 600 megawatts of data center capacity already under contract and plans to scale its manufacturing capacity more than tenfold within 2026. In a post-earnings interview, Feldman said AI demand is "through the roof" and that customers are willing to pay a premium for the company's specialized inference chips.

What to Take Away From This

A headline revenue miss isn't automatically bad news. When a company like Cerebras has GAAP revenue and underlying business metrics diverging because of contract-timing or accounting structure, selling purely off the headline number can mean missing the real growth story underneath. Always check what "core" metric a company is emphasizing and why.
Analyst rating and price-target changes matter more when they move against the stock. When a stock drops 14% and major investment banks hold their ratings or raise targets anyway, that's a signal of a gap between the market's immediate reaction and professional fundamental analysis. Analysts can be wrong too, but the absence of downgrades to Sell is itself useful information.
Hardware revenue and services/subscription revenue carry different volatility profiles. Hardware sales can swing sharply quarter to quarter depending on when a single large contract gets delivered and recognized, while cloud/subscription revenue tends to show a steadier growth trend. Separating the two matters when analyzing a growth stock's earnings.
Newly public companies face a credibility test in their first few earnings reports. This was only Cerebras's second quarterly report as a public company. Early-stage volatility in a freshly IPO'd company's results often reflects the market still learning the company's reporting patterns and business cycle rather than a fundamental problem.
Comparing reactions across peers helps diagnose what a selloff actually means. Intel and AMD trading higher the same day Cerebras fell tells you the market interpreted this as a company-specific issue, not a sector-wide AI chip demand slowdown. Checking how peers in the same industry traded during a single-stock selloff is a useful way to judge what's really driving it.

FAQ

Why did Cerebras stock fall 14% after earnings?

Second-quarter revenue of $180 million came in about 7% below the $194 million analysts expected. The main driver was a 23% year-over-year drop in hardware sales, largely tied to the timing of revenue recognition on large supply contracts shifting into future quarters rather than a drop in underlying demand.

If revenue missed, why did Wall Street raise its price targets?

Cerebras's separately disclosed "core revenue" metric rose 103% year over year, and core cloud revenue jumped 287%. Adjusted loss per share also beat expectations, and the company raised its full-year core revenue guidance. Firms including Morgan Stanley, Wedbush and UBS pointed to those underlying figures in treating the selloff as overdone and raising their price targets.

What does Cerebras do, and how is it different from Nvidia?

Founded in 2015 by Andrew Feldman, Cerebras builds AI chips using its Wafer-Scale Engine, which uses an entire silicon wafer as a single chip rather than networking multiple smaller chips together the way Nvidia's GPU-based systems do. That design reduces data bottlenecks and has proven particularly effective for AI inference workloads that require fast response times. The company went public on the Nasdaq in May 2026.

Sources

This article is an original synthesis and analysis based on the reporting below, not a reproduction of the original articles. Please check the source articles directly for the most current figures.

⚠️ This article is for informational purposes only and is not investment advice. Market conditions change constantly - always verify the latest data before making investment decisions.