Why Marvell’s Earnings Beat Wasn’t Big Enough for Such Great Expectations
A semiconductor company reports record revenue. Earnings beat Wall Street’s estimates. The company raises its guidance for the rest of the year, and not just by a little, but by a meaningful margin. Data center revenue is exploding, up 46% year-over-year. The CEO sounds confident. The fundamentals are undeniably strong.
And the stock drops 8%.
Welcome to Marvell Technology (NASDAQ: MRVL) in 2026, where beating expectations isn’t enough anymore. Where “good” has become the new “disappointing.” Where the market’s appetite for AI-fueled growth has become so ravenous that even a perfectly solid quarter can leave investors asking, “Is that it?”
This is the story of how Marvell found itself trapped by its own success. And why the AI semiconductor trade has entered a new, far more complicated phase.
Let’s dig in.
The Numbers That Looked Great on Paper
First, let’s get the facts straight. Because by any historical measure, Marvell’s fiscal second-quarter results were excellent.
Revenue: $2.74 Billion, Up 37% Year-over-Year
Marvell reported Q2 CY2026 revenue of $2.739 billion, up 37% from the same quarter last year and 13% sequentially. That topped the analyst consensus of $2.71 billion, a beat, albeit a modest one at about 1%.
Think about that for a second. Thirty-seven percent year-over-year growth. In any other industry, that would be headline news. In any other company, that would trigger a celebration. But for Marvell, in this market, it was… fine. Just fine.
EPS: $0.94, A Beat, But Barely
Non-GAAP earnings per share came in at $0.94, beating the $0.93 consensus estimate. That’s a $0.01 beat. A penny. The smallest possible margin of victory.
Now, I’m not saying a beat is a beat, it absolutely is. But when your stock has nearly tripled in a year, investors aren’t looking for pennies. They’re looking for fireworks.
Data Center: The 800-Pound Gorilla
Here’s where things get interesting. Data center revenue hit $2.17 billion, up 46% year-over-year and now representing a staggering 79% of total revenue.
That’s the headline within the headline. Marvell is no longer a diversified chip company, it’s an AI infrastructure company, full stop. The data center business alone is bigger than many chipmakers’ total quarterly revenue. Optical DSPs, custom silicon, switching, broadband analog, CXL memory expansion, all contributed.
CEO Matt Murphy called it “continued strong demand” across the data center division. And he wasn’t wrong.
The company also raised its full-year fiscal 2027 revenue outlook to about $12 billion and lifted its fiscal 2028 target to approximately $18 billion. It now expects data center revenue to grow about 60% in fiscal 2027 and over 60% in 2028.
If you’re scoring at home: record revenue, earnings beat, guidance raised, data center exploding.
So why did the stock drop?
So Why Did the Stock Drop 8%?
This is where the story gets interesting. Because the answer isn’t about Marvell’s performance. It’s about the expectations wrapped around that performance.
The “Beat” Wasn’t Big Enough
The sharp decline was attributed primarily to “disappointment over the magnitude of the earnings beat”. Let that sink in. The market wasn’t disappointed that Marvell missed — it was disappointed that Marvell didn’t blow past expectations.
With a beat margin of only about 1.1%, investors had been anticipating something much bigger. The options market had priced in a blowout. When Marvell delivered “just” a solid beat, profit-taking kicked in.
In other words: the market wasn’t punishing Marvell for failing. It was punishing Marvell for not over-delivering enough.
The Valuation Problem: When P/E Hits 84x
Here’s the uncomfortable truth that no one wants to talk about: Marvell’s stock had soared more than 220% over the past year heading into the report. Its price-to-earnings ratio sat above 84x.
Let me put that in perspective. The forward P/E was around 60x — making Marvell one of the 15 most expensive stocks in the S&P 500. For comparison, rival Broadcom trades at about 32x forward earnings.
When you’re priced for perfection, you have to deliver perfection. And perfection, in this market, means a massive beat, not a penny.
“Buyers had already banked a 178% gain this year,” one analysis noted. “Leaving little room for results that merely exceeded forecasts rather than dramatically resetting expectations”.
Jim Cramer put it bluntly: the selloff reflected “valuation and positioning rather than a weak quarter”.
The 178% Run-Up That Changed Everything
Marvell had already gained roughly 178% in 2026 alone before the earnings report. That’s not a stock, that’s a rocket ship.
When a stock runs that hard, that fast, the expectations bar doesn’t just rise, it launches. Every piece of good news becomes “priced in.” Every beat becomes “expected.” Every “strong quarter” becomes “the baseline.”
And suddenly, you find yourself in a situation where 50% revenue growth guidance — which Marvell delivered for Q3, is met with a shrug.
“Growth that almost no company in the market can match got sold because the price had already extrapolated further”.
The Google Deal Hangover
If valuation was the main course, the Google partnership was the bitter dessert.
$120 Billion Promise, But When?
Just a week before earnings, Marvell announced a massive expanded partnership with Google involving custom AI chips. The deal could represent up to $120 billion in revenue over six years if all planned targets are met.
That’s a staggering number. Life-changing. Company-defining.
But here’s the catch: that $120 billion figure reflects a “conditional maximum potential rather than guaranteed orders”. And the revenue contribution? It’s back-loaded.
Investors learned that the real economic payoff lands in FY29. The Google relationship was more “eventually” than “right now”.
“Concerns over the timing of revenue from Marvell’s massive Google AI chip partnership overshadowed strong quarterly results”.
In plain English: the market wanted Google revenue now. Marvell said “soon.” The market didn’t like that answer.
Market Wanted Certainty, Got Potential
Marvell’s guidance didn’t include any upside surprise from the Google deal because the customer had already been incorporated into prior forecasts. There was no “Google bump” in the numbers.
Investors were looking for fresh clues on long-term growth and didn’t get them. Instead, they got confirmation of what they already knew: the Google partnership is huge, but it’s going to take time.
And in a market that had already priced in perfection, “it’s going to take time” is a four-letter word.
What the Analysts Are Saying
Wall Street is divided, which, if you’ve been paying attention, is exactly what you’d expect.
The Bulls: “Ignore the Expectation Mismatch”
Bank of America, UBS, Barclays, Wells Fargo, and Citi all reiterated buy or buy-equivalent ratings after the report.
BofA analyst Vivek Arya kept a $365 price target on the stock, implying 51% upside. UBS held a $310 target. The consensus price target among analysts is around $271-$299, implying significant upside from current levels.
Arya’s argument? “We ignore this expectation mismatch,” he wrote, arguing Marvell is positioned to accelerate revenue growth as its compute, networking, optics, security, and storage business expands.
The Bears: Valuation Concerns Are Real
On the other side, Goldman Sachs analyst James Schneider reiterated a Neutral rating and $195 price target. The stock could remain “range bound” after results that largely matched expectations, he said.
Morgan Stanley also kept an Equalweight rating, raising its target to just $246, only about 2% above Marvell’s Thursday close. The bank noted that “fundamentals are accelerating, but expectations already appear high”.
The Middle Ground: Range-Bound Until More Clarity
Goldman wants to see how quickly Marvell’s custom compute business grows in the second half of fiscal 2027. The firm also wants more detail on the Google partnership and is watching the ASIC revenue ramp at Microsoft.
Morgan Stanley, meanwhile, highlighted Marvell’s growing emphasis on “XPU attach” businesses, optical connectivity, and scale-up opportunities, suggesting the strategy could offer greater visibility than competing directly with Nvidia.
The next major catalyst? Marvell’s October 6 Investor Day, where management is expected to detail its longer-term AI infrastructure strategy.
The Bigger Picture: What This Means for AI Semiconductors
Marvell’s earnings story isn’t just about one company. It’s a warning sign for the entire AI semiconductor trade.
The AI Trade Has Entered a New Phase
For the past two years, AI stocks have enjoyed a “show me the growth” phase. Any company that could credibly tie itself to AI infrastructure saw its stock soar. The bar was low. The narrative was everything.
That phase is over.
We’ve now entered the “show me the profitable growth” phase, and more importantly, the “show me growth that exceeds already-stretched expectations” phase. The narrative alone isn’t enough anymore. Execution must be flawless. And even then, it might not be enough if the stock has already run too far.
Execution Is the New Narrative
Marvell’s data center business is thriving. AI demand is real. The Google partnership is massive. The long-term outlook is compelling.
But none of that mattered on Friday because the stock had already priced in all of it — and then some.
This is the brutal math of high-valuation growth investing. When you’re trading at 60x forward earnings, the market isn’t buying your current performance. It’s buying your future performance, and it wants that future to arrive yesterday.
What to Watch Next
If you’re trying to figure out where Marvell goes from here, here’s what I’m watching.
October 6 Investor Day
This is the big one. Management is expected to provide additional detail on the long-term growth trajectory and the impact of recent large-scale commercial agreements. If there’s going to be a catalyst that resets expectations, this is it.
Morgan Stanley’s Joseph Moore said he would be “tactically long” the stock into the Investor Day if shares sell off.
Custom Silicon Ramp
Marvell’s custom business, including XPU and XPU-attached products, is expected to more than double in fiscal 2028. The question is: can that ramp happen quickly enough to justify the current valuation?
Goldman is waiting for evidence that “growth in the company’s largest business can continue at a pace that supports higher expectations”.
Gross Margin Trajectory
Here’s a subtle but important detail: Marvell guided Q3 non-GAAP gross margin to 57.5%-58.5%, slightly lower than Q2’s 58.9%. The reason? The expanding custom silicon product portfolio has different margin characteristics.
As custom silicon scales, margins may face near-term pressure. That’s not a deal-breaker, but it’s something investors will watch closely.
So here’s the bottom line.
Marvell delivered a genuinely strong quarter. Revenue was up 37%. Data center revenue jumped 46%. EPS beat estimates. Guidance was raised. The long-term outlook improved.
And the stock dropped 8%.
This isn’t a story about Marvell failing. It’s a story about the market’s expectations becoming so inflated that even a beat-and-raise quarter can feel like a letdown. It’s a story about valuation catching up with reality. It’s a story about the AI trade entering a new, more demanding phase.
If you’re a Marvell investor, the fundamentals are still sound. The AI thesis is still intact. The Google partnership is still massive. But you need to be honest with yourself about what you’re buying: a company priced for perfection, in a market that demands nothing less.
The next few months will be critical. Watch the Investor Day. Watch the custom silicon ramp. Watch the margins.
And remember: in a market where expectations have gone parabolic, sometimes a beat isn’t a beat. Sometimes, it’s just not enough.
What’s your take on Marvell’s earnings? Are you buying the dip or waiting for more clarity? Let me know in the comments, and don’t forget to subscribe for more AI semiconductor analysis.
Comments
Post a Comment