“Marvell tech had $1 billion more in sales than anyone thought. It’s involved directly with the data center. It’s got optical, it’s got fantastic equipment.” That was Jim Cramer on Friday morning, pointing to Marvell Technology as the standout report from the prior night. The stock was up sharply in pre-market trading, and Cramer’s framing raises a real question for investors: what exactly does a beat like this mean, and is Marvell actually worth owning as a data center play?
The verdict is that Cramer’s enthusiasm is directionally correct, but the thesis requires more precision than a one-liner about beating sales expectations. Marvell is a legitimate AI infrastructure company with a specific and defensible role in the buildout, and the Q4 numbers back that up. But the stock also carries a concentration risk that investors need to understand before treating it as a simple data center buy.
What the Beat Actually Looked Like
Marvell reported Q4 FY2026 revenue of $2.22 billion, beating Wall Street expectations near $2.21 billion. Non-GAAP EPS came in at $0.80 versus the $0.79 consensus estimate. Those are modest beats on the headline numbers, but Cramer’s “$1 billion more in sales than anyone thought” framing appears to reference the magnitude of data center revenue relative to where expectations stood earlier in the year, not the quarter-over-quarter margin of outperformance.
The more meaningful signal was in guidance. Management forecast Q1 FY2027 revenue of approximately $2.4 billion, comfortably ahead of Wall Street expectations near $2.28 billion. CEO Matt Murphy signaled that year-over-year revenue growth is expected to accelerate each quarter in fiscal 2027, driven by continued strength in data center. That forward guidance, more than the Q4 beat itself, is what sent shares up roughly 12% in pre-market trading.
The Custom Silicon Angle Cramer Is Pointing At
Cramer specifically called out optical and data center equipment. That maps directly to Marvell’s strategic positioning. The company makes custom AI chips (ASICs) for hyperscalers like Amazon Web Services, Microsoft, and Google, and it recently acquired Celestial AI to strengthen its optical interconnect capabilities. Data center revenue now accounts for roughly 74% of total revenue, a concentration that has been consistent across the past several quarters.
This is different from what NVIDIA does. NVIDIA sells GPU compute at scale to anyone who needs it. Marvell designs custom silicon specifically for individual hyperscaler workloads, and it builds the high-speed networking and optical interconnect that moves data between those chips. CEO Matt Murphy described the company’s custom AI design pipeline in Q2 FY2026: “Our custom AI design activity is at an all-time high, with the Marvell team now engaged in over 50 new opportunities across more than 10 customers.” That pipeline is what gives the guidance credibility.
Cramer also flagged NVIDIA in the same breath, noting that “nvidia under 180… has now become right now it’s just an S&P play.” NVIDIA is trading near $181 today, essentially flat year-to-date after a 64.5% gain over the past year. The implication is that Marvell, at a lower valuation and with a more targeted data center story, may have more room to move from here.
The Concentration Risk Investors Cannot Ignore
Cramer’s call is sound as a directional trade thesis, but it understates one structural risk. When 73% to 76% of Marvell’s revenue comes from data center in any given quarter, the company’s fortunes are tightly tied to hyperscaler AI capital expenditure decisions. If Amazon, Microsoft, or Google slow their infrastructure spending, Marvell has limited diversification to absorb the impact.
The automotive and industrial segment, which could have provided some balance, was largely divested. Marvell sold its automotive ethernet business to Infineon for $2.5 billion, closing in August 2025. That sale sharpened the portfolio and generated cash, but it also removed a revenue buffer. The Celestial AI acquisition and the pending XConn Technologies deal are both data center plays, meaning the concentration is deepening, not broadening.
For investors who believe AI infrastructure spending continues to accelerate through 2026 and 2027, that concentration is a feature. For investors who are uncertain about the pace of hyperscaler capex, it is a risk that needs to be priced in.
Who Should Pay Attention to This Call
Cramer’s framing works best for investors who are specifically looking for AI infrastructure exposure that is less correlated to NVIDIA’s daily moves. Marvell trades at a forward price-to-sales ratio of 7.67x, a premium to peers, but one that reflects its concentrated AI infrastructure positioning. Analyst price targets range from $105 from RBC Capital to $135 from Melius Research at RBC Capital to Melius Research’s higher target, both well above the current price near $87.73, suggesting the market has not yet fully priced in the bull case if data center growth holds. That gap between current price and analyst targets is the core of the opportunity Cramer is pointing at.
The industry average P/E of 43x provides context on where Marvell’s valuation sits relative to the broader semiconductor sector, a useful benchmark for investors weighing whether the current entry point is justified by the growth trajectory.
The call is less suited for investors who need sector diversification or who are uncomfortable with a stock where one bad quarter of hyperscaler capex guidance could erase months of gains. Institutional ownership sits at 83.51%, which means the stock is heavily held by professional money managers who will move quickly on any change in the AI spending narrative.
Marvell shares are up 18% over the past month and about 21% over the past year. The stock pulled back significantly from its late 2025 highs near $100 before this week’s earnings catalyst, which means investors are buying back in at a lower entry point than the peak enthusiasm of last December.
What to Watch Next
The most important number to track going forward is data center revenue growth on a year-over-year basis. Murphy committed to accelerating growth each quarter in FY2027, which means the Q1 FY2027 report will be the first real test of that claim. If data center revenue comes in at or above the implied trajectory from the $2.4 billion total guidance, the bull case stays intact. If it misses, the concentration risk becomes a live problem rather than a theoretical one.
The Celestial AI optical interconnect acquisition is also worth monitoring. CEO Murphy called it “a transformational milestone that accelerates our scale-up roadmap for interconnect” and a way to strengthen Marvell’s position in one of the fastest-growing opportunities in AI datacenter infrastructure. Revenue contribution from that deal is not expected until the second half of fiscal 2028, so near-term results will still be driven by the existing custom silicon and networking portfolio.
Cramer is right that Marvell’s Q4 report was the strongest number of the night. The more useful takeaway is understanding why: the company has built a specific, defensible position in AI infrastructure that is not dependent on winning the GPU arms race. Whether that position justifies buying the stock at today’s price depends entirely on your conviction that hyperscaler AI spending does not slow materially in the next 12 to 18 months.
Top Gaining Stocks
Top Losing Stocks