Why is Marvell Technology stock tumbling today?

Marvell shares fell 7.9% pre-open after Q2 FY2027 results beat estimates but failed to impress on outlook. Revenue rose 37% to $2.739B and non-GAAP EPS was $0.94, while FY2027 revenue guidance was raised to ~ $12B (and FY2028 to ~ $18B), yet investors wanted stronger long-term growth acceleration from a major Google AI chip agreement. Gross margin guidance for Q3 (non-GAAP 57.5%–58.5%) implies a ~90 bps sequential decline as custom AI silicon mix weighs on near-term profitability.

Analysis

The near-term loser is the name most exposed to investor impatience around AI monetization: when a company raises numbers but not the long-range slope, the market reads it as evidence that the low-hanging re-rating is already spent. That pressure can spill into other custom-silicon and “AI-near-term” beneficiaries where valuation depends on a straight-line adoption curve, while diversified names with broader product mix and pricing power should hold up better.

Mechanically, the key issue is margin mix. A larger share of custom silicon can grow revenue but still compress gross profit dollars versus higher-margin networking/merchant silicon, so the stock is being marked on quality of growth rather than growth itself. The broader AI capex tape still looks intact, which is constructive for the ecosystem over 6-18 months, but it also means hyperscalers will keep forcing suppliers to compete on economics, not just access.

The consensus may be missing that this is less a demand problem than a sequencing problem: investors wanted evidence that the new customer win would pull revenue forward by quarters, not years. If that ramp shows up in the next 1-2 earnings prints, the drawdown can reverse quickly; if not, the multiple can stay under pressure even with beat-and-raise quarters. Falsifiers are simple: a second straight quarter of guide acceleration, stabilization in gross margin, or evidence that the custom program becomes a broader platform win rather than a one-off design.

Key Decisions for Investors

• Short MRVL vs long AVGO for 1-3 months: the pair isolates execution/valuation risk in custom AI silicon while keeping exposure to the broader AI infrastructure spend; target 10-15% relative underperformance in MRVL if margin pressure persists.
• If already long MRVL, trim into strength on any intraday bounce and wait for the next guide update; the current setup favors selling rallies until management proves the Google-related ramp is pulling revenue into the next two quarters.
• Watch GOOGL as a stealth beneficiary: if custom silicon lowers inference costs faster than expected, Google can improve AI economics without proportionate capex growth; consider a tactical long only if cloud commentary confirms deployment acceleration within 1-2 quarters.
• Use SOXX/SMH as a hedge rather than a directional bet: this is more a stock-specific multiple reset than a sector-wide demand shock, so broad semis should outperform MRVL unless AI capex commentary deteriorates across peers.