Marvell Technology: Greater Clarity Does Not Impress

Marvell’s shares fell 10% after earnings as investors questioned the upside from the Alphabet/Google deal despite strong results. Q2 revenue rose 37% to $2.74B, and Q3 guidance was $3.15B with 50% YoY growth expected, but valuation is stretched at ~20x sales and the implied incremental revenue from the Google agreement appears closer to ~$6B (vs up to ~$20B annually). The mix of strong growth with “less-than-expected” deal-driven revenue expectations drove the post-earnings selloff.

Analysis

MRVL’s downside is less about the reported quarter and more about the reset in what the market was willing to capitalize into the stock. At ~20x sales, investors were paying for a near-vertical AI ramp and a marquee customer that would de-risk the story; the disappointment now is that the path looks more incremental, which invites multiple compression before any earnings revision shows up. In the next few weeks, the stock likely trades as a sentiment/multiple name rather than a fundamentals name, with sell-side model cuts and de-grossing doing more damage than the underlying revenue run-rate.

Second-order, this is a relative loser for the broader custom-silicon complex if the market starts demanding proof of ROI on hyperscaler ASICs rather than rewarding press-release scale. That would spill into names like AVGO and even parts of the AI infrastructure basket where backlog quality matters more than headline design wins. GOOGL is the quieter beneficiary: if the silicon program is real, the economic upside is lower inference cost and tighter stack control, but that payoff is 6-18 months out and likely not enough to move near-term consensus earnings by itself.

The contrarian take is that the selloff may be overdone if the market is confusing timing with cancellation. A smaller 2025 contribution does not negate a multi-year attach rate inside Google, and once custom silicon is embedded, replacement and support revenue can be durable. What would falsify the bearish view is either a clearer backlog conversion path on the next call or evidence that hyperscaler AI capex is re-accelerating rather than being optimized for efficiency.

Key Decisions for Investors

• Short MRVL on any post-earnings bounce; target 1-3 months of multiple compression if the Google ramp remains vague. Risk/reward is favorable as long as the stock stays anchored to a premium sales multiple; cover if management quantifies a tighter 2025-26 revenue bridge.
• Pair trade: short MRVL / long AVGO for 6-12 weeks to stay AI-silicon exposed while isolating customer-concentration and expectation-risk in MRVL. AVGO’s diversified cash flow should be more resilient if the market reprices custom silicon timelines across the group.
• Do not chase GOOGL on this print alone; instead, set a watch item for the next capex/infrastructure update. If Google discloses lower inference cost or better AI serving economics, add a small long as a 6-18 month beneficiary of in-house silicon leverage.
• If MRVL reclaims the pre-earnings level without better Google visibility, fade the rally. That would imply the market is ignoring a model-cut cycle and you likely have another leg lower once estimates catch down.
• Use SOXX as a hedge if holding MRVL short; the thesis is company-specific multiple compression, not a broad semiconductor demand shock.