Marvell: It's Down 10%, But This May Not Be The Dip To Buy

Marvell (MRVL) delivered solid Q2 ’27 results, beating revenue and EPS expectations, and raised FY’27 revenue guidance to $12B and FY’28 to $18B. Attention shifted to its transformative Alphabet/Google deal, which could generate up to $120B in cumulative revenue over 6–6.5 years, with warrant vesting tied to incremental sales milestones. However, gross margins are expected to decline as custom-chip programs ramp, tempering near-term profitability.

Analysis

GOOGL is the cleaner beneficiary because the strategic value is not the incremental revenue line item; it is the ability to lower unit compute costs, improve cloud gross margin, and reduce dependence on external accelerators. If the custom silicon program scales, Alphabet can choose to subsidize AI usage more aggressively than peers, which should help workload retention and enterprise attach rates over the next 6-18 months.

MRVL gets top-line visibility, but the market will eventually reprice the quality of that revenue. Custom-chip programs are typically lower-multiple than merchant silicon, more customer-concentrated, and more exposed to yield/packaging slippage, so the real risk is margin dilution and a lower terminal multiple even as reported sales accelerate. That creates a second-order loser set in higher-margin networking and accelerator vendors if hyperscalers keep internalizing more of the stack.

The consensus may be overcalling the durability of the upside for MRVL and undercalling the strategic benefit to GOOGL. In the next 1-3 months, the key falsifier is whether management can show that incremental custom volume does not compress gross margin more than expected; if margins roll over faster than guidance implies, the narrative shifts from "transformative" to "low-quality growth." For GOOGL, the thesis weakens only if deployment is delayed, yields disappoint, or AI demand growth slows enough that internal silicon savings do not flow through to cloud economics.

Key Decisions for Investors

• Long GOOGL / short MRVL as a 2-4 month pair: express the view that Alphabet captures the economics while Marvell bears the mix-down risk; target 8-12% relative outperformance if the market re-rates quality of earnings.
• Fade MRVL strength on rallies above the post-news range with a defined stop if gross-margin commentary stabilizes; the risk/reward favors waiting for the first analyst-note cycle before adding size.
• Add GOOGL on pullbacks into the next earnings revision window; the setup is stronger on 6-18 month margin expansion than on immediate revenue optics.
• Watch NVDA and SOXX as indirect pressure valves: if hyperscaler custom-silicon adoption broadens, rotate toward beneficiaries of AI demand growth rather than pure accelerator share.
• Alert item: if MRVL can hold margin guidance while expanding the custom pipeline, cover the short and reassess; that would signal the market is underestimating the economics of the business mix.