Marvell (MRVL) Q2 2027 Earnings Call Transcript

Marvell reported record Q2 FY2027 revenue of $2.739B (+37% YoY, +13% sequential) and non-GAAP EPS of $0.94 (above the guidance midpoint, +40% YoY), with the data center segment at $2.17B (+46% YoY) driven by AI demand. The company raised its outlook to $12B for FY2027 (up from $11.5B) and $18B for FY2028 (up $1.5B), with data center growth guided ~60% for FY2027 and >60% for FY2028; Q3 revenue guidance is $3.15B at the midpoint (+15% sequential, >50% YoY). Margins were supported by operating leverage (non-GAAP operating margin 36.6%, +180 bps YoY) as Marvell committed $1B of capacity prepayments for FY2027 supply constraints and continued capital returns ($200M buybacks in Q2). While communications revenue is expected to be lumpy and custom ramp creates a gross-margin sequential headwind in Q3, management expects non-GAAP gross margins of 57.5%-58.5% and long-term operating margin of 38%-40% (targeting entry in Q4 FY2027).

Analysis

MRVL is transitioning from a quarterly beat story to a longer-duration AI infrastructure claim on the market. The important mechanism is not the current guide raise; it is that revenue visibility is now stretching far enough out that the stock should trade more on 2028/29 optionality and operating leverage than on near-term EPS. That usually supports multiple expansion, but only if investors believe the mix shift to custom silicon can coexist with acceptable gross margins.

The second-order winners are hyperscalers, especially GOOGL, because bespoke silicon plus higher-bandwidth networking lowers cost per token and improves AI ROIC even if capex stays elevated. The losers are narrow point-solution vendors in optics/switching that cannot bundle connectivity, memory, and custom attach; Marvell’s integrated stack raises the bar on pricing power across the ecosystem. The capacity prepayment stance also signals supply is the binding constraint, so foundry/packaging/optics vendors may see better volume certainty, but Marvell’s free cash flow will likely lag reported growth for a few quarters.

Contrarian risk: consensus may be over-anchoring on the headline growth rates and underweighting mix drag plus customer concentration. The thesis breaks if gross margin cannot hold the current range or if the October 6 Investor Day fails to add incremental FY29+ custom detail beyond what is already embedded. If AI capex pauses, this becomes a long-dated promise trade, and the market will quickly haircut the warrant narrative.

Key Decisions for Investors

• Buy MRVL on a 3%-5% post-earnings pullback; use the October 6 Investor Day as the next catalyst. Falsifier: gross margin slips below the guided band or management does not raise FY29 visibility.
• Use MRVL Jan-2027 call spreads rather than outright stock if you want upside with defined risk; the setup is best for a volatility reset followed by an Analyst Day re-rating.
• Tactical relative value: long MRVL / short NVDA in small size only as a hedge against AI spend rotating from compute to networking and custom attach; cover if NVDA networking attach or scale-up optics commentary re-accelerates.
• Watch GOOGL as an indirect beneficiary of lower AI infrastructure cost, but wait for capex commentary before adding exposure; the stock benefits only if custom silicon improves cloud margins faster than it raises spend.