Marvell Q2 FY2027 Earnings — A Guided Quarter Decided by Gross Margin and the Google Deal's Fine Print
The Call
Revenue is not really in question this quarter — Marvell guided $2.7B ±5% at the Q1 call and consensus sits almost exactly on that midpoint, so the street has embedded no beat/miss premium at all. What actually moves the stock is whether non-GAAP gross margin holds inside the 58.25-59.25% guided band (a standing bear watch trigger fires below 58% for two consecutive quarters), whether management puts any near-term dollar figure on the Google warrant deal disclosed three weeks ago, and whether the independently-sourced (non-IR) reporting that Marvell lost the AWS Trainium3/4 backend design to Alchip gets addressed with anything more specific than a general denial.
| Metric | Our Estimate | Company Guidance | Consensus |
|---|---|---|---|
| RevenueGAAP +35.8% YoY vs Q2 FY2026's $2,006.1M. Guide issued 2026-05-27; consensus sits on the guide midpoint. | 2,725 USD M | 2,700 (±5%, 2,565-2,835) USD M | 2,710 USD M |
| Gross profitGAAP Revenue × GAAP gross margin estimate. GAAP gross margin is not separately guided by the company (only non-GAAP GM is guided) — see the non-GAAP gross margin businessMetric for the guided/consensus-comparable figure. | 1,425 USD M | — | — |
| Gross marginGAAP Four-quarter GAAP trend extrapolation, cross-checked against the non-GAAP guide less Q1 FY2027's realized GAAP/non-GAAP gap. | 52.3% | — | — |
| Operating profitGAAP Q1 FY2027 operating profit ($339.4M) bridged forward with a ~30% incremental margin on the ~$307M of incremental revenue. | 432 USD M | — | — |
| Operating marginGAAP | 15.9% | — | — |
| Net incomeGAAP Operating profit less an anchored ~$300M GAAP-to-non-GAAP bridge (SBC + acquisition-related intangible amortization + net interest + tax), held near Q1 FY2027's realized $304.9M gap. | 132 USD M | — | — |
| Diluted EPSGAAP Net income / ~897M estimated diluted shares. GAAP EPS is structurally depressed by purchase-accounting amortization and SBC — see the non-GAAP diluted EPS businessMetric ($0.97 est vs $0.93 consensus) for the figure the market actually prices the print against. | 0.15 USD | — | — |
- → Non-GAAP gross margin prints below 58% — triggers the standing two-consecutive-quarter bear watch and signals more mix pressure from the lower-margin custom-silicon ramp than guided, likely dragging GAAP EPS well below the $0.15 estimate even if revenue holds near guide.
- → Management quantifies near-term (FY2027) revenue from the Google warrant deal — none of the external evidence found puts a dollar figure on Google's contribution before FY2028; a material FY2027 number disclosed on the call would mean the revenue and Data Center estimates here are too low.
- → Management confirms, with specifics, the SemiAnalysis/Benchmark reporting that Marvell lost the AWS Trainium4 backend design to Alchip — a live, independently-sourced (non-IR) risk to the ~50% FY2027 Data Center growth guide that the company has so far only addressed with a general denial ('we didn't lose any business'), not specifics.
- → Teralynx T100 CPO is disclosed as having lost a hyperscaler socket to Broadcom's already-shipping Tomahawk 6/Davisson CPO switch — this is the single most-watched open bull trigger, and a loss (rather than continued silence or a win) would compress the >70% FY2027 interconnect growth guide.
Behaviour model, not arithmetic, and deliberately low confidence. Built from a normal sequential guide step (Q2's guide was ~12% above the Q1 print) applied off this report's Q2 estimate — seasonally reasonable for an H2 ramp into the ~$11.5B FY2027 total. This pipeline's guidance calls have come in low twice already this cycle on other issuers (STX FQ1 FY2027 -9.3%, AMD Q3 2026 -3.1%), both understating management in an up-cycle, and Marvell just added a fresh, market-moving customer catalyst (the Google deal) that a mechanical sequential-step model cannot anticipate — so the actual guide is more likely to surprise above this figure than below it.
Evidence (8)
Positioning Suggestions
→ Consensus ($2.71B revenue, $0.93 EPS) sits almost exactly on Marvell's own guidance midpoint — there is no embedded beat/miss premium in the street, so this print carries binary-outcome risk rather than a graded one. Size accordingly, especially with the stock trading near the top of its 25-45x historical multiple band after the Google-driven rally.
→ AVGO already fell ~5% on the Google-deal headline and reports its own print after Marvell (2026-09-02) — a small MRVL long / AVGO short pair captures the custom-silicon share-shift narrative with less single-name earnings risk than a standalone MRVL position into the print.
→ Non-GAAP gross margin is the metric most likely to move the stock independent of revenue, given the standing sub-58%-for-two-quarters bear trigger — weight the margin line over the headline revenue beat/miss when sizing a post-print reaction.
Predictions
8MRVL Q2 FY2027 revenue meets or beats the $2.7B guidance midpoint
MRVL non-GAAP diluted EPS beats the $0.93 consensus estimate
MRVL non-GAAP gross margin prints at or above 58.5%
Management discloses a named hyperscaler design win for the Teralynx T100 CPO switch on the call
If confirmed, the interconnect bull case gains its first hard evidence point — consider adding to MRVL / trimming AVGO on the relative-value pair.
Management quantifies specific FY2027 (near-term) revenue contribution from the Google custom-silicon warrant deal
Management directly addresses (confirms or specifically denies with detail) the AWS Trainium3/4 design-loss-to-Alchip reporting on the call
MRVL raises its FY2028 revenue outlook above the current $16.5B
Q3 FY2027 revenue guidance midpoint comes in above $2.90B
A guide meaningfully above $2.90B alongside the Google catalyst would be the clearer signal to add than the Q2 print itself.
Key Questions
- Does non-GAAP gross margin hold at or above 58%, or does the standing two-consecutive-quarter bear watch trigger fire?
- Does management put any dollar figure on the Google warrant deal's FY2027 revenue contribution, or confirm it is entirely a FY2028+ ramp?
- Does the Teralynx T100 CPO switch get a named hyperscaler design win, or does the silence extend into a third quarter?
- Does management address the SemiAnalysis/Benchmark reporting of an AWS Trainium3/4 design loss to Alchip with any specificity, beyond the CEO's general denial?
- Does the Data Center segment's YoY growth rate actually accelerate toward the ~50% FY2027 target, or does it stay closer to Q1's 27%?