Marvell (MRVL) Q2 FY27: The $18 Billion FY28 Guide Already Contains the Google Revenue
Marvell closed Thursday down 1.30% at $241.93 and then surrendered another 7.09% in the aftermarket to $224.34, on a quarter that beat every line management guides to. Revenue of $2.739 billion was a record, up 37% year over year and 13% sequentially, against consensus near $2.71 billion. Non-GAAP EPS of $0.94 cleared $0.93. Data center revenue reached $2.17 billion, 79% of the total, up 46%. The third-quarter guide of $3.15 billion plus or minus 5% sits roughly 4% above the street, and the $1.10 EPS midpoint clears $1.07. Management raised fiscal 2027 revenue by $500 million to approximately $12 billion and fiscal 2028 by $1.5 billion to approximately $18 billion — the second consecutive $1.5 billion raise to the FY28 number in two quarters. None of that is what the print was about.
The question carried into August 27 was whether Marvell would disclose Google custom-products revenue as its own reported series. It did not, and the reason it did not is the more consequential disclosure. Asked directly about the custom opportunity for FY29, Murphy said revenue from the warrant is already reflected in fiscal 2028 guidance and that the material impact arrives in fiscal 2029 and beyond. That single sentence converted the July 29 agreement from an upward revision waiting to happen into an input already spent. The market had spent nine sessions since August 18 pricing the warrant as incremental. Management priced it as included.
What actually produced the $1.5 billion raise is more interesting than the Google line, and the sell-side attribution on the call is unambiguous: scale-up optics. Murphy identified a meaningfully higher FY28 outlook for scale-up optics as a key driver, with NPO and CPO both in adoption and CXL described as deploying at multiple hyperscalers for memory expansion and inference. Custom is guided to more than double next year, but the raise was called broad-based rather than custom-led. This matters because it is the attach franchise, not the accelerator socket, that carries the durable economics. A hyperscaler that in-sources its compute die still buys SerDes, optical DSP, retimers, and the 800G and 1.6T interconnect around it. The scope of the Google agreement itself concedes the point — inference accelerators alongside storage controllers, NICs, memory interface controllers, and near-memory compute. Most of that list is attach. The warrant is buying the perimeter, not just the die, which is the correct read of why Nvidia took roughly 2.5% in convertible preferred: exposure to Marvell wins containing no Nvidia silicon.
The margin line is where the mix arrives. Q2 non-GAAP gross margin of 58.9% came in near the top of guidance. The Q3 midpoint of 58.0% is 90 basis points lower sequentially, and that is custom silicon entering the mix at custom silicon’s margin structure. Durn’s answer is that operating margin enters the 38% to 40% long-term range in Q4 and reaches the high end through FY28, with FY28 operating expenses growing at roughly half the rate of revenue. Read plainly: Marvell is trading gross margin for volume and defending earnings with operating leverage below the gross line. That works while revenue is accelerating. It is brittle if revenue merely grows. R&D rose 43% to $741.1 million and stock-based compensation more than doubled to $326.2 million in the quarter that is supposed to be demonstrating restraint.
The balance sheet is doing work that does not show up in EPS. Operating cash flow of $606 million fell sequentially, and the reason given was higher prepayments — Marvell remains on pace for approximately $1 billion of capacity prepayments to suppliers in FY27 against what Murphy called pervasive industry-wide constraints. Cash ended at $3.93 billion against $4.96 billion of long-term debt. Growth is being pre-funded, in cash, into a supply chain that is rationing. That is a reasonable use of the balance sheet and it is also the mechanism by which a demand disappointment would arrive as a working-capital problem before it arrives as a revenue problem.
At $224.34 the company is worth roughly $189 billion, about 10.5 times the FY28 revenue it just guided to. Running the FY28 arithmetic on management’s own framework — $18 billion at the high end of a 38% to 40% operating margin, net of interest and a low-teens tax rate — produces something near $7 per share, so the aftermarket price is approximately 31 times a number two fiscal years forward that requires the custom ramp to arrive on schedule. The stock is up roughly 184% year to date and 222% over twelve months, joined the S&P 500 in June, and set an all-time high of $329.88 that same month. The current street average sits at $295.42 with no sell ratings. That is not a distribution that leaves room for a good quarter.
Base case is $220 to $270 into the October 6 investor day, which is now carrying more weight than the print did — management explicitly declined to give an FY29 framework and deferred it there, while conceding that the $10 billion-plus FY29 custom target is likely higher. Bull case retests $329.88 on an FY29 number that breaks custom into compute and attach and shows the attach line growing independently of socket wins. Bear case is cohort derating rather than anything company-specific: Broadcom, Astera Labs and Marvell all reprice together on the same forward multiple compression, and a move to 20 times the same FY28 earnings puts Marvell at $140 to $150 without a single guidance cut.
The decision-relevant disclosure is in the Q2 10-Q, not the press release: the accounting treatment of the warrant. Each of the 240 tranches of 240,042 shares vests against $500 million of recognized Google custom-products revenue. If the fair value of vested tranches is recognized as consideration payable to a customer, it reduces reported revenue rather than sitting in equity or operating expense — and it is measured against a share price that rises when the thesis works. In that case the approximately $18 billion FY28 figure is quoted gross of a charge that scales with the stock, and every $500 million of Google revenue that validates the franchise books at less than $500 million. The footnote answers whether the dilution schedule is a cap table event or an income statement one.