$MRVL: Marvell Reported a Perfect Quarter. The Stock Fell Anyway. Here Is What the Market Is Actually Saying.
test
Marvell Technology ($MRVL) reported fiscal Q2 2027 results on August 27. Revenue came in at a record $2.739 billion, up 37% year over year and 13% sequentially. Data center revenue hit a record $2.17 billion, up 46% year over year. Non-GAAP earnings per share of $0.94 beat management's own guidance midpoint by a penny and rose 40% year over year. The company raised its full-year fiscal 2027 revenue outlook to approximately $12 billion and its fiscal 2028 outlook to approximately $18 billion, representing 45% and 50% compound growth respectively. And CEO Matt Murphy disclosed an expanded commercial agreement with Google covering custom XPUs, inference accelerators, storage controllers, network interface controllers, and near-memory compute, with cumulative revenue potential that Murphy described as in the range of $120 billion over six years.
The stock fell roughly 7% after hours.
That is the setup. Not a bad quarter, a perfect one by most conventional measures. And the reaction was negative. Understanding why requires understanding what investors actually owned going into this print.
What the market already knew
Marvell entered this week up approximately 187% year-to-date. That is not a run that prices in a good quarter. It is a run that prices in a decade of AI infrastructure dominance, a dominant position in custom silicon for the world's largest hyperscalers, and a gross margin profile that expands as scale arrives. When you own a stock at those levels, an earnings beat is confirmation, not a catalyst. The question the market was asking was: did the future just get materially bigger, or did it just confirm what we already thought?
The honest answer: incrementally bigger, but not by enough to justify another leg higher. The Google commercial agreement is strategically significant but its major revenue impact lands in fiscal 2029 and beyond. Fiscal 2027 and 2028 guidance was raised, but the raises were in line with what buyside models had already embedded after prior upside prints. And gross margin guidance for fiscal Q3 stepped down to 57.5%-58.5% from 58.9%, a signal that the Custom silicon ramp, while accelerating, creates near-term margin pressure as volume scales before pricing power normalizes.
So the after-hours move was not a verdict on the quarter. It was a verdict on the price paid for a future that moved further away.
The Google deal: what $120 billion actually means
The most discussed item on the call was the expanded commercial agreement with Google. Murphy was careful with the language, describing the engagement as "very broad-based," covering "a number of product lines including inference accelerators, storage controllers, NICs, memory interface controllers, and near-memory compute." He said Marvell "pioneered the XPU attach category" and that "all projections to date have been under-called."
The warrant and commercial structure for the agreement through fiscal 2028 is already reflected in current revenue targets. The $120 billion cumulative figure Murphy described is the envelope for fiscal 2029 and beyond. Murphy deferred detailed sizing to Marvell's Investor Day on October 6, 2026, calling it "massive" and "game-changing" for the company over "the next 6 to 6.5 years."
For investors with a two-year horizon, the operative number from this disclosure is not the headline $120 billion. It is that the Custom business, which is expected to more than double in fiscal 2028, already had a prior fiscal 2029 target of $10 billion-plus that Murphy said now has "significant upside bias." The exact revision to that target, along with the fiscal 2030 through fiscal 2032 buildout, will be the story at the October Investor Day.
The gross margin signal
The one number that genuinely surprised the market was the Q3 gross margin guide. Marvell posted 58.9% non-GAAP gross margin in Q2. For Q3, the midpoint is 58%, and the range is 57.5%-58.5%.
The reason is product mix. The Custom business, which is ramping at significant scale beginning in the second half of fiscal 2027, carries lower initial gross margins than Marvell's established connectivity and ethernet products. Custom silicon is designed to the customer's specification, and first-generation programs price differently than mature, high-volume commodity products. Management expects this pressure to normalize over time, and is still targeting a non-GAAP operating margin range of 38%-40% by Q4 of fiscal 2027.
The read-through: margins are not deteriorating structurally. They are compressing temporarily because the fastest-growing segment of the business is in its early ramp phase. The question is how long "temporary" turns out to be.
There is also a non-data-center story in the revenue mix that is easy to miss.
The segment no one is talking about
Marvell's communications and other segment generated $568 million in Q2, up 10% year over year but down 3% sequentially. For Q3, management is guiding this segment down low-to-mid teens both sequentially and year over year. This is 21% of current revenue declining in a quarter when the headline is "50%-plus year-over-year total growth."
The communications segment includes carrier infrastructure, enterprise networking, and industrial applications. These markets are in a different cycle than data center AI. Carrier capex has been soft, enterprise networking digested elevated inventory from the post-COVID cycle, and industrial demand remains uneven. The segment is not small enough to ignore at $2.27 billion in annualized revenue.
The full story of Marvell's near-term earnings power is data center growing north of 60% in fiscal 2027 while communications declines mid-teens, and the aggregate blends to that 45%-growth headline. It is a strong print, but a two-speed company with one engine on fire and one in repair mode.
Briefings like this land in members' inboxes before the market prices them in. Join free
The question the selloff asks
The reaction Thursday evening was rational. After a 187% year-to-date run, investors who bought on AI infrastructure thesis are now asking whether Marvell at this price is a hold or a trim. The gross margin dip gave permission to take profit. The Custom timing pushed the really big numbers two to three years out. And management transitions -- a new CFO and new IR leadership -- introduced execution uncertainty at an operationally critical moment.
None of that means the thesis is broken. It means the re-rating that took the stock from post-2022 lows to a $215 billion market capitalization has compressed the margin of safety. The same company that looked cheap in 2023 at 15x earnings now trades at the multiple that only makes sense if fiscal 2028 and fiscal 2029 land at or above guidance.
Behind the paywall, we work through the Custom revenue timing analysis, the gross margin trajectory and operating leverage math, the dilution picture from the March 2026 convertible preferred issuance, and the three scenario-by-scenario price zones that frame the distribution of outcomes from here.
The rest of this briefing is for paid members: the Custom revenue timing analysis and what $120 billion over six years actually means for near-term holders, the gross margin compression math and when it reverses, the dilution picture from the March 2026 convertible preferred, and the bull/base/bear scenario price zones.
AlphaBriefing Paid gets you every investment thesis, scenario framework, and catalyst brief we publish -- the analysis private intel clients pay four figures for, at a fraction of that.