Signal of the Week / Week of August 31, 2026
Record quarter. Guidance raised twice. The stock fell 10%.
Marvell posted its highest revenue ever, beat on both lines, and lifted its outlook for two consecutive fiscal years. The market sold it anyway, on gross margin. But the margin that supposedly disappointed came in above the midpoint of guidance, and the reason it compresses next year is that Marvell is winning the business it wants. That is the setup, and it is why this one gets defined risk before it gets a target.
What actually happened Thursday night
Start with the quarter, because it was a very good one. Revenue of 2.739 billion dollars was a company record, up 37% year over year and 13% sequentially, and it landed 39 million above the midpoint of guidance. Non GAAP earnings of 0.94 per share were up 40%. Operating margin expanded to 36.6%. Cash flow from operations came in at 605.5 million, and the company returned 254 million to shareholders including 200 million of buybacks.
Data center revenue reached 2.172 billion, up 46% year over year and 18% sequentially, and now represents 79% of total sales against 74% a year ago. This is no longer a diversified chip company with an AI segment. It is an AI infrastructure company with a legacy business attached.
Then the guidance. Third quarter revenue is guided to 3.15 billion plus or minus 5%, up 15% sequentially and better than 50% year over year, with non GAAP earnings of 1.05 to 1.15. Full year fiscal 2027 was raised to approximately 12 billion, about 45% growth, up from a prior outlook near 11.5 billion. And fiscal 2028 was raised to approximately 18 billion, roughly 50% growth, up from 16.5 billion.
The stock fell about 10%.
Two of the most important companies in AI infrastructure reported 24 hours apart and said the same thing. Demand is not the constraint. Supply and capacity are. Nvidia is prepaying for it. Marvell plans roughly 1 billion dollars in supplier capacity prepayments this fiscal year for exactly the same reason.
The margin story, read properly
above the midpoint of the company’s own guidance
raised from approximately 16.5 billion
expanded Tier 1 hyperscaler agreement, milestone dependent
Here is the part the headlines skipped. The quarter’s non GAAP gross margin of 58.9% came in slightly above the midpoint of guidance. The company did not miss on margin. What spooked the market is the forward band, which sits below trailing levels because the revenue mix is shifting toward custom silicon.
Custom silicon carries a lower gross margin than merchant silicon. That is simply true. But it also carries scale, multi year design win visibility, and customer lock in that merchant parts never provide. Marvell is trading percentage points of gross margin for a much larger and much stickier revenue base, and management guided fiscal 2028 operating expenses to grow at roughly half the rate of revenue, pushing operating margin toward the upper end of its 38 to 40 percent target.
Lower gross margin, higher operating margin, far more revenue. That is not deterioration. That is a business model choice, and the market priced it as a warning.
And then there is the agreement almost nobody led with. Marvell disclosed an expanded commercial arrangement with a Tier 1 hyperscaler covering custom AI inference accelerators, storage controllers, network interface cards and memory interface controllers, with potential revenue of approximately 120 billion dollars over six and a half years if all milestones are met.
What can break this trade
The demand thesis was independently confirmed by Nvidia 24 hours earlier, and Marvell’s own numbers agreed with it. Record revenue, data center up 46%, and outlooks raised for two consecutive fiscal years is not a company with a problem.
The selloff came on forward mix, not on execution or demand. A 10% drawdown in a name whose FY28 revenue outlook just went up by roughly 1.5 billion dollars is a repricing of margin assumptions, and those are the kind of assumptions that get revisited quarter by quarter.
The valuation had a lot in it. The stock ran hard into this print after a very strong year, and a name trading at a rich forward multiple gets punished violently when any assumption moves against it. A 10% drop does not automatically make it cheap.
The margin compression is real, not imagined. Custom silicon genuinely earns less per dollar, and if the mix shifts faster than operating leverage arrives, the model gets worse before it gets better. Roughly 1 billion in capacity prepayments consumes cash up front. The 120 billion figure is milestone dependent over six and a half years, not booked revenue, and it concentrates a large share of the future in a single customer relationship.
So this is a name to size deliberately rather than aggressively. A stock that just moved 10% on a guidance nuance can move 10% again on the next one. Position size is the risk control here, and the stop is not optional.
The trade specifics
Members receive the full alert with entry zone, protective stop, profit targets and the exact contract before the position is opened, along with our plan for trading the post earnings drift in this name.
Every MTOptions alert carries a defined entry, a defined exit and a protective stop. On the winners and on the ones that go against us. That is what 25 years of published trade logs looks like.
Figures referenced are from Marvell Technology’s second quarter fiscal 2027 results and management commentary released August 27, 2026, and from NVIDIA Corporation’s fiscal second quarter results released August 26, 2026. Forward outlooks are company guidance and are subject to revision. The referenced hyperscaler agreement figure represents potential revenue over approximately six and a half years contingent on milestones and is not booked revenue. Prices and market reaction figures are as of the August 28, 2026 session and will change.
MTOptions is a publication of Pinpoint Financial Group LLC. Content is provided for educational and informational purposes only and is not individualized investment advice or a recommendation to buy or sell any security. Options carry substantial risk and are not suitable for every investor. Past performance does not guarantee future results.
