MRVL
Marvell’s AI party got a little louder
Marvell is back with another reminder that it wants a bigger seat at the AI infrastructure table. The company rolled out second-gen Structera X memory products, flashed a wider infrastructure portfolio, and kept waving around a long-term revenue target that suddenly looks maybe, possibly, not ambitious enough.
The new memory gear is built to help cloud and enterprise customers wrangle larger AI workloads without having to rip out half their existing setup like you’re remodeling a kitchen with a sledgehammer. Marvell says the products can speed up responses for large AI models by 40%-50%, or roughly double context at similar speeds versus the prior version.
The number that caught Wall Street’s eye
That’s where BNP Paribas analyst Karl Ackerman came in. He said Marvell could generate about $33.40 a share in earnings by 2030, which is above management’s own target of more than $30 and the current consensus estimate of $24.91.
That’s a pretty strong flex, especially since Marvell already raised its long-term outlook at investor day:
- fiscal 2028 revenue outlook: about $20 billion
- fiscal 2031 revenue range: $70 billion to $90 billion
- gross margin target: 56%-59%
- operating margin target: 44%-46%
In other words, the Street is now arguing Marvell may have been conservative while the company was already sounding pretty confident. Classic Wall Street: never satisfied, always revising the spreadsheet.
Why investors still care
The stock was down 3.22% Monday as some traders took profits after the recent AI sprint. But the bigger picture is that Marvell is trying to sell investors on a full-stack AI infrastructure story — memory, networking, storage, connectivity, security, the whole buffet.
And if it can keep turning those partnerships with AMD, Arm, Alibaba, Samsung Electronics and SK Hynix into real adoption, the market may keep giving it credit for being more than just another chip name riding the AI wave.
Big picture: Marvell’s AI ambition is getting so large that the debate has shifted from “Can it grow?” to “How much growth did management leave on the table?”