MRVL
NVDA
Jensen Huang just turned up the volume
Marvell Technology got the kind of shoutout that makes Wall Street sit up straight. At Computex 2026 on June 1st, Nvidia CEO Jensen Huang called Marvell a potential “next trillion-dollar company” while appearing onstage with Marvell CEO Matthew Murphy in Taipei.
The market did what the market does: it heard “AI infrastructure,” saw a shiny new storyline, and sent the stock roughly 17% higher overnight before the gains kept rolling into Tuesday. Cute? Sure. But also meaningful. Because behind the headline-grabbing compliment is a business that has actually been growing like it means it.
Not just a one-night pep rally
Marvell isn’t riding on vibes alone. The company posted record fiscal 2026 revenue of $8.195 billion, up 42% year over year, then followed that with a record Q1 fiscal 2027 showing: $2.418 billion in revenue and $638.8 million in operating cash flow.
And the company didn’t exactly whisper its outlook, either. Marvell guided Q2 fiscal 2027 revenue to $2.7 billion at the midpoint and raised its revenue outlook for fiscal 2027 and fiscal 2028. In other words: the bridge to “trillion-dollar company” is long, but management is at least laying down some fresh asphalt.
The Nvidia tie-up is doing real work
This is where the story gets less celebrity cameo and more strategy chess. On March 31, 2026, Nvidia and Marvell announced a partnership linking Marvell to Nvidia’s AI factory and AI-RAN ecosystem through NVLink Fusion, with Nvidia also investing $2 billion in Marvell as part of the deal.
That matters because it puts Marvell inside the dominant AI compute stack instead of outside it looking in. Marvell supplies custom XPUs and compatible scale-up networking, while Nvidia brings the rest of the toolbox:
- Vera CPUs
- ConnectX NICs
- BlueField DPUs
- Spectrum-X switches
For hyperscalers, that means semi-custom AI infrastructure with fewer blank sheets of paper. For Marvell, it means a seat at a very expensive table.
Why investors are watching so closely
Marvell’s data center business is now the heavy lifter, making up 76% of total revenue in Q1 fiscal 2027. The company also designs chips for Amazon’s Trainium and Inferentia programs and Microsoft’s Maia accelerator, while reportedly talking with Google about two more AI chips.
That’s the kind of customer lineup that can make a chip company look less like a one-hit wonder and more like infrastructure plumbing. Broadcom is still the heavyweight in custom AI chips, but Marvell’s growing role suggests the market may be opening up faster than people thought.
Big picture: Huang’s praise didn’t create Marvell’s AI story — it just gave it a brighter stage light. And in this market, stage light can turn into valuation fuel very quickly.