Marvell (MRVL): Does Jensen's Trillion-Dollar Call Hold Up?
When the CEO of the most important company in AI points at you and says "next trillion-dollar company," the market listens. Our model? It nodded, then frowned at the price tag.
Why we're even talking about this
If you blinked on June 2, you missed it. Jensen Huang called Marvell the next trillion-dollar firm at Computex, and the stock ripped 32.52% — Marvell's biggest one-day gain ever, narrowly topping its previous best from May 2023. That move took the market cap to just over $250 billion. Cute number, but a long way from thirteen zeros.
The endorsement isn't a press release flourish. Nvidia recently committed a $2 billion investment into Marvell, and is also funding other photonics outfits that move data with light instead of electricity — a not-so-subtle hint about where the AI bottleneck is heading. Huang's pitch is that as clusters get huge, the connections between chips become as valuable as the chips themselves. Marvell makes those connections.
So is this a $1T-in-waiting, or a great story that's already priced in? That's what the three gates are for.
The model run, three gates
Survival — 100/100. Nothing to argue with here. Marvell delivered record fiscal 2026 revenue of $8.195 billion, GAAP net income of $2.670 billion ($3.07 per diluted share), and non-GAAP EPS of $2.84. Q1 FY27 produced record operating cash flow of $639 million, with $200M of buybacks and $54M of dividends returned in the quarter. Debt-to-equity sits at 0.29 with a current ratio of 3.28. This is a profitable, cash-generative business with a clean balance sheet. "Quality Compounder" fits.
Growth — 72/100. Strong, with a but. The top-line trajectory is genuinely impressive: fiscal 2026 revenue grew 42% YoY, with non-GAAP EPS up 81% YoY, and Q1 FY27 delivered record revenue of $2.418 billion — up 28% YoY — with data center revenue setting a record at $1.833 billion. Management now expects full-year fiscal 2027 data center revenue to grow roughly 50%, led by interconnect up more than 70%, with scale-out switch revenue forecast to exceed $600 million in fiscal 2027 and reach $1 billion in fiscal 2028. The score isn't higher because much of that growth is concentrated in a small number of hyperscaler custom-silicon programs — great when they ramp, painful if one slips.
Valuation — 39/100. This is where the verdict gets written. The trailing P/E is 91, the forward P/E is 58, EV/EBITDA is 85.5 and EV/FCF is 139. The stock is up more than 2.3x year to date, with price-to-sales re-rating from under 10x in January to near 22x today. You can talk yourself into those multiples if the FY28 guide lands, but "if the guide lands" is doing a lot of work.
The bull case — and what would actually have to happen
The bull math isn't crazy. Marvell has guided fiscal 2027 revenue to approach $11 billion (more than 30% growth) and fiscal 2028 to approximately $15 billion, with non-GAAP EPS "well over $5." If you believe that, the forward multiple compresses fast. Management has also flagged that the custom chip business alone could generate more than $10 billion in annual revenue by fiscal 2029.
The strategic position is real, too. Marvell has expanded its collaboration with Nvidia across optics, NVLink Fusion and AI RAN, and bolted on Polariton, Celestial AI and XConn to accelerate scale-up optics, custom silicon and switching. If the interconnect layer is the next bottleneck — and the photonics dollars from Nvidia suggest the industry thinks it is — Marvell sits on the right tile.
The bear case our valuation gate is sniffing
Two things. First, scale. Broadcom does over $63B in revenue versus Marvell's roughly $8B, and trades around 41x forward 2026 earnings versus Marvell at 24x on that same lens. Marvell trading at a discount to Broadcom is rational; Marvell trading at Broadcom-like dollars per share of FY28 hope is where the model gets twitchy. When Broadcom crossed the $1T line, it was already within striking distance; Marvell is operating from a much lower starting point.
Second, the endorsement itself. "Next" is doing a lot of heavy lifting — at a $264 billion valuation, Marvell remains far from the trillion-dollar club, and Huang's comments are better read as a statement about opportunity than a forecast that a trillion-dollar cap is imminent. That's not a knock on the company. It's a reminder that a CEO shout-out is not a discounted cash flow.
What would flip our read? A demonstrable jump in gross margin toward the high-50s/low-60s without giving back data-center share, or simply more time at this revenue run-rate so the multiples grow into the price rather than the price growing into the multiples. A material guidance cut, or a failure to clear the high end of growth expectations in the next two quarters, would do the opposite.
The bottom line
MEH isn't dismissive — it's the model saying "good company, full price." Survival is pristine. Growth is one of the better stories in semis. Valuation is where bulls and bears are really fighting, and at a 39, our gate sides with the bears on the margin. The 1-year analyst consensus target sits around $235.70, which is roughly where the stock already trades — i.e., even the sell side isn't underwriting another leg from here without new news. Jensen may well be right about the destination. The question this scan is asking is about the entry point. Not advice — just the numbers as we read them.