Jensen Huang Just Named Marvell the Next $1 Trillion Stock. Is the Stock a Buy Following a 129% Surge?

Marvell stock has surged 129% over the past three months on investors’ focus on AI-infrastructure demand, with an added catalyst after Nvidia’s Jensen Huang suggested Marvell could be next to join the trillion-dollar club. The article cites Goldman estimates that custom ASIC shipments could reach parity with GPU sales by next year, while Marvell expects custom ASIC revenue to more than double in the next fiscal year (vs ~+20% this fiscal year) and projects a 70% increase in data center interconnect revenue; switching is expected to reach $1B in fiscal 2028 from $600M this year. Valuation is flagged as stretched (trailing P/E ~94, forward P/E ~67, P/S ~27), but the bullish case hinges on continued high growth to justify the premium.

Analysis

MRVL is now trading like a scarce AI-infrastructure option, not a normal semiconductor name. That means the stock can keep working if hyperscaler capex broadens, but the bar for incremental upside is high: the market is already paying for a long runway, so any sign that design wins are concentrated, delayed, or lower-margin than expected would trigger fast multiple compression. The real competitive tension is not just versus other chip designers; it is also between custom silicon and merchant solutions, where the winner may be the vendor that can bundle silicon, optics, and switching into a lower-friction platform.

Second-order beneficiaries sit in the optical bottleneck: names exposed to high-speed interconnect content can get a longer demand tail than pure accelerator suppliers because network upgrades are required even if GPU/ASIC growth pauses. Conversely, if hyperscalers standardize around fewer architectures, suppliers with weaker scale or less software adjacency can lose pricing power quickly. Broadcom is the closest quality comp, but it has a better ability to absorb mix shifts and protect margins, which matters if the market starts rewarding durability over pure growth.

The near-term setup is momentum-driven over days, but the 1-3 month risk is digestion after a parabolic rerate. What would falsify the bull case is not macro noise; it is a guide that fails to show acceleration in custom ASIC or interconnect revenue, or evidence that backlog is not converting into shipments at the promised rate. Over 6-18 months, the key question is whether MRVL becomes a repeatable platform winner or remains a cyclical beneficiary of one AI spend wave.

Key Decisions for Investors

• Do not chase MRVL after the recent rerate; wait for a 10-15% pullback or a post-earnings consolidation before initiating exposure.
• If establishing exposure, use a 6-9 month call spread on MRVL rather than cash equity to cap valuation risk and preserve upside if AI demand re-accelerates.
• Relative-value: long CRDO or COHR against a smaller MRVL position if you want purer exposure to the optical bottleneck, since network content may have a longer tail than custom silicon alone.
• Use AVGO as the higher-quality AI infrastructure hedge against an MRVL long; if MRVL execution wobbles, AVGO should prove more resilient on margin and free-cash-flow durability.
• Set a fundamental alert: if the next guide does not show clear sequential acceleration in data-center revenue or margin leverage, treat the AI narrative as fully priced and reduce exposure.