BofA raises Marvell stock price target to $200 on AI networking growth

BofA raised Marvell’s price target to $200 from $125 and increased its AI networking TAM forecast from roughly $6 billion to $14 billion for 2026-2030, implying stronger long-term demand for Ethernet transceivers and DSPs. The firm estimates Marvell could capture 60% to 70% share in 800G and 1.6T generations, potentially adding more than $1.2 billion in DSP opportunities. Additional support came from Marvell’s Polariton acquisition and recent target hikes from RBC, Barclays, and other brokers.

Analysis

This is less about Marvell’s near-term print and more about the market repricing the duration of AI interconnect demand. If the networking TAM expansion holds, the biggest second-order winner is not just MRVL’s optics DSP franchise but the broader capex stack: fiber, lasers, test equipment, and the contract manufacturers that can actually ramp 800G/1.6T volume without yield deterioration. The implication is also mildly negative for vendors exposed to slower-cycle legacy switching spend, because budget is shifting from compute adjacency into connectivity attach rates.

The key risk is that the market is extrapolating a clean conversion from TAM to earnings before the production curve is proven. Optical attach historically looks linear in slides and lumpy in shipments; any slip in cloud deployment cadence, yield issues, or custom accelerator mix shifts could compress the 2026-2028 ramp and force multiple compression in a name already re-rated on expectation. In other words, the upside is real over 12-24 months, but the first derivative can still be choppy over the next 1-2 quarters if investors decide this is another AI infrastructure “promise now, monetize later” story.

There is also a subtle competitive dynamic: if Marvell’s share assumptions prove right, the real loser may be the second-source ecosystem, not the marquee hyperscalers. That argues for caution on names whose bull case depends on broad AI networking share gains without differentiated silicon content. The China-related Nvidia catalyst is directionally supportive for semis sentiment, but it does little to change Marvell’s core debate; if anything it reinforces that the market is willing to pay for strategic AI exposure even when the end market is policy-constrained.

Key Decisions for Investors

• Stay long MRVL into the next 3-6 months, but finance part of the position with covered calls or a collar; the thesis is still improving, but the stock is vulnerable to valuation air pockets if execution evidence lags TAM rhetoric.
• Pair trade: long MRVL / short a slower-growth networking or legacy switching peer over 6-12 months; the relative value is strongest if optical attach accelerates while non-AI infrastructure spending remains flat.
• Buy 6-12 month call spreads on MRVL rather than outright calls to express upside from 800G/1.6T adoption while limiting multiple-risk if the market derates overvalued AI names.
• Use pullbacks after any headline-driven spike to add, not chase; best entry is on 5-8% weakness tied to broader semiconductor volatility, since the catalyst path is multi-quarter rather than immediate.