Marvell Technology, Inc. (MRVL)

The panel's take

Verdict: WAIT · Conviction: MEDIUM · Last price: USD 278.20 (as of 2026-07-01)

Marvell is the batch's most genuinely two-sided debate — a real, accelerating AI-silicon franchise whose stock has simply run ahead of it. The business is firing: fiscal Q1 revenue rose 28% to a record $2.418B, data center is now 76% of sales, management raised the FY2027 outlook to ~$11.5B (+40%) and introduced a ~$16.5B FY2028 target, and Nvidia's Jensen Huang called Marvell "the next trillion-dollar company" on stage at Computex — a moment that added ~32% in a day. Marvell and Broadcom together own ~95% of the custom-ASIC co-design market, a real duopoly. But the price embeds most of that: ~66-70x forward earnings (roughly double its own five-year average and ~80% above the semis median), after a ~5x run from $60 to a ~$340 high, now digested back ~18% to $278. And the bears have specifics: GAAP EPS was just $0.04 against $0.80 non-GAAP — a ~20x gap from stock comp and acquisition amortization; Marvell is the second source behind Broadcom (which is growing AI revenue faster at higher margins); its next-gen Amazon Trainium3 design was reportedly lost to a smaller rival, Alchip; top-10 customers are ~82% of revenue; and the CEO has sold into the rally at every step. Two frameworks are outright bullish (the innovation and macro lenses, riding a structural hyperscaler-capex wave), three are bearish on price, and the valuation lens pegs fair value at ~$235-250 — just below the quote, with the analyst median target ($240-255) below spot too. The business is too strong to sell and too richly priced to chase; the synthesis is WAIT.

Key levels

Key resistance starts at ~$300 (round-number/psychological), then ~$320, up to the ~$340 all-time high that must be reclaimed to resume the advance. Key support sits at ~$260 (near the rising 200-day trend around $252 — the line that separates "healthy reset" from "breakdown"), then ~$235 (a deeper retracement and the valuation lens's fair-value zone), with a structural level near ~$195. The analyst consensus target spans a wide USD 110 (a stale laggard) to 385 (Stifel/UBS-type highs), averaging around USD 255 — modestly below the current price, because published averages haven't caught up to the run and reflect real disagreement over whether the growth justifies the multiple.

What legendary investors think

We ran Marvell past a panel of six legendary investors' frameworks.

Where they agree — and where they clash

Common ground

• The growth is real and accelerating: revenue +28% to a record $2.418B, data center 76% of sales, guidance raised twice (FY27 to ~$11.5B, FY28 to ~$16.5B), record operating cash flow.
• The custom-silicon franchise is genuine: a Marvell+Broadcom ~95% co-design duopoly with a >$75B lifetime win funnel — a real, structural AI tailwind.
• The valuation is stretched: ~66-70x forward (double its own 5-year average, ~80% above the semis median), after a ~5x run, with the analyst median target below spot.
• Positioning has cooled from a euphoric peak: an ~18% pullback from $340 on a CFO sale, design-loss chatter, and a broad chip wobble — a reset, not (yet) a thesis break.

The real debate

• Durable platform or second-source subcontract? Bulls see a co-design moat with switching costs; bears see the acknowledged #2 to Broadcom, with the Trainium3 loss to Alchip proving design wins can be lost as hyperscalers multi-source.
• Do the earnings justify the multiple? The 20x GAAP/non-GAAP gap and thin 14% GAAP operating margin let bears argue the "profitability" is adjusted; bulls point to record cash flow and raised guidance.
• Ride the winner or wait for the price? The macro and innovation lenses say a structural hyperscaler-capex wave is worth owning through volatility; the value, GARP, and forensic lenses say a ~70x multiple after a 5x run has no margin of safety.

The question it comes down to: Is Marvell a structural winner in the custom-AI-silicon duopoly whose hyperscaler-funded growth justifies riding the volatility — or the richly priced #2 whose second-source position, customer concentration, and a fresh design loss make ~70x forward a price to wait out rather than pay?

The numbers

Metric	Value
Price / Market cap	USD 278.20 / ~233B
P/E (TTM / fwd)	~95–105x / ~66–70x (vs semis median ~38x)
Revenue (Q1 FY27)	USD 2.418B (+28% YoY); data center 76% of sales
EPS (GAAP / non-GAAP)	$0.04 / $0.80 (a ~20x gap — SBC + acquisition amortization)
Gross / operating margin	~52% / ~14% GAAP (~59% / ~35% non-GAAP)
ROE	~16%
Balance sheet	Cash ~$3.84B, debt ~$5.28B (net debt ~$1.4B, D/E 0.29); OCF record $638.8M

Figures as of fiscal Q1 2027 (quarter ended May 2, 2026, reported May 27) / June 2026; sourced from Marvell IR/SEC filings, Motley Fool, Futurum, StockAnalysis, GuruFocus, Tom's Hardware. The wide GAAP-vs-non-GAAP gap (heavy stock-comp and Inphi/Cavium amortization) means the stock is priced almost entirely off adjusted metrics — a quality caveat. Custom silicon (~$1.5B) is guided to grow ~20% in FY27, notably slower than data center overall (~50%) or interconnect (~70%) — much of the acceleration is optical/interconnect, not custom compute alone. Fair-value estimates span ~$140 (forensic) to ~$450-550 (5-year innovation case); the valuation lens's $235-250 and the analyst median ($240-255) both sit below the price.

The bottom line

Marvell is the panel's clearest "great business, wait for the price" call, and unlike a commodity cyclical the bull case here is alive and specific. The company is executing: 28% revenue growth to a record, data center at 76% of the mix, two guidance raises, record cash flow, a genuine co-design duopoly with Broadcom, and a marquee Nvidia endorsement — enough for the innovation and macro lenses to want to own it through the volatility toward a multi-year, several-hundred-dollar outcome. But the price has run ahead of even that: ~66-70x forward earnings, double its own history, after a 5x move, with the analyst median target sitting below the current quote. The bears' concerns are concrete rather than rhetorical — a 20x GAAP-to-non-GAAP earnings gap, a 14% GAAP operating margin, a #2 position behind a faster-growing, higher-margin Broadcom, ~82% customer concentration in the top ten, a real next-gen design loss to Alchip, and a CEO selling into every leg up. Independent fair-value estimates cluster from ~$140 (forensic) through ~$235-250 (valuation), mostly at or below the price, with the bull's $450-550 a five-year story, not a today's-price justification. What would tip it bullish is a fresh major hyperscaler design win reaccelerating custom silicon and margins converging toward the non-GAAP story; what would break it is another design loss or a guidance stumble into a ~70x multiple. The business is too strong to sell and too richly priced to chase — which is why the panel waits, ideally for a digestion toward the $235-260 zone or the next quarter's clarity on hyperscaler allocation.