Marvell's 5% Bounce Hides a $270 Target Below Where It Already Traded

Generated by AI agentVivian QiReviewed byThe Newsroom
2min read

- Piper SandlerPIPR-- initiates MarvellMRVL-- coverage with an Overweight rating and $270 price target, 13% above current levels but below Wall Street's $301 average.

- Marvell's custom AI chips for AmazonAMZN--, MicrosoftMSFT--, and GoogleGOOGL-- drive record $2.74B Q2 revenue, with Google's $12.2B warrant unlocking equity as purchases grow.

- Valuation risks emerge as Marvell trades at 22x sales (vs. 17x for Nvidia) and 79x earnings, pricing in $45B 2030 revenue projections despite recent 40% stock decline.

- Analysts urge patience until October's strategy day to validate growth potential before committing to high multiples, emphasizing momentum over speculative bets.

Marvell jumped about 5% Thursday after Piper Sandler kicked off coverage with an Overweight rating and a $270 target, pushing the stock to roughly $239. But the number worth your attention isn't the 5%. It's the $270 — a target barely 13% above the current price, below Wall Street's average of about $301, and far below the near-$330 MarvellMRVL-- fetched earlier this year.

That gap is the whole story in one view.

The business is genuinely improving

Marvell isn't an off-the-shelf chip company. It designs custom AI processors for the three largest U.S. cloud operators — Amazon's Trainium, Microsoft's Maia, and, under an agreement disclosed in August, Google's TPUs — the bespoke alternative to Nvidia's standard-socket GPUs. That makes it the only custom-chip designer working with all three hyperscalers.

Quick Backtesting Tool

Symbol
Strategy
Backtest Range

Fresh data backs the story. In the quarter reported Aug. 27, revenue hit a record $2.74 billion, up 37% year over year, with data-center revenue climbing 46% and driving most of that growth. Management raised its annual outlook for the second straight quarter and guided third-quarter sales to $3.15 billion.

The Google deal is the centerpiece of the bull case. Google received a warrant to buy up to $12.2 billion of Marvell stock at $206.58 a share, one that vests only as Google actually buys chips — tranches unlock as custom-product revenue accrues. If fully exercised, the structure implies roughly $120 billion of cumulative purchases by fiscal 2033. It is equity structured as a sales incentive: buy more, own more.

None of this is cheap.

The multiple is the entire risk

Through a sector-relative lens, Marvell's growth, momentum, and balance sheet all confirm. Revenue growth is top-quartile and accelerating, guidance has gone up two quarters in a row, momentum still sits above its 50- and 200-day averages after a violent summer, and the balance sheet is nearly clean — net debt of about $1 billion against cash near $4 billion, roughly what an asset-light design house should carry.

Valuation is the one weak link, and it's heavy. Marvell trades near 22x trailing sales — the richest sales multiple in the group, ahead of Nvidia's ~17x, Broadcom's ~20x, and AMD's ~20x — and at roughly 79x trailing earnings, about double Broadcom and nearly triple Nvidia. The market isn't paying for what Marvell earns today; it's paying for what a custom-chip ramp will produce years out. Piper's own framing makes that explicit, valuing the stock on about 18x a 2030 earnings figure and projecting EPS near $19 by then on revenue of roughly $45 billion.

That is why the stock was so fragile this summer. It topped near $330 in June, then fell roughly 40% in about a month as the semiconductor complex de-rated — including a drop of about 39% between mid-June and mid-July on a mix of Fidelity trimming its stake and dilution worries over the Google warrant. The business kept improving the whole time. The price had simply run ahead of the facts, and high-multiple growth names get repriced on any wobble even when the fundamentals are fine.

What to do with a price that's already paid for the ramp

The five-factor read here is lopsided in one specific way: the confirmers — growth, safety, momentum — all point up, while valuation carries the entire risk. In process terms, that makes this a momentum-and-growth sleeve holding, not a hold-forever core. It belongs in a barbell, paired with durable cash-flow or dividend names, and sized so that a repricing doesn't own the outcome.

The honest reading of Thursday's "catalyst" is that it's modest. Piper's $270 is a validation of the story at roughly today's price, not a new high-water mark. The real test is Oct. 6, when Marvell hosts an analyst day expected to lay out its long-term custom-silicon model — the same day that will tell investors whether the numbers can keep improving fast enough to grow into a multiple that is already paid for.

Chasing the 5% bounce and waiting to see the ramp are two different decisions. The factor stack says the business deserves watching. The valuation says let the analyst day prove the trajectory before you pay for it twice.