Marvell (MRVL) at $1 Trillion: $78 Billion of Revenue at Today's Multiple, $145 Billion at Broadcom's

Marvell closed the earnings session at $241.42 with 876.93 million shares outstanding and a market capitalization of $211.7 billion, then traded to $224.34 after hours on a fiscal second quarter that beat on both lines and raised fiscal 2027 and fiscal 2028 revenue for the second consecutive quarter. The stock is up roughly 184% year to date against a 52-week range of $61.44 to $329.88. Reaching $1 trillion from here requires 4.72 times the current market capitalization, which is $1,140 a share on the current count and $1,069 a share if Google's warrant vests in full and the denominator becomes 935.9 million. Both of those numbers are arithmetic. Everything interesting about the question is in what has to happen to produce them.

Start with what produced the last move, because it is the wrong template. Marvell's market capitalization is up about 238% over the past twelve months. Revenue over the same twelve months is up 37%. The gap is re-rating. The multiple did roughly two and a half times the work and the business did the rest, which is the normal shape of a stock that goes from unloved to consensus long inside a year. The forward multiple is now 53 times the next twelve months and roughly 37 times the fiscal 2028 number management guided to on Thursday. That is the constraint on repeating the trick. Getting from $212 billion to $1 trillion the way Marvell got from $63 billion to $212 billion would require the forward multiple to reach the mid-200s. It will not. The next 4.72 times has to come from earnings, and that changes the question from a sentiment question into an arithmetic one with only one free variable.

The free variable is the terminal multiple, and the sensitivity is brutal. At the roughly 37 times fiscal 2028 earnings the market is paying today, $1 trillion requires about $27 billion of net income. At the non-GAAP net margin Marvell is converging on, call it 34.5% once operating margin sits at the high end of the 38% to 40% band, that is roughly $78 billion of revenue. At 28 times, a multiple appropriate to a business growing 25% rather than 45%, $1 trillion requires about $36 billion of net income and roughly $103 billion of revenue. At 20 times, which is exactly what Broadcom trades at today with a $1.71 trillion market capitalization and 64% expected revenue growth next fiscal year, $1 trillion requires $50 billion of net income and roughly $145 billion of revenue. The same milestone costs Marvell $78 billion or $145 billion of revenue depending entirely on what the market decides to pay for it when it gets there, and the entire history of semiconductor companies says the multiple compresses on the way up, not down. Broadcom is the proof standing in the room: the incumbent in this exact business, eight times Marvell's size, growing faster next year, at 20 times.

Compound it forward from the $18 billion management guided for fiscal 2028. Reaching $78 billion takes 4.3 years at a sustained 40% revenue CAGR and 3.6 years at 50%. Reaching $103 billion takes 5.2 years at 40%. Reaching $145 billion takes 6.2 years at 40% and 8 years at 30%. Equity markets price forward, so the market capitalization arrives roughly a year to eighteen months before the revenue does. Netting that out: the earliest credible date for a $1 trillion Marvell is fiscal 2031, and it requires a sustained 50% revenue CAGR for five straight years with the forward multiple holding above 35 times, which is to say it requires the company to grow like Nvidia and be valued like it while doing so. The central case is fiscal 2033 to fiscal 2034 at a 35% to 40% CAGR with the multiple settling near the high twenties. At 25% to 30% growth with convergence toward Broadcom's multiple, it does not happen this decade, because the revenue requirement grows faster than the revenue does.

There is an uncomfortable check on all three of those paths, and it is not a multiple argument. Management targets roughly 20% of the custom silicon market by 2028 and more than $10 billion of custom revenue in fiscal 2029. Take those two statements together and the implied total custom market is around $50 billion. The lowest bar to $1 trillion is $78 billion of Marvell revenue. Marvell could take one hundred percent of the custom AI silicon market as it currently sizes it and still be short. The trillion-dollar path therefore does not run through the custom market as management has defined it. It runs through the custom market being several times larger than management's own framing, or through Marvell capturing a much greater share of the interconnect and switching spend that surrounds it, or both.

Which puts the moat question at the center rather than the periphery. Marvell's defensible asset is SerDes, optical DSP, the 800G and 1.6T interconnect ramp guided above 70% growth this year, and the 51.2T switching franchise with scale-out switch revenue projected past $600 million. That attach revenue survives a hyperscaler bringing its compute die in house, which is why it deserves a premium multiple. The problem for the $1 trillion question is that the attach franchise is not large enough to be a $78 billion revenue base by itself. The business that is large enough is the compute socket, and the compute socket is the one Marvell has no structural moat in, competes with Broadcom for, and cannot defend against a customer's internal team. The moat that justifies today's multiple is too small to reach the destination, and the business large enough to reach the destination does not justify the multiple. That tension is the actual answer to the question, and no investor day resolves it.

The Google warrant is the company's own timeline written down. It runs to fiscal 2033, covers up to $120 billion of custom product revenue in 240 tranches of $500 million, and vests 58,970,907 shares at $206.58 along the way. That ceiling averages roughly $18.5 billion a year from one account, which is approximately the entire fiscal 2028 revenue base. It is also, usefully, self-financing dilution: full vesting lowers the share price required for a $1 trillion cap from $1,140 to $1,069, a 6.3% discount, in exchange for the revenue that gets there. Existing holders reach the milestone at 4.43 times rather than 4.72 times. The warrant makes $1 trillion arrive marginally sooner as a market capitalization and marginally later as a per-share outcome, and that fiscal 2033 expiry sits directly on top of the central case for arrival.

Against 38 covering analysts, no sells, an average target near $257 and a range running from $126 to $400, the most bullish published target on the street implies a $351 billion market capitalization. That is 35% of the way to $1 trillion, and it is the single most useful calibration available. Nobody who covers this stock professionally has a twelve-month number that gets a third of the distance. The bull case is fiscal 2031 on a 50% sustained CAGR and a defended multiple, catalyzed by the October 6 investor day breaking the socket count into compute versus attach and raising the fiscal 2029 custom figure well above $10 billion. The base case is fiscal 2033 to fiscal 2034 on 35% to 40% growth into a high-twenties multiple. The bear case is not a Marvell case at all, it is cohort derating: if the custom silicon complex re-rates toward the hardware multiples the sector historically carried, Marvell converges on Broadcom's 20 times, the revenue requirement moves to $145 billion, and the milestone recedes faster than the company can grow toward it.

The decision-relevant number is not the growth rate. It is the multiple Broadcom carries when Marvell reaches $40 billion of revenue. If the incumbent is still at 20 times at that point, the honest answer to how long Marvell takes to reach $1 trillion is longer than any current holder's time horizon.