On 6 October Marvell set a target of growing revenue from $8.2 billion to $70–90 billion in five years. The stock rose 5.8%. I own it, so I tried to break the plan instead of cheering it: four stress tests, what the share price already assumes, three scenarios with odds, and the checklist I'll use for the next Investor Day.
Disclosure. I own MRVL. It is about 9% of my portfolio, and I'm up about 26% on it. That is exactly why this page tries to argue against the plan, not for it.
Source: Interactive Brokers daily bars. Up 210% over the period, yet the 6 Oct close is still 9% below the June peak close ($316.43; intraday high $329.88).
| Target | Number | What it means |
|---|---|---|
| FY28 revenue | $20B | +67% on FY27's expected $12B. Was $18B in August. |
| FY28 data center | ~$18B | The FY29 data-center target, pulled in a year early. |
| FY29 custom silicon | $12B+ | Up from a $10B target. The deck says more than 3x FY28. |
| FY31 revenue | $70–90B | 55–60% a year for five years. |
| FY31 model | 44–46% | Non-GAAP operating margin; gross margin 56–59%, tax 15%. |
| FY31 EPS | >$30 | Non-GAAP, "implied" by the model above. |
| Market size | ~$400B | Marvell's addressable market in 2030, ~45% a year. |
The $16.5B and $18B figures come from coverage of the 27–28 Aug results. "Analyst average" is the Alpha Vantage consensus of 37 analysts on 7 Oct; it has not caught up with the new target yet.
Their record helps them. In FY21 Marvell set a 15–20% revenue growth target and delivered 23% a year. A 2024 data-center target of $15B for FY29 was raised to about $18B, and has now been pulled forward to FY28. This is a management team that has beaten its own long-term numbers. That earns the plan some credit. It does not earn it full credit.
The deck gives a 2030 market size and a five-year growth rate for each business. Put the FY31 revenue next to the market it sells into, and you get the market share the plan quietly requires.
FY31 revenue by business from the deck's midpoint ($37.5B interconnect, $30B custom, $10B switching and storage, $2.5B comms). Market sizes from the deck: ~$65B, ~$235B, ~$85B, ~$15B. Working back from the deck's growth rates gives FY26 revenue of about $8.4B, close to the reported $8.2B, so the split holds together.
The flag is interconnect. Marvell says it has been first to market at every generation of the chips inside optical links (DSPs, drivers, amplifiers). The plan needs it to take almost six of every ten dollars spent in that whole market in 2030, including newer near-package and co-packaged optics for scale-up that the deck mostly dates to 2027, against rivals such as Broadcom and Credo. Custom chips at ~13% of their market is the more believable leg.
FY28 is an optics year. If custom silicon is under $4B in FY28 (the deck puts FY29's $12B+ at more than 3x FY28) and comms is ~$2B, then about $14B has to come from interconnect plus switching. That is roughly three times my ~$4.6B estimate for FY26, in two years. Raymond James notes that the ~$2B Marvell added to its FY28 outlook comes from scale-out optics, scale-up optics and switching.
FY29 is a custom year. Custom silicon has to more than triple in a single year. Google's program is part of that ramp, and the CEO said meaningful revenue from it starts in FY29. A large part of the five-year plan rests on that ramp landing on time.
Bar heights scale 0–20%. Diluted share count rose from 870M to 921M over the same year. Non-GAAP EPS excludes all of this: Q2 FY27 was $0.94 non-GAAP and $0.33 GAAP.
Google holds a warrant on up to 58,970,907 Marvell shares at $206.58, expiring in August 2033. About 1.4M shares vest in quarterly steps over the first year. The rest vest in 240 equal slices, one for every $500M of custom products that Google and its affiliates buy from Q3 FY27 through FY33. The 8-K describes these purchases as discretionary and states no minimum purchase commitment. Buying all $120B would vest everything.
The dilution is small: about 1–1.5% by FY31 if Google's purchases add up to around $40B (my estimate). The economics are the point: at a $400 share price a fully vested warrant would be worth about $11B, roughly 9.5% of the $120B. In effect, a volume rebate paid in stock.
What neither the deck nor the 8-K says: whether the warrant is booked against revenue, and whether the FY31 ">$30" includes it. If stock pay settles near 5% of revenue, it is worth about $3.40 a share of that $30 (my estimate).
Instead of asking what Marvell is worth if the plan works, turn it around: what does the plan have to deliver for today's buyer to earn a decent return? I used 12% a year to early 2030, then valued the stock on FY31 earnings.
| Multiple in early 2030 | FY31 EPS needed | Revenue that implies | vs the $70–90B target |
|---|---|---|---|
| 25x | ~$17 | ~$45B | About two-thirds of the low end |
| 20x | ~$21 | ~$56B | About 80% of the low end |
| 15x | ~$28 | ~$75B | Just above the low end |
$287.01 grown at 12% a year for 3.3 years is about $417. Revenue converted at a 44% operating margin, 15% tax and about 1.0B shares. My estimates, not company figures.
Scale runs 0x to 60x. Broadcom: $375.81 ÷ $19.39 (year to Oct 2027). Marvell: $287.01 ÷ $6.76 (year to Jan 2028). On Marvell's own $20B FY28 outlook my EPS estimate is $6.90–7.70, which brings the multiple down to roughly 37–41x. Analyst averages from Alpha Vantage, 7 Oct.
Reading. At a 20x exit, the price already assumes Marvell gets most of the way to its plan. You are paying for the plan mostly working, and you get paid extra only if it fully works. That is a fair bet, not a cheap one.
My read. Weighting the cases 25 / 50 / 25 gives about $318, around 11% above the $287 close. The analyst average target on TipRanks after the event, $313.84, lands in almost the same place. Investor Day made the story stronger. It did not make the stock cheap.
What I'm doing. Holding. Not adding at this price; I'd add only on a pullback to around $240–250. My own trim rule: if it closes above about $345, I sell 20–25% of the position.
Scorecard. After the Q3 FY27 report I'll come back and score the near-term checks above. The scenarios get scored against FY28 results (early 2028) and FY31. The original text stays as written; the commit history is public.
Divide the revenue target by the market size the company itself gives. A share above about a third deserves a reason.
Interconnect needs ~58% of its 2030 market. Custom needs ~13%.
Five-year plans are rarely smooth. Find the year carrying the biggest step-up and ask what has to happen in it.
Custom more than triples in FY29, the year the CEO says meaningful Google revenue starts.
Stock-based pay and customer incentives are real costs even when the headline numbers leave them out.
Stock pay reached 11.9% of revenue, and at a $400 share price Google's fully vested warrant would be worth about 9.5% of the purchases it vests on.
Ask what the price needs, not what the plan could be worth. The gap between the two is your margin of safety.
At 20x in 2030, $287 needs ~$56B of FY31 revenue against a $70–90B target.
A stronger story and a higher price can cancel out. Holding and adding are different decisions.
The thesis improved, the stock rose with it, and the value I get from my scenarios is only ~11% above the price.
A management team that has beaten its targets earns a hearing, not a free pass on the next, much bigger one.
Marvell beat its FY21 target of 15–20% a year with 23%. The new plan asks for 55–60% a year.
Run it on any five-year plan before letting it change a position.
| # | Question | Marvell, 6 Oct 2026 |
|---|---|---|
| 01 | Is every business's implied share of its own market under about a third? | No, interconnect ~58% |
| 02 | Is the biggest step-up backed by a signed contract with committed volumes? | Signed, but purchases discretionary |
| 03 | Is stock-based pay under ~8% of revenue? | No, 11.9% |
| 04 | Are customer incentives (warrants, rebates) and their accounting disclosed in the plan? | Warrant yes, accounting no |
| 05 | Does the price need less than the low end of the plan? | Yes, ~80% of it at 20x |
| 06 | Has management met its previous long-term targets? | Yes |
| 07 | Did you write your add and trim prices before the event? | Investor check |
Row 07 is about you, not the company. Three red, two green, one partial: a real plan with stretched pieces, at a price that already assumes most of it.
The story got better. The price got there first.