One quarter of revenue is $2.739 billion. The five-year target is $80 billion at the midpoint. Those two numbers belong to the same company, and as of Tuesday’s close, the market added roughly $13 billion of market cap to celebrate the distance between them.
Marvell Technology held its 2026 Investor Day on October 6, and CEO Matt Murphy did not hedge. The company raised its fiscal year 2028 revenue target to $20 billion, up from Wall Street consensus of $18.2 billion and current FY27 forecasts of roughly $12 billion. Then came the longer range: revenues between $70 billion and $90 billion by fiscal year 2031. Shares gained 5.8% on October 6, prompting several analysts to raise their price targets. Jefferies moved first and hardest, lifting its target to $450 from $325 while maintaining a Buy rating.
What Has to Go Right
The FY28 number is the more defensible of the two. Data center revenue alone is expected to approach $18 billion in FY28. That is ambitious but not implausible given the current trajectory: Q2 net revenue hit a record $2.739 billion, up 37% year over year, driven by data center growth that accelerated to 46%. The quarter beat guidance by $39 million. The company has beaten for two straight quarters and raised both near- and long-term outlooks each time.
The FY31 number is a different conversation. The company aims to nearly octuple its fiscal 2026 revenue of $8.2 billion to a range between $70 billion and $90 billion. The $80 billion midpoint vastly exceeds Wall Street consensus of roughly $47 billion in revenue. Management’s breakdown of how they get there: approximately $37.5 billion of interconnect, about $10 billion from switching and storage, approximately $30 billion from custom products, and roughly $2.5 billion from communications and other businesses.
The custom silicon slice is where execution risk concentrates. Custom silicon currently generates approximately $1.5 billion annually. Management expects the business to grow more than 20% in FY27 and more than double in FY28, but getting from there to $30 billion by FY31 requires sustained hyperscaler commitment across multiple program generations. COO Chris Koopmans said the FY31 forecast is based on hundreds of products across top hyperscale customers and does not rely on a major new customer or a single home-run design win. That is a reassuring framing. It is also a statement that has to be true for nearly a decade to pay off.
The Connectivity Angle Most Are Missing
Crucially, Marvell’s near-term FY28 raise was not driven by custom chips. The incremental roughly $2 billion added to the FY28 outlook is being driven by continued strength in scale-out optics, scale-up optics, and switching, rather than an immediate acceleration in custom silicon. Marvell is winning on connectivity first, custom second. Jefferies analyst Blayne Curtis noted that connectivity sits at the center of the growth story, with interconnect revenue potentially reaching $37.5 billion at the midpoint of FY31. That is a $37.5 billion business that barely existed five years ago.
The Risks
Marvell’s valuation is the obvious risk: the stock is priced for unusually strong execution. Customer concentration adds a second layer: four customers represented 72% of gross accounts receivable as of August 1, 2026. A single hyperscaler pulling back its capex cycle or pivoting to an in-house silicon architecture could shave a meaningful portion of the FY31 model before management finishes a slide deck.
Broadcom sits ahead of Marvell in custom silicon scale today, and that gap does not close without flawless execution across multiple program ramps simultaneously. The FY31 targets also assume a total addressable market that management pegged at $400 billion by 2030. If AI infrastructure spend disappoints or consolidates around fewer suppliers, the denominator of Marvell’s share calculation shrinks.
What to Watch
The FY28 custom revenue figure is the most important near-term checkpoint. Marvell expects its custom business to exceed $12 billion in FY29, which means the ramp has to become visible in reported data within the next few quarters. If optics and switching carry the next two or three earnings beats while custom lags, the credibility of the FY31 model erodes regardless of the connectivity story.
Murphy put extraordinary numbers on the board Tuesday. The stock’s reaction tells you the market wants to believe them. Whether the math holds depends almost entirely on whether four hyperscalers keep ordering, keep ramping, and keep Marvell as a preferred partner through a five-year window when AI architectures will change several times over.
