- Marvell's Oct. 6 investor day is the event investors expect to deliver another lift to long-term AI guidance, with RBC Capital Markets forecasting at least a $2 billion increase to the FY2029 AI revenue outlook from the current $10 billion.
- Management has already reset expectations twice: August guidance lifted FY27 revenue to $12 billion from $11.5 billion and FY28 to $18 billion from $16.5 billion, up from earlier targets of $11 billion and $15 billion.
- The company is pushing into the second wave of AI spend, showcasing what it calls the industry's first 2nm optical interconnect demonstrations at ECOC 2026 and expanding its AT&S partnership to lock in advanced IC substrate capacity.
- Marvell targets raising data center market share from 13% to 20% by 2028, with data center customers already at 74% of revenue — a concentration risk that puts sustained data center revenue growth and stable margins through fiscal 2027 at the center of the proof.
- Custom silicon is forecast to more than double by FY2028 from $1.5 billion annually, and analysts want clearer operating-margin goals and larger data-center market size assumptions alongside the revenue upgrades.
- The stock has more than tripled in 2026 but sits about 16% below its June 18 high at roughly 86 times trailing and 62 times forward earnings, leaving little cushion for a merely positive presentation — a modest earnings beat previously produced an 11.62% decline.
- Micron's recent results reinforced that the AI infrastructure buildout is still alive, but Marvell's test is whether custom silicon and the network fabric binding data centers together keep deepening.
Quotes
“the industry's first 2nm optical interconnect demonstrations”
“sustained data center revenue growth and stable or improving margins in quarterly results through fiscal 2027”
“RBC specifically expects a raise of at least $2 billion in the FY2029 AI revenue outlook .”
“a modest earnings beat previously resulted in an 11.62% stock decline”
“the company to use its Oct. 6 investor day to lift long-term AI guidance again”