Who's Really The King of Custom Silicon, Broadcom or Marvell Technology?
Broadcom (AVGO) and Marvell (MRVL) reported earnings tied to custom AI silicon. Broadcom said quarterly revenue was $22.2B, including $10.8B from AI chips, and guided Q3 AI revenue of $16B. Marvell reported $2.418B revenue, with data center at 76%, and raised FY2027 to about $11.5B. The article compares their AI networking and custom ASIC roadmaps and valuations.
How this was made
The 30-second read
Why it matters
Both names are presented as having delivered earnings that clarify competitive positioning, with AVGO leaning on scale and networking mix and MRVL leaning on narrower but faster-growing custom engagement and raised multi-year targets.
Market read
Traders can use the reported AI revenue mix and raised guidance to update near-term expectations for AI-silicon demand and execution risk into Q4.
What to watch
Valuation gap (AVGO ~21x vs MRVL ~55x forward) implies MRVL is more sensitive to any AI capex timing wobble or photonics/XPU schedule delays.
Background
The article frames Broadcom versus Marvell as the two leading custom AI silicon suppliers, emphasizing AI-chip revenue, networking attach, and hyperscaler co-design.
Ticker impact
Broadcom reports $22.2B quarterly revenue with $10.8B from AI chips and guides Q3 AI revenue to $16B+.
Near-term bias to hold strength while investors track whether AI networking stays near the ~40% mix.
The article provides concrete earnings and guidance figures, but it is still framed as comparative analysis rather than a fresh, independently verified catalyst beyond the reported print.
Marvell reports $2.418B revenue with data center at 76% of mix and raises FY2027 revenue to about $11.5B.
Potential upside follow-through if execution on Celestial AI photonics and the flagship XPU program avoids slippage.
The text includes specific raised guidance and execution checkpoints, but it does not provide new deal terms or incremental disclosures beyond the earnings framing.
Market effects
Reinforces investor focus on custom AI silicon co-design, AI networking attach, and hyperscaler lock-in versus dual-sourcing.
Limited direct regional read-through; primarily US-listed semis sentiment.
Highlights global hyperscaler demand signals (Google, Meta, OpenAI, AWS, Microsoft) that can influence broader AI infrastructure capex expectations.
Counterpoint
The comparison may overstate “pecking order” by extrapolating guidance and customer engagement claims without addressing execution risk or competitive pricing pressure.
Key entities
- companyBroadcom
Reports $22.2B quarterly revenue, $10.8B from AI chips, and Q3 AI revenue guidance of $16B+.
- companyMarvell Technology
Reports $2.418B revenue with 76% data center mix and raises FY2027 revenue to about $11.5B.
