Marvell and Synopsys both profit from the AI chip boom, but one company is quietly handing hyperscalers the tools to cut the other out entirely. The tension between them reveals something uncomfortable about where custom silicon money actually flows.
This post may contain links from our sponsors and affiliates, and Flywheel Publishing may receive compensation for actions taken through them.
Marvell Technology (NASDAQ:MRVL | MRVL Price Prediction) and Synopsys (NASDAQ:SNPS) reported earnings a day apart in late August. Both ride the AI infrastructure wave from opposite ends of the stack. Marvell sells custom silicon and optics to hyperscalers. Synopsys sells the EDA software and IP those same hyperscalers use to design chips in-house. That overlap is exactly why the article title frames Synopsys as a threat.
Custom Silicon Lifts Marvell. Ansys Lifts Synopsys.
Marvell posted record Q2 FY27 revenue of $2.739 billion, up 36.55% year over year, with Data Center growing 46% and reaching 79% of the mix. CEO Matt Murphy said “AI-related bookings remain exceptionally robust” and guided Q3 to $3.15 billion. The blockbuster item was an expanded Google custom silicon agreement covering inference accelerators, storage controllers, NICs, and near-memory compute, paired with a warrant allowing Google to acquire up to 7% of Marvell shares tied to revenue milestones.
Synopsys delivered revenue of $2.477 billion, up 42.37%, and non-GAAP EPS of $3.91. Design IP finally returned to roughly 11% growth, and Ansys is starting to pay off through the new Multiphysics Fusion product. Sassine Ghazi said Synopsys is “executing with focus, extending our leadership and gaining momentum” one year after the acquisition closed.
| Driver | Marvell | Synopsys |
| Main Engine | Data Center silicon, custom XPU | EDA software, IP, Ansys physics |
| Non-GAAP Op Margin | 36.6% | 41.6% |
| YoY Revenue Growth | 36.55% | 42.37% |
Where the Two Businesses Actually Collide
Marvell explicitly flags the risk that customers develop their own solutions or vertically integrate. Synopsys is the shovel seller enabling that path. Ghazi described a “Factory 2” model of licensing plus royalties, where hyperscalers get customized IP to build differentiated in-house silicon. He put it bluntly: “These chips will not happen without our interface IP.”
Marvell counters with portfolio breadth. Murphy argued “Point solutions at this juncture, we believe, are not going to get it done.” Marvell is stacking 800G and 1.6T optics, 51.2T scale-out switches, Celestial AI photonics, and XPU-attached silicon into an end-to-end offer that any single Synopsys IP block cannot replicate on its own.
What I Am Watching Through 2027
For Marvell, the tell is custom revenue in fiscal 2029, which Murphy indicated could push beyond the previously cited $10 billion plus figure once the Google programs ramp. Investor Day on October 6, 2026 should reset those numbers. For Synopsys, I want to see Multiphysics Fusion actually contribute to EDA growth in 2027 and Factory 2 deals close with named hyperscalers. Sentiment has already diverged sharply: MRVL is up 143.33% year to date, while SNPS is down 11.44%.
Why I Lean Toward Marvell for Now
I find Marvell’s setup more compelling despite the Synopsys overhang. The Google warrant is a rare signal of customer commitment. If a hyperscaler is willing to take equity exposure, that reads like a partner planning to stay. Synopsys still fits a different investor. If you want software-like margins near 41.6%, a $10.9 billion backlog, and a stock that has already reset, SNPS looks like a patient compounder. I would rethink my Marvell view if custom bookings soften at Investor Day or if hyperscalers accelerate their own designs faster than the roadmap suggests.
Contact [email protected] for any questions or corrections.
