Jun. 2 at 10:37 PM
$LWLG $SIVEF $MRVL $NVDA $GFS
The architecture, timing, and supply-chain logic are lining up.
Jensen reportedly framed Marvell as the next trillion-dollar AI optics/connectivity platform. That matters because Marvell now owns Polariton — the plasmonic modulator architecture that publicly demonstrated high-speed performance using LWLG’s Perkinamine material.
This is the key: Marvell/Polariton may own the modulator architecture, but Perkinamine may be the complex EO material layer that enables the speed, power, footprint, and scalability that made the architecture valuable.
If MRVL is preparing a serious Polariton ramp, the Perkinamine material/IP layer would need to be locked down well ahead of production — qualified, licensed, reliability-tested, integrated into the foundry/process flow, and supply-chain planned before any 2027 high-volume ramp.
That is why Yves’ earnings-call language matters. LWLG disclosed it is negotiating “a new material supply and licensing agreement with one of our lead customers” to support “high-volume production” anticipated in 2027.
Marvell was not named. But the pattern is hard to ignore: long-time LWLG partner Polariton demonstrates Perkinamine-enabled performance, Marvell acquires Polariton, Jensen elevates Marvell as a major AI optics platform, and LWLG is now negotiating a material supply/licensing agreement ahead of a 2027 production ramp.
The Sivers/GF announcement adds another layer. The optical stack is forming around lasers, foundries, modulators, packaging, and critical material IP. Sivers brings the light. GF brings the manufacturing ecosystem. Polariton brings the modulator. Marvell brings scale. LWLG may bring the performance material inside the bottleneck.
And Perkinamine is not exclusive to MRVL. If multiple Tier-1s are qualifying LWLG’s material through foundry-accessible flows, this becomes bigger than one customer. It becomes a platform-material story.
Once a critical EO material layer is qualified and embedded into optical design flows, adoption can escalate quickly. LWLG does not need to win the entire optical engine. It can monetize the material/IP layer across multiple modulator platforms, customers, and AI interconnect generations.
MRVL may be the most visible validation path. But if Perkinamine becomes reusable performance IP across the AI optical fabric, the TAM expands far beyond one customer.
Not confirmed — but the alignment is getting difficult to dismiss.
https://www.stocktwits.com/Stocktwits/message/655089503
