They own the physics for 1.6T lasers, yet they just wrote off $36M to quit the DSP rat race.
Thesis: Lumentum controls the Indium Phosphide bottleneck essential for 200G/lane AI transceivers. They are effectively sold out for 18 months. The current margin compression is a temporary artifact of the mix shift from legacy telecom to AI datacom. By cutting the failed DSP division, they are concentrating all capital on their only true edge: the laser fab. This is a volume play on physical layer scarcity.
Verdict: LONG — Conviction: MEDIUM
Catalyst: Thailand facility ramping volume shipments in Calendar Q1 2025, validating the capacity expansion thesis.
Key Risk: Execution on the 40% capacity expansion. If they fail to bring tools online, they miss the window, and customers like Nvidia will force qualification of competitors.
The Tell: The DSP exit. 'We decided to stop our in-house development of certain communications ASICs... We believe we can meet customer needs using ASICs from third-party partners.' They realized they couldn't compete with the merchant silicon giants and capitulated to save cash for the fab.
Friction Level: HIGH_FRICTION — Bulls see a bottleneck monopoly with pricing power; Bears see a commoditized supplier unable to expand margins despite being sold out.
Report not found
The report data is no longer available. Please return to the archive.