LITE research

Position in the Physical Layer of AI thesis

LITE is the Photonics pivot. The Physical Layer thesis says datacenter optical interconnects are a hard constraint on AI training cluster scaling. Lumentum makes the EML (electro-absorption modulated laser) chips that go inside every 800G+ datacenter optical transceiver. Lumentum is the only public-market merchant supplier of 200G-per-lane EMLs at production volume.

Quinn's book opened LITE on 2026-05-05. The 2026-05-05 snapshot had LITE at +0.40% (same-day entry). Sized at 1.9% of book. The position is small but the thesis weight is high.

Recent catalysts (last 30 to 60 days)

NVIDIA $2B deal disclosed. Nvidia confirmed Lumentum as a strategic optical-component supplier with a $2B multi-year commitment in Q1. The disclosure included specific 200G/lane EML SKUs.
800G transceiver design wins. Lumentum's EMLs are designed into 800G transceivers shipping from at least 3 of the 4 major hyperscalers' next-gen builds.
Coherent Inc. acquisition complete. The Coherent merger (closed 2025) integrated complementary laser portfolios. Synergies tracking.
Q3 FY26 print solid. Revenue beat, guidance raised. Datacenter mix continued to grow as legacy telecom mix declined.

The thesis (what has to be true)

1. 800G stays the dominant datacenter optical generation through 2027. If hyperscalers skip 800G and jump to 1.6T faster than expected, Lumentum's 200G/lane EML window compresses. 2. Lumentum stays the lead merchant EML supplier. Coherent (now part of Lumentum) was the second supplier; the merger removed one competitor and consolidated share. 3. 1.6T optics transitions to InP-based PICs on schedule (2027-2028). Lumentum has an InP roadmap, but the next-gen competition includes ALMU and AXTI on the substrate side. 4. Datacenter mix keeps growing. Telecom mix is the legacy drag.

Kill vectors (what would break the thesis)

1.6T accelerated adoption. Hyperscalers skipping the back half of the 800G cycle and pulling forward 1.6T deployments would compress LITE's EML window.
Coherent integration friction. Operational integration of the acquired business is the execution risk.
Foundry capacity squeeze. LITE outsources some EML production; TSMC or other foundry reallocations could squeeze.
Customer concentration. Nvidia is now the largest customer. Any pivot from Nvidia is the biggest single risk.

Layer context

Photonics layer. Sister names: ALMU (anchor, InP PIC builder), SIVEF (MEMS timing oscillators), AXTI (compound semi substrates), LPTH (small Photonics). LITE is the highest-revenue-quality Photonics name; ALMU is the highest-asym smaller-cap; SIVEF is the niche.

Position discipline

Pivot sleeve rules.
Trim trigger: +100 percent from cost.
4-quarter thesis-fail rule applies. If Nvidia pivots or 1.6T pulls forward, the clock starts.

Moat 8. Asym 7. The moat is high because EML manufacturing at scale is patented + capital-intensive. The asym is moderate because Lumentum is already large; the upside is share + mix expansion, not category creation.

Real money. Real position. Real receipts.