## Tape / setup
Datacom and industrial photonics—chips, modules, and subsystems. When investors talk “AI infra,” the read-through here is **optical connectivity and power** (ZR/ZR+, CWDM/DWDM), not a generic software story.
## What matters
- **Datacom mix vs industrial / telecom**: growth and margin live in how much revenue is hyperscale-oriented vs legacy telco spend.
- **Co-packaged optics / high-speed pluggable narrative**: wins, design-ins, and whether pricing holds or gets competed down.
- **Execution on integration** (post-Coherent carve dynamics are mostly in the rear view, but cost and mix still drive the multiple).
## Bull case
Hyperscale and AI cluster build-outs keep high-speed optical demand on a long runway; LITE’s component/module position captures upgrades in **speed and power efficiency** with a cleaner story than a commodity module vendor if mix shifts right.
## Bear case
Customer concentration, pricing pressure on “obligatory” optics specs, and any capex pause at the big buyers hit revenue faster than the story suggests.
## Invalidation
Sustained guide cuts on datacom or margin compression that doesn’t recover in a quarter—especially if peers aren’t seeing the same—means the AI/optical bid isn’t sticking for this name.
## Trade framing (not advice)
If you’re vol-aware: the trade is often **event path** (guide, hyperscale datapoints) more than quarter-point estimate noise; structures should match whether you’re playing **catalyst** or **slow grind**.