$CIEN beat, raised, and still fell 11%.
Revenue: +37% YoY
Adjusted EPS: +215%
FY26 revenue guide: raised to $6.42B
So the interesting part isn’t whether AI optical demand is real.
It’s what happens next.
Q3 non-GAAP gross margin: 46.4%
Q4 guide midpoint: 45.0%
Q3 non-GAAP operating margin: 22.5%
Q4 guide midpoint: 20.0%
Revenue is still expected to grow.
Margins are expected to step down.
And two customers represented 41.7% of revenue.
Demand ≠ margin durability ≠ valuation.
Evidence before prediction.
#AIInfrastructure #OpticalNetworking #Semiconductors
The part I’d pressure-test is whether the bottleneck migrates faster than capex can monetize it.
“100% booked” and “supply constrained” still don’t tell us who captures the rent.
The proof chain I’d watch is:
booked capacity → qualified output → sustained utilization → pricing → free cash flow.
If new supply comes online under fixed LTAs while power, networking and capex costs rise, revenue can stay strong even as marginal ROIC compresses.
So the next gate isn’t more bookings. It’s whether the next wave of capacity converts into durable margins and cash flow.
The piece I’d add is that chokepoint power should probably be measured in qualification time, not market share.
A 90% share matters much less if a second source can be qualified in six months.
The real leverage is:
inventory runway vs. substitution lead time.
That’s where I’d look for the actual asymmetry.