Optical Comms Just Got Repriced — Here's the Real Reason

The Entire Optical Communications Sector Was Mispriced. Its Stocks Are Rebounding for a Reason

After months of being overlooked, US-listed optical communications names have staged a real comeback — and it's not a bounce off oversold levels. Three narrative shifts are rewriting how the Street sizes this sector's TAM, margin profile, and its ties to the AI buildout.

Shift 1: From Module Supplier to AI Cluster Bottleneck

The old story was simple: 800G → 1.6T transceiver upgrades, tied to Nvidia GPU shipment cycles. The new story is Optical Circuit Switching (OCS) — the dynamic optical routing layer inside massive GPU superclusters.

$LITE is the poster child. OCS shipments doubled QoQ in FY26 Q4, with management guiding for $400M+ in OCS revenue in H2 2026. By early FY27, Lumentum is on track to become a core external OCS supplier for top-tier cloud providers. This isn't incremental — it opens a new TAM that decouples growth from the traditional GPU shipment cadence.

Shift 2: The Margin Story Flips

Gross margins aren't a supply-demand windfall anymore — they're structural. $LITE just broke past 47.43% gross margin. The old assumption was you needed $2B/quarter in revenue to hit those numbers; top players are now getting there on $1B.

Why: vertical integration (in-house InP lasers), a richer mix of high-value CW laser chips, and premium OCS systems scaling up. That's bringing fundamental money back into the space, not just momentum flows.

Shift 3: CPO Goes From "Someday" to Locked-In Roadmap

Co-Packaged Optics used to be treated as a distant, speculative concept — largely excluded from core valuation models. That changed when $NVDA's Spectrum-X CPO switches hit full mass production on August 14, 2026.

Add Nvidia's equity stake and long-term supply deals with Lumentum for external CW laser chips, and the CPO supply chain is now real. Scaled deployments land 2027–2028, setting up a second demand wave for laser chips — and resetting the sector's valuation floor with a genuine growth premium attached.

What the 13Fs Say - Q2 2026, filed for quarter ended June 30

Institutions were net buyers across the optical space last quarter, but positioning tells three different stories:

→ $POET — Broad, synchronized institutional conviction. Citadel added ~9.74M shares (+201.5%), Point72 added ~1.12M (+188%), Renaissance opened a fresh ~3.47M share position, Two Sigma added ~92K (+149%). Millennium was the lone seller, trimming ~70K shares. When this many funds move the same direction on the same name, that's consensus, not noise.

→ $COHR — A genuine tug-of-war. D.E. Shaw added 1.67M shares (+1,364.4%) and Two Sigma added 700%+ (+49K shares) — but Millennium cut 74% (-740K), Point72 cut 84% (-570K), Citadel trimmed 14% (-360K), and Bridgewater all but exited (-90%+, -150K shares). Sharp divergence like this usually means the street is genuinely split on the setup, not just rotating.

→ $LITE — Modest repositioning both ways. D.E. Shaw added ~770K (+57%) and Citadel added a smaller ~110K, while Two Sigma, Point72, and Millennium trimmed ~120K / ~54K / ~29K shares respectively — profit-taking after the run, not a thesis change.

Bottom line: the market spent a year pricing optical comms as a cyclical transceiver story. It's being repriced as a structural AI-infrastructure bottleneck asset — new TAM (OCS), durable margins, and a CPO roadmap that's no longer hypothetical. Watch how POET's consensus buying and COHR's split positioning resolve into Q3 prints.