Citi Maps an $11B Optical Switching Boom — The Real Bottleneck Sits Below Ciena

Citi Research analyst Ronald Josey now sizes the optical circuit switching (OCS) market at $11 billion by 2030, with Nvidia set to join Alphabet as a primary customer of suppliers including Lumentum and Coherent. The number was flagged in coverage of Citi's 2026 Global TMT Conference (Sept 8–10, 2026) and tracks an earlier July 19, 2026 Citi note that raised 2026 hyperscaler data-center capex to $730.9B — a path Citi now pencils in as $400.2B (2025) → $730.9B (2026) → $970.1B (2027) → $1.16T (2028).

That distinction matters because most optical coverage conflates OCS, optical transmission, transceivers, and lasers. Citi frames AI networking as entering \"Stage 4\" — the early adoption phase of near-packet optics (NPO) and co-packaged optics (CPO) replacing earlier bandwidth-expansion stages — and names Lumentum, Ciena, and Keysight as the headline beneficiaries. Cigna AI's parallel forecast (raised July 30, 2026) calls for 30%+ CAGR OCS revenue growth through 2030, anchored by Google's seed orders to Lumentum, Coherent, and Huber Suhner.

Alphabet spent an estimated $500M–$1B on OCS over the past five years through Project Apollo, its in-house 3D-torus network using MEMS-based optical switches. That small base is what Citi is now extrapolating forward. The new variable is Nvidia: with the Quantum-X Photonics and Spectrum-X silicon-photonics switch platforms, Nvidia is converting OCS from a Google-only internal architecture into a merchant market for hyperscalers broadly.

Nvidia's March 2, 2026 strategic moves put money behind that thesis. The company committed $2 billion each to Lumentum and Coherent — a total $4B into US-based laser and photonics manufacturing — alongside multiyear purchase commitments and capacity-access rights for advanced laser components, per the NVIDIA-Lumentum press release. The implicit message: the optical bottleneck is upstream, not at the system level.

Ciena is the most-traded name in optical networking, and the Citi note validated that — Q3 FY2026 revenue of $1.671B was up 37% YoY, lifting full-year FY2026 guidance to $6.42B at the midpoint, with Ciena's optical-transport share >25% globally. But Ciena's customers buy transmission systems; the more constrained supply chain — and the more leveraged profits — sit two tiers upstream.

Lumentum's Q4 FY2026 (quarter ended June 27, 2026) delivered net revenue of $1.006B, up 109% YoY, with non-GAAP gross margin of 50.4% and non-GAAP operating margin of 36.6%. Components — the laser, EML, CW, and InP lane products inside the data-center interconnect — drove $649.4M of the quarter, up sharply from a year earlier. Q1 FY2027 guidance of $1.225–$1.275B, per the Lumentum Q4 FY2026 release, implies another 20%+ sequential step.

Coherent is the broader photonics platform play: Q3 FY2026 revenue of $1.81B was up 21% YoY (27% pro forma), GAAP gross margin 37.7% (non-GAAP 39.6%), and the company just launched its PhotonLink integrated-optics platform bundling lasers, specialty fiber, silicon photonics, and end-to-end assembly for CPO/NPO. Management's PhotonLink launch lifted the 2030 integrated-optics TAM by $30B to $90B — an explicit bet that the value migrates from pluggable transceivers into the laser-plus-package stack.

The Chinese optical-module complex is the other leg. Zhongji Innolight, which supplies more than half of Nvidia's optical-module wallet and is the first to qualify 1.6T modules, posted H1 2026 revenue of ¥41.78B (+182% YoY) and net profit of ¥13.65B (+242%), with module gross margin of 46.6%, per the Quartr summary of Zhongji Innolight filings. Goldman Sachs estimates 1.6T shipments reach 25.5M units in 2026 and 45.7M units in 2027, and Nvidia's 1.6T procurement plan was raised to as high as 20M units for 2026. Eoptolink is the second China peer, with H1 2026 optical interconnect revenue of ¥20.88B and gross margin 48.46%.

Citi's Stage 4 framework — NPO and CPO replacing pluggable optics — is essentially a reordering of who captures the dollar. The investable trade is not \"optical networking as a category\"; it's the specific sub-tier inside that category where supply is constrained and pricing power lives. Reading the Citi note plus the most recent quarterly reports, four layers stand out.

The layer that justifies the $11B OCS forecast is the first row — lasers and InP wafers. Lumentum's own OFC 2026 briefing projected roughly 85% CAGR in InP optical lane volume for EML, CW, and ultra-high-power CPO lasers. That is the rate that converts an \"$11B by 2030\" headline into the EBITDA expansion visible in the Q4 FY2026 report.

The next two earnings cycles — Lumentum Q1 FY2027 (early November 2026) and Coherent Q4 FY2026 (already reported) — are the immediate catalysts. Lumentum has guided $1.225–$1.275B for Q1, which puts the company at a $5B+ revenue run-rate exiting fiscal 2027 versus the $3.014B FY2026 print. Coherent's Q3 FY2026 call already flagged accelerating AI datacenter demand with growing backlog visibility.

Three near-term signals to watch: (1) whether Nvidia's NVLink and Quantum-X roadmaps confirm OCS as a merchant-market component versus a vertically integrated Nvidia-only product; (2) FCC and BIS decisions on Chinese optical-module exports, which directly affect Zhongji Innolight and Eoptolink — Innolight reportedly generated 64.1% of H1 2026 revenue from the US, per the FCC targeting coverage; and (3) Ciena's Q4 FY2026 update on optical-transmission share gains.

If Citi's $11B OCS forecast holds, the 2030 optical stack looks meaningfully different from today. NPO and CPO will have moved from \"early adoption\" (Citi's Stage 4 in 2026) into mainstream shipment by 2028, which means pluggable transceivers lose intra-rack share while laser and InP wafer suppliers capture more dollars per rack. Coherent's $90B integrated-optics TAM and Lumentum's 85% InP-lane CAGR projection are the two most aggressive long-dated sell-side numbers in the supply chain.

Three structural risks frame the thesis. First, supply chain concentration — Lumentum, Coherent, and Furukawa Electric are the only credible merchant EML and ultra-high-power laser sources, so a single yield disruption moves the entire sector. Second, Chinese-module export controls could bifurcate the global market between merchant Zhongji Innolight supply and US-domestic Coherent/Lumentum capacity. Third, the Stage 5 transition — whatever follows NPO/CPO — could compress the optical premium the way 100G Lambda once did.