Nomura sees a sustained 1.6T optical-transceiver cycle and rising AI-networking opportunity, constrained by high-end component shortages

Overview

Nomura's CIOE 2026 conference takeaways examine the AI data-center optical-network supply chain. The report argues that 1.6T transceivers remain the near-term growth pillar, while shortages in advanced components and the early development of NPO, CPO and OCS create both execution risks and opportunities for technology-leading suppliers.

Core views

Nomura reports that 1.6T transceivers have become the mainstream product for scale-out networks, with multiple vendors displaying products at CIOE 2026. Improved automation could raise assembly efficiency and support faster capacity expansion at transceiver manufacturers. However, component availability remains the central bottleneck. The institution believes incumbent 1.6T suppliers, particularly InnoLight, can retain leading positions through effective supply-chain management. The report expects the 1.6T cycle to persist because 2.4T and 3.2T transceivers still face substantial challenges in the 400G-per-lane era. Rather than relying only on the traditional silicon-photonics path, next-generation products may require new materials such as thin-film lithium niobate (TFLN) or new modulation technologies such as Coherent Lite. Nomura notes progress by suppliers including Murata in TFLN modulators, but presents these technologies as solutions still addressing the next-stage technical hurdles. NPO was a major focus at the exhibition. Suppliers displayed differing solutions, including Hisilicon's VCSEL-based and internally sourced-light CPO offerings, CIG's silicon-photonics CPO with an external light source, and Accelink's 6.4T NPO solution for scale-up networking. Nomura says there is no industry-wide NPO standard yet. It expects early movers that can provide customized solutions and win orders from large global AI customers to achieve higher value content and margins during initial commercialization. The report identifies shortages across the supply chain. While 70mW CW lasers and 100G EML/PD chips are already in mass production and are supporting robust 800G shipments, 100mW-plus CW lasers and 200G EML/PD producers still need higher yields to close the demand-supply gap for 1.6T transceivers and NPO products. Nomura also identifies 3nm DSPs, mSAP-based advanced PCBs, multicore fiber and high-precision ferrules as potential new chokepoints. These constraints could reduce high-end, including 1.6T, shipments relative to expectations, but may also create opportunities for domestic suppliers to gain global share. Nomura argues that Chinese optical-communications companies already play important roles in the global AIDC supply chain, especially in transceivers and components, while upstream materials, lasers and equipment offer further room for domestic substitution. It cites progress in InP and SiGe substrates, high-end EML/CW-laser/PD chips, multicore and hollow-core fiber, and transceiver automation, coupling and testing equipment. The report favors Zhongji InnoLight and Accelink in high-end optical transceivers and NPO, TFC in NPO/CPO components, and YOFC and Shenzhen T&S in high-end AIDC fiber and fiber connectors. OCS is another emerging theme. Nomura notes that its potential use may extend beyond Google's scale-out network to several network segments. Cigna AI projects OCS revenue will exceed USD2bn in 2026 and USD8bn by 2030E, with scale-up OCS associated with GPUs potentially surpassing scale-out OCS applications by 2030E. Yet architectures remain fragmented across liquid-crystal, MEMS and silicon-photonics approaches, and Nomura believes broader commercialization requires more work across the supply chain. NVIDIA's recently discussed scale-in network is also presented as a potential future demand driver for AI networking as enterprise AI factories become more widely adopted.

Analysis framework

Nomura bases its conclusions on CIOE 2026 forum attendance, meetings with dozens of companies across the optical-communications supply chain, and observations of exhibited products. It traces demand from AI data-center network architectures into transceiver technology roadmaps, component supply constraints, domestic substitution opportunities and emerging NPO, CPO and OCS applications.

Methodology notes

• Industry AnalysisSupply-demand framework

  Demand-supply assessment of AIDC optical-network products and critical components

  The report links growing demand for 1.6T transceivers and NPO products to shortages and yield limitations in lasers, optical chips, DSPs, PCBs and fiber-related components.

• Competition & strategyValue chain analysis

  Optical-communications supply-chain analysis

  Nomura assesses transceivers, components, chips, materials, fiber and equipment to identify where supply bottlenecks and domestic-substitution opportunities sit.

• Industry AnalysisUpstream-Midstream-Downstream Transmission

  Technology and network-architecture transmission

  The report explains how AI network evolution from scale-out to scale-up, scale-across and scale-in affects component requirements, supplier barriers and commercialization prospects.

Asset mapping & comparison

Structured mapping from thesis to named assets (strengths, weaknesses, peers, risks).

• Zhongji InnoLight (300308 CH / 3308 HK)

  Preferred high-end optical-transceiver and NPO solution provider

  Strengths

  Nomura cites its leading position in high-end transceivers and effective supply-chain management.

  Comparison

  Identified alongside Accelink as a preferred high-end optical-network supplier.

  Risks

  Exposure to high-end component shortages and potential lower-than-expected 1.6T shipments.

• Accelink (002281 CH)

  Preferred NPO and high-end optical-network supplier

  Strengths

  Displayed a 6.4T NPO solution for scale-up networks.

  Weaknesses

  NPO standards and architecture remain fragmented.

  Comparison

  Competes across NPO/CPO solution approaches shown by Hisilicon and CIG.

  Risks

  Commercialization depends on securing large global AI-customer orders.

• TFC (300394 CH)

  Preferred NPO/CPO component maker

  Strengths

  Nomura identifies TFC as a leading Chinese player in optical components.

  Comparison

  Positioned in the component layer of the NPO/CPO supply chain.

  Risks

  High-end component bottlenecks and uncertain NPO standardization.

• YOFC (6869 HK)

  Preferred high-end AIDC fiber and fiber-connector supplier

  Strengths

  The report cites progress in multicore and hollow-core fiber.

  Comparison

  Mentioned alongside Hengtong and China ZTT in advanced fiber development.

  Risks

  Multicore-fiber and precision-ferrule constraints may affect supply-chain execution.

• Shenzhen T&S (300570 CH)

  Preferred high-end AIDC fiber and fiber-connector supplier

  Strengths

  Included in Nomura's preferred suppliers for high-end AIDC fiber and connectors.

  Risks

  Exposure to supply-chain bottlenecks in advanced fiber-related components.

Key data

• CIOE 2026 dates9-10 September 2026Nomura attended the Shenzhen exhibition, technology forums and company meetings.
• Mainstream transceiver product1.6TNomura identifies 1.6T as mainstream in scale-out networks.
• Next-generation transceiver speeds2.4T / 3.2TProducts still face technology challenges in the 400G-per-lane era.
• Existing mass-production optical components70mW CW laser and 100G EML / PD chipsThe report says these support robust 800G transceiver shipment growth.
• OCS market revenue forecastAbove USD2bn in 2026; above USD8bn by 2030ECigna AI presentation cited by Nomura.
• Potential scale-up OCS timingBy 2030ECigna AI expects scale-up OCS tied to GPUs to exceed scale-out OCS applications.

Impact & implications

Nomura believes the AIDC optical-network upcycle favors companies that remain ahead of technology changes and manage constrained supply chains effectively. It sees a longer 1.6T opportunity window, incremental domestic-substitution potential upstream, and longer-term optionality from NPO, CPO, OCS and scale-in networks, but emphasizes that technical yield challenges, supply constraints and fragmented architectures can delay adoption.

Risks

• Yield improvements in 100mW-plus CW lasers and 200G EML/PD chips may lag, leaving a demand-supply gap for 1.6T and NPO products.
• Shortages in 3nm DSPs, advanced PCBs, multicore fiber and high-precision ferrules could lead to lower-than-expected high-end transceiver shipments.
• NPO and OCS architectures remain fragmented, which may delay standardization and commercialization.

What to watch

• Progress in domestic substitution for optical materials, lasers and equipment.
• Penetration and commercialization of OCS and CPO.
• Development of scale-in network architectures and enterprise AI-factory adoption.
• Yield progress for high-power CW lasers and 200G EML/PD chips.