The Accelerating Cycle of Speed Upgrades
The iteration rhythm of the optical transceiver industry is undergoing a fundamental shift. According to data from the China Academy of Information and Communications Technology (CAICT), driven by the rapid development of AI, the optical transceiver iteration cycle has shortened from the earlier 3–4 years to approximately 2 years today. This accelerating trend means that even as the industry is still digesting the transition from 800G to 1.6T, the footsteps of 3.2T are already clearly audible.
The 2026 OFC exhibition sent a clear signal: multiple companies confirmed that 1.6T products have entered the mass production phase, while Broadcom‘s release of the industry’s first 400G/lane DSP chip has laid the technical foundation for 3.2T optical module applications. The technology roadmap from 1.6T to 3.2T is unfolding at a pace that exceeds industry expectations. For data center planners and optical transceiver purchasers, understanding the direction of technology evolution over the next five years is not only a matter of forward-looking planning but also essential for mitigating the risks associated with technology roadmap decisions.
1.6T: Opportunities and Challenges in the Year of Mass Production
2026 is widely regarded within the industry as the “year of mass production” for 1.6T optical transceivers. Yet the reality behind the term “mass production” is far more nuanced than the words suggest. On the demand side, annual demand for 1.6T optical transceivers in 2026 has already surpassed 25 million units, leading the industry in growth rate. The two giants, NVIDIA (approximately 15 million units) and Google (approximately 10 million units), account for the vast majority of demand, directly shaping the trajectory of industry prosperity. As the world‘s largest purchaser of optical chips, Google’s demand is projected to surge from 200 million chips in 2026 to 600–800 million chips by 2028, signaling extremely robust long-term demand.
The supply side, however, presents a considerably more sobering picture. Constrained by shortages of upstream EML laser chips, isolators, and other core components, 1.6T optical transceiver shipments in 2026 are projected at only about 15 million units, leaving a supply-demand gap of approximately 10 million units. This tightness is expected to persist into 2027, when 1.6T demand is projected to exceed 40 million units while shipments are still anticipated to fall short of 30 million units. The widening gap between supply and demand means that 1.6T optical transceivers are currently, in practical terms, in a state of “structural undersupply.”
For optical transceiver purchasers, this translates to sustained premium pricing and extended lead times for 1.6T products in the near term. Until 1.6T supply constraints are comprehensively alleviated, 800G will continue to play a predominant role in many AI cluster deployments. On the procurement strategy front, a dual-track approach of “800G for baseline coverage plus 1.6T for incremental capacity” is advisable, complemented by long-term agreements to lock in capacity allocation from key suppliers in order to navigate supply uncertainty over the next one to two years.
3.2T: Technology Pre-Research and Standardization Progress
Even as 1.6T is still ramping toward mass production, technology pre-research and standardization efforts for 3.2T are accelerating. At OFC 2026, Broadcom‘s 400G/lane DSP chip was recognized as a critical milestone for the commercialization of 3.2T optical transceivers. The industry broadly expects 3.2T products to enter the commercial validation phase between 2027 and 2028.
There are two primary technical pathways for implementing 3.2T. The first continues the traditional pluggable approach, building 3.2T transceivers based on an 8×400G lane architecture — an approach that offers the advantage of good compatibility with existing data center switch interfaces and cabling schemes. The second pathway involves deeper integration of 3.2T with CPO technology, leveraging the bandwidth density advantages of co-packaged optics to achieve more compact system integration. Based on current technical discussions, these two pathways are not mutually exclusive — pluggable 3.2T is better suited for intra-data center scale-out networks, while CPO-based 3.2T will have greater advantages in GPU-to-GPU interconnect scale-up networks.
From a standardization perspective, the Optical Internetworking Forum (OIF) has initiated work on CEI-448Gbps signaling standards, defining technical specifications for electrical interfaces above 400Gbps per lane — a prerequisite for multi-vendor interoperability of 3.2T optical transceivers. Meanwhile, the Ethernet Alliance is updating its 2026 Ethernet Roadmap, paving the way for higher-bandwidth Ethernet standards. For optical transceiver purchasers, monitoring the progress of these standardization bodies helps gauge the interoperability and supply chain ecosystem maturity of 3.2T products.
Coherent Optics: The Growth Engine of the DCI Market
Beyond intra-data center pluggable optical transceivers, coherent optical transceivers are emerging as another high-growth segment worthy of close attention. TrendForce notes that as the 1.6T generation enters mass production and demand for edge computing and data center interconnect (DCI) takes shape, the markets for 800G and 1.6T ZR/ZR+ coherent optical modules will expand concurrently.
The core value proposition of coherent optical transceivers lies in their ability to achieve high-bandwidth data transmission over long distances (typically 80 km to 120 km or more), whereas traditional IM-DD (intensity modulation-direct detection) approaches are generally limited to distances under 10 km. This characteristic makes coherent transceivers the ideal choice for interconnecting data centers. As AI clusters evolve from single-data-center deployments to cross-data-center distributed deployments, DCI requirements are shifting from “optional” to “mandatory.” LightCounting has raised its sales forecast for 800G ZR/ZR+ optical transceivers for 2026–2027, primarily because major customers are accelerating the shift from on-board solutions to pluggable DWDM modules and using these modules directly on switches and routers.
From a market structure perspective, product forms in the DCI space are evolving from traditional dedicated transmission equipment toward pluggable coherent modules. 800G ZR/ZR+ modules can be inserted directly into standard switch ports without requiring additional transmission equipment, greatly simplifying the deployment and maintenance complexity of DCI networks. For data center operators, this means that cross-data-center optical interconnects can be deployed as simply as ordinary optical transceivers, lowering the barriers to DCI network construction and operation. For optical transceiver manufacturers, coherent modules typically command higher gross margins than intra-data center modules of equivalent speeds, making this a high-value segment worth prioritizing.
Hollow Core Fiber: A Potential Disruption in Transmission Media
Even as optical transceiver technology evolves rapidly, the transmission medium itself is also undergoing transformation. Hollow Core Fiber (HCF), an emerging fiber technology, is offering data center providers a new solution to address both capacity and latency requirements.
Unlike traditional solid-core fiber where optical signals propagate through glass, hollow-core fiber guides light through an air- or vacuum-filled central core surrounded by a microstructured cladding. This design yields two key advantages: first, the speed of light in air is approximately 30% faster than in glass, translating into significantly reduced end-to-end latency — a critical factor for scenarios such as financial trading and real-time inference that require distributed AI training across data centers; second, hollow-core fiber exhibits extremely low nonlinear effects, enabling it to support higher optical power transmission without signal distortion, thereby allowing longer unrepeatered transmission distances.
Although hollow-core fiber remains in the early stages of industrialization, its potential value should not be overlooked. At OFC 2026, hollow-core fiber was listed as one of five key trends. For optical transceiver purchasers and data center planners, the maturation of hollow-core fiber could alter the cost structure and performance boundaries of data center interconnects within the next three to five years. While it will not directly impact optical transceiver selection decisions in the immediate term, maintaining awareness of this technology will enable more forward-looking decisions in medium- to long-term network planning.
Market Size Outlook: The Path to $60 Billion
From a longer-term perspective, the outlook for the optical transceiver market is compelling. Under LightCounting‘s neutral scenario, the global optical transceiver market is expected to approach $60 billion by 2031, corresponding to a CAGR of over 20% from 2025 to 2031. This projection implies that the optical transceiver industry will, within the next five years, complete a leap from a roughly $20 billion market to a nearly $60 billion market, establishing itself as a truly significant core technology sector.
The core drivers underpinning this long-term growth include at least the following four factors: first, the continued scaling of AI large-model training clusters, which drives demand for higher-bandwidth, higher-port-count optical interconnects; second, the evolution of data centers from centralized to distributed architectures, spurring sustained growth in DCI coherent optical transceivers; third, the maturation of edge computing scenarios such as autonomous driving and industrial IoT, which open entirely new market spaces for optical transceivers beyond traditional data centers; and fourth, the ongoing “optical replacing copper” trend penetrating ever-broader application scenarios, with optical interconnects extending from long-reach to medium- and short-reach, and from inter-device to intra-board and even chip-to-chip domains.
It is worth noting that while the growth outlook is optimistic, the risk of cyclical market fluctuations should not be underestimated. LightCounting specifically notes in its forecast that the optical transceiver market tends to experience a downturn approximately every three years, with the two most recent occurrences in 2019 and 2022, driven primarily by slowing capital expenditure growth among leading cloud companies and inventory accumulation across supply chain tiers. For industry participants, maintaining sensitivity to capital expenditure and inventory cycles while embracing growth, and building resilient supply chain management capabilities, are essential for navigating through cycles successfully.
From the production ramp-up of 1.6T, to technology pre-research for 3.2T, to the market expansion of coherent optical transceivers and the potential breakthrough of hollow-core fiber, the optical transceiver industry stands at a critical window of simultaneous acceleration in both technology iteration and market expansion. Over the next five years, the industry is poised to grow from a market of approximately $26 billion today into a massive industry approaching $60 billion. For optical transceiver purchasers, the key imperative is to make prudent yet forward-looking selection decisions in an environment of rapid technological change — ensuring that current deployment needs are met while preserving sufficient compatibility and expansion headroom for technology upgrades over the next two to three years. For optical transceiver suppliers, sustained investment in cutting-edge technology R&D, deepening supply chain integration, and cultivating long-term customer relationships will be essential to achieving steady and enduring success in this marathon of high-speed growth.
