Sales of Ethernet Transceivers Increased 33% in 2020. Expect More Growth to Come, Despite Co-packaged Optics on the Horizon.
LightCounting releases its March 2021 High Speed Ethernet Optics Report
Sales of Ethernet optical transceivers set a new record in 2020, reaching $3.7 billion – up 33% from a slow 2019, when the segment declined by 18%. Suppliers of optics recovered promptly in Q2 2020 from the disruptions caused by COVID-19, and demand for all products ranging from 1GbE to 400GbE exceeded expectations for 2020 in total. LightCounting’s newly updated High Speed Ethernet Optics Report includes detailed historical shipments of these products from 2016-2020 and an equally detailed forecast for 2021-2026.
Service providers and enterprise customers scrambled to add enough bandwidth to keep up with data traffic growth in their networks in early 2020, but slowed spending in the second half of the year. Consequently it was the Cloud companies that had the greatest impact on the demand for optics in 2020, as they continued spending apace through year-end.
Data collected by LightCounting shows that sales of legacy products resumed gradual declines in the second half of 2020, while sales of 400G products continued to grow. Innolight reported its first shipments of 2x400GbE transceivers, as Google started evaluations of these modules for deployment in 2022. Sales of 200GbE FR4 transceivers also picked up in the last quarter of 2020 and Facebook plans to ramp consumption of these modules sharply in 2021-2022.
Sales 400G and 2x400G (800G) modules will sustain the growth of the global market for Ethernet transceivers in 2021-2026 at a CAGR of just above 10%. Sales of 100GbE modules are projected to remain steady at just above $2 billion per year. The total market for Ethernet connectivity is projected to reach $6.8 billion in 2026, not including sales of optics co-packaged with switching ASICs, called Co-Packaged Optics (CPO).
CPO is an exciting technology, but it is important to set realistic expectations for its adoption by the market. Apart from numerous manufacturing challenges and meeting the targets set for lower power consumption, end users have to accept CPO as a viable approach for continuing cost reduction. Facebook and Microsoft advocate for creating a new eco-system around CPO and industry standards for manufacturing optical engines, but initial products will be based on proprietary designs. This will be a significant barrier for the largest customers that are spending $0.5 billion or more on optics annually.
The figure below presents our forecast for shipments of CPO engines (a.k.a. ‘chiplets’) and pluggable Ethernet optical transceivers for 800G applications. CPO enthusiasts may argue that this view is too conservative, but we suspect that it may be too optimistic. Forecasters sometimes underestimate the time required by an industry to change direction in design/technology.
For example, when Cisco acquired Lightwire in 2010, some analysts predicted that Silicon Photonics (SiP) will replace all other optical chip technologies in the next 3 years. A decade later, SiP-based products account for less than 20% of optical transceiver sales, but adoption of this technology is finally accelerating now. An updated forecast for SiP-based products will be published in May 2021 with our Report on Integrated Optical Devices.
It’s certain that customers will start evaluating CPO as soon as it is available, and very likely that Cloud companies and very large enterprises will make limited deployments of CPO in 2025-2026, but massive adoption of CPO may have to wait until well after 2026. Pluggable optical transceivers are set to remain the dominant solution for Ethernet connectivity for the next 5 years and probably for a lot longer than that.
More information on the report is available at: https://www.lightcounting.com/products/HSEO/