The Market Opportunity for Photonics

The trends that will drive the evolution of markets revolve around higher data rates in optical transceivers, more semiconductor content, higher levels of photonic integration and increasing market share for silicon photonics.

By Eric Higham, Director Advanced Semiconductor Applications and Advanced Defense Systems Strategy Analytics

An ever-expanding web of connectivity enmeshes users. The promise and vision of emerging 5G wireless networks and devices intrigues us as we wonder how this wireless standard will enrich our lives. We embrace more and higher definition video capabilities and the world shrinks as social networks grow. These capabilities are not ubiquitous, but companies like Google, Facebook, Amazon and others are developing solutions to provide broadband connectivity to roughly half of the global population that is currently unserved.

Figure 1: IP Data Traffic Forecast

The result of these developments is an explosive increase in data traffic. The solid line in Figure 1 shows the impressive growth of data since 2010, extrapolated to 2030. This is total data, so it has both a wireless and a wireline component. The message in this chart is that every bit of information, regardless of where or how it originates must travel on a network. The networks, access methods, physical transport media and the servers that make up “the cloud” define the trends for the optical market, but data traffic is the growth engine. Of course, no analysis is complete without a discussion of the effects of the COVID pandemic. A rough estimate of the increase in data traffic brought about by the shift to more home-centric activities is shown in the dashed line in Figure 1. The data traffic increase is obvious, but this COVID-induced traffic increase will cause operators to accelerate their network capacity plans by about 3 years.

The optical market has a long history and this segment was notable for its volatility. Traditionally, telecom companies and MSOs (Multiple Systems Operator) have managed the networks supporting the data traffic. This transport network consists primarily of fiber and much of that network is underground. Adding fiber is an expensive and disruptive process, as streets and yards must be excavated. Because of the cost and scope of this effort, the operators typically added excess fiber capacity and waited for that capacity to be absorbed before repeating that cycle.

Figure 2: Data Center Traffic Forecast

The growing importance of data companies and data centers has broken this telco/MSO-driven boom or bust cycle for the optical industry. Exhibit 2 shows Cisco’s 2018 estimate for data center traffic extrapolated to 2030, with the COVID increase shown in the dashed line. This traffic is about six times larger than the data traffic shown in Figure 1, with a similar anticipated growth rate. The details of this traffic lead to some very interesting optical opportunities. Not only is the data traffic larger than the transport network traffic, roughly 75% of that data remains within the data center. In addition, this estimate does not address “rack local” data that stays within the same server rack. This traffic is estimated to be up to twice as large as the data traffic within the data center, creating a very large opportunity for short-reach, high data rate connections and transmission.

Data center traffic has smoothed the growth trajectory of the optical market, removing much of the cyclical nature described earlier. The datacom companies (like Facebook, Amazon, Netflix and Google) do not worry about “monetizing the pipe”. To these companies, the pipe is a conduit expanding and extending the reach of the services they offer. Their business model is getting internet access to more people to enable more clicks, video streams, searches and internet orders. This value has become apparent as we moved home during the pandemic and the home, rather than the office became the center of data traffic. Once a data center is built, additional capacity means buying and installing more servers, not digging up streets. This change in business model has created an inflection point for this market and has reduced the volatility and uncertainty substantially.

The Opportunities
The optical/photonic opportunity is very large. Figure 3 shows an average of recent industry forecasts for optical transceiver revenue. The consensus is for flat revenue in 2021, as a result of the pandemic’s influence on supply chains and the recent silicon chip shortage. Like most electronics markets, this one is heavily dependent on China and the final revenue trajectory will be influenced by factors like how quickly China’s economy recovers from COVID and the timeline for 5G deployments.

Figure 3: Optical Transceiver Revenue

Also of interest are the sources of the discontinuous traffic growth created by COVID. Figure 4 shows the enormous growth of non-traditional sources of data traffic, along with quantifying the shift to “at home” services resulting from the COVID-imposed lockdowns. The growth of these services, along with the expected increase in 5G deployments has operators thinking that the discontinuity caused by COVID will become the new future trajectory.



Figure 4: Why is Data Growing?

Monetizing the Network and Transport Opportunities
So, how does the supply chain capture and monetize these opportunities? As data traffic increases and consumers and businesses demand faster data rates and more capacity, devices, networks, transport and data centers must respond with higher performance. While fiber is an excellent transmission medium for light, it has loss and dispersion. At regular intervals and when signals are injected or removed, the signal is transformed from the optical domain to the electrical domain to be reconditioned before it is converted back to the optical domain for further transmission.

Figure 5: Optical Transceiver Architecture

Figure 5 shows a simplified block diagram of a 400 Gbps optical module using a direct detection scheme. This module architecture illustrates a few notable features of the optical transport architecture. The input “line rate”, in this case 400 Gbps, is demultiplexed into lower speed lanes. These lanes contain the reconditioning and silicon control functions for these demultiplexed signals. After conditioning, the lanes are multiplexed back to the higher line rate. The number of lanes multiplies the opportunity for silicon and compound semiconductor devices.

This architecture also illustrates one of the fundamental challenges. The rapid growth of data traffic demands higher line rates, but there is a cost and complexity trade off to consider. As the number of lanes in a direct detect receiver become prohibitive, module manufacturers are refining coherent detection techniques that achieve higher data rates per lane but depend much more heavily on DSPs and error correction techniques.

Figure 6: Optical Transceiver Module Data Rate Evolution

The chart in Figure 6 shows the expected speed evolution in optical modules. As line rates have increased, module speeds and lane rates have gone up in response. As the chart shows, 100 Gbps modules have become the workhorse of the industry, but we believe 200 Gbps and 400 Gbps modules are growing quickly. Figure 6 shows revenue, and the higher speed modules must improve the $/bit cost metric to see the widespread deployment that is being forecast.

Data Centers and Photonic integrated Circuits
In Figure 2, we saw that data center traffic is large and growing explosively. The data that stays within the data center and the server racks presents an enormous opportunity for short reach technologies that can offer advantages over the legacy copper and optical cable connections. This application looks poised to be a significant driver for photonic integrated circuits (PIC) and silicon photonics (SiPh).

Figure 7 shows a forecast for semiconductor revenue in optical transceiver applications. This market opportunity makes use of several different semiconductor technologies and this illustrates the PIC opportunity. The integration opportunity arises because GaAs, InP and silicon all have different advantages and different technology combinations meet different performance requirements. The result of these requirements is a large opportunity for InP-based integration and devices.

Figure 7: Semiconductor Technology in Optical Transceivers

Silicon photonics will play an increasingly significant role by taking advantage of the manufacturing and cost advantages of silicon. There are significant product, process and packaging efforts underway to take advantage of the benefits of silicon for optical transceivers. Silicon is a very poor lasing device, so a silicon-based transceiver relies on another technology, increasingly InP, for the laser. We expect that the silicon photonics-based revenue will almost triple over the forecast.

Conclusions
Data traffic continues to explode as users demand increasingly sophisticated wireless and video applications, workers embrace “work from home” and more services go into the cloud. Data traffic will also increase as 5G network and device deployment expands. There will be growth opportunities for the networks that enable and transport the data and the data centers that represent the physical manifestation of “the cloud”. To support this data tsunami, the optical content in access and transport networks, along with data center interconnections is also increasing. The trends that will continue to drive the evolution of this market revolve around higher data rates in optical transceivers, more semiconductor content, higher levels of photonic integration and increasing market share for silicon photonics.