1.4 Datacom and Telecom
Datacom and Telecom
Datacom and Telecom
You can place or change feedback until 31 December 2026, 17:00
The data explosion generated by the growth in social networks and digital entertainment, cloud-computing, and IoT are radically driving the growth of data centers and the need for high bandwidth, low-latency, low power consumption fiber-based communications. Since, 2022 this growth has accelerated to unprecedented levels due to that rapid adoption of AI in general and Generative AI in particular. This has changed the dynamics of the market to the point that telecommunication companies are not driving next-generation technology platforms and metrics, but the datacom (amongst other segments), the datacenter industry (Web 2.0) social media companies (such as Google, Facebook, Microsoft, Amazon etc.) are. These forces in general are transforming the data center architectures to a higher level of integration of photonics components.
Internet traffic is one of the important metrics that is being used to show activity in telecommunications as well as data communications (which includes datacenters and high-performance computing). Internet Protocol (IP) traffic has typically been used to gauge the amount of data that is being transferred through the internet. To accommodate the strong growth in Internet data traffic, the fiber-optic infrastructure that allows data to be communicated between network nodes and within datacenters, must be upgraded. Photonic components are used to build the fiber-optic infrastructure; they comprise devices like laser diodes, photodetectors, multipliers, modulators, and transceivers. These are known as discrete components, while a mix of these components that are integrated or connected on a single substrate (such as silicon, InP or GaAs) are known as PICs or photonic integrated circuits.
Three major segments exist in the datacom and telecom market: Telecom core/metro, telecom access, and datacom. While this report will continue to use the phrase “datacom”, it is important to point out interchangeably using the phrase “datacenter” would be acceptable given the dominance of this sub-sector in driving revenue and innovation in this segment. The growth of the telecom core/metro and datacom segments are expected to experience very strong growth over the next decade. One of the key metrics that is needed for any market analysis is how the market for photonics components will grow over the next decade from a PIC perspective. This is important since the trend to integrate photonics components in PICs is beginning to accelerate. This trend is driven by applications that require smaller photonic component solutions, lower power consumption, high data rates, hybrid packaging, longer interconnect lengths, and scaled economics in terms of $/Gbps.
While the rise of PIC based technologies is exciting, equally exciting in the photonics component market is the increase of the market for transceivers. Transceivers are small boxes at the end of each fiber-optic link that comprise photonics components and PIC components. The photonics transceivers sub-segment is expected to grow to quickly over the next decade from the rapidly ramping the deployment of 800Gbps systems using 60-80GHz bandwidth photonic devices. Within 5 years, 1600Gbps and 3200Gbps are expected.
The obvious yet most difficult next move is to revisit increasing the optoelectronic device speed, and those speeds in particular that are driven not from 60-80 GHz analogue optical bandwidth, but beyond 100+ GHz (typically for most optical communication systems, 40 GHz analogue bandwidth roughly corresponds to approximately 50 Gbaud NRZ using standing coding techniques (and 100Gbps PAM4); while 70 GHz analogue bandwidth corresponds to ~100 Gbaud NRZ, and 200Gbps PAM4. Subsequently, 100GHz analogue bandwidth corresponds to ~150 Gbaud NRZ, and 300Gbps PAM4, which is commonly referred to as 300G per lane.
Further, as is topical today, PAM4[i] encoding is popular today and this technique effectively doubles the data rate, so for example, 250 GHz analogue bandwidth can produce not only ~300 Gbaud NRZ, but also 600 Gbps using PAM4 (which is a lane rate of 600G, 3X today’s lane rate of 200G). At the same time, these new optoelectronic devices must be very small, and operate with very low voltage to keep power consumption low.
The transceiver/modules roadmap that displays the purple brick walls is shown below in Figure 1 and Figure 2. The first figure shows the individual line purple brick walls for each technology segment. The following figure shows the purple brick wall areas that require focus to addesss the challenges to achieve the metrics laid out in the transceiver/module roadmap.
| Purple Brick Wall | Description |
| Product vehicle | >3.2Tbps and 6.4Tbps optical transceiver using multichannels |
| Form factor | Higher density and smaller form factor than OSFP today |
| Lasers | High operating temp (>150C); wavelength stable, uncooled 1310nm |
| Low noise integrated opt amplifiers | Noise factors for optical amplifiers <5dB |
| High speed, low power modulators | >100GHz 3dB bandwidth (EO); 1Volt or less for direct drive from CMOS |
| Waveguide loss | Low loss in visible wavelengths <0.01dB/cm |
| Temperature of operation | 0-85C uncooled |
| Power consumption of TxRx | <1Tbps/10W |
| Full PIC implementation | Includes both Active and passive integrated components on PIC platform |
The main roadmap challenge for the communications industry is to plan future data line rates that exceed 200G and extend towards 300G and 400G. These line rates are needed to achieve pluggable transceiver and/or co-packaged transceiver aggregate data rates of 400Gbps, 800Gbps, 1600Gbps (or 1.6Tbps), 3200Gbps, 6400Gbps and beyond. While companies are looking to complete the designs for 200G line rates in 2023-2025 timeframe, they are also looking how to extend the data rates and line rates further[ii]. Interest in line rates of 300G and 400G are still at an early stage, however, active electro-optic polymers offer modulator device demonstrations today that achieve these goals[iii].
Several vendors are looking at how to address 200G line rates, and are planning spatial multiplexing (adding fibers), wavelength division multiplexing, (adding wavelengths in a single fiber), encoding with more complex symbols per bit
(PAM, QAM etc), designing optical and electrical devices for higher bandwidth, lower voltage (power), and smaller footprint (size) as part of a PIC platform. These higher performance drivers are being accelerated by data hungry customers such as data centers, high performance computing, and shorter reach telecommunications.
The main areas of growth in Datacenters and Telecom will be to support the increasing internet traffic of 50% CAGR primarily because of the growth of Generative-AI, “G-AI”, applications but all as a result of additional growth in cloud computing, video streaming, mobile data traffic and the general expectation by consumers to have higher connectivity options. This growth will also require limited increases in fronthaul and backhaul network capacity. As datacenters evolve from single, core megacenter locations to clusters of small datacenters, disaggregation is taking place requiring growth of high-capacity links between the smaller datacenters. Integrated photonics is of great interest as an enabling technology for disaggregation. Disaggregation promises improved efficiency and data capacity for the data center but is limited by the cost and performance of the links. Latency requirements, in particular, impose hard limits on distances over which certain resources can be disaggregated.
The trends identified by Bloomberg and other analysts noting the explosive growth of G-AI, as well as trends noted in the IPSR roadmap have accelerated over the past decade. While the IPSR roadmap primarily focuses on the number of devices connected to the data center that could lead to the adoption of billions of IoT devices for appliances as diverse as HVAC, medical instruments and factory automation, Bloomberg and others focus on the need to accommodate increasingly complex photonic devices that will be widely deployed in either edge or core datacenters.
The massive amount of data from these fast-growing elements are unstructured and the demand for analytics and fast movement of data is also accelerating. In addition, the importance of security has been repeatedly highlighted with multiple high-profile situations gathering world-wide attention.
The Data Center of today is seeing a transformation from independent computing, storage and networking systems to integrated system units. As noted above, this change is driven by the twin demands of comples G-AI focused devices and mass quantity IoT devices. These devices will increase the number of devices that are interconnected thereby increasing the amount of data that is transmitted, sorted, analyzed and distributed.
Issues involving processors, storage, networking and switching fabrics and their future trend toward increased integration are all driving the packaging requirements for data centers. The physical entities that constrain or direct development of packaging for these subsystems are the thermal, electrical, photonic and mechanical metrics that define the system characteristics.
Thermal and power management of the system continue to be major challenges for Data Centers. A key component of the total cost of ownership is the cost of the energy to power the high-end equipment and, in some locations, the amount of energy that can be delivered to a data center is at the limits of what the utilities can deliver. Thus integrated silicon photonic systems must equal or better electronic systems in terms of cost, energy use, and performance if they are to achieve widespread utilization in the data centers.
To achieve reduced energy use, the American Society of Heating, Refrigerating and Air-Conditioning Engineers (ASHRAE) is creating standards where the temperature and humidity can be higher than has been traditionally required for high-end equipment. In doing so, there is less energy required to cool the data center. Therefore, there is a need for better thermal technologies within these systems to maintain acceptable junction temperatures. Further, ASHRAE recommends that server inlet temperatures be between 18 and 27 degrees Celsius (64.4 to 80.6 degrees Fahrenheit), with relative humidity anywhere between 20 and 80 percent.
Likewise, the active power management of the components that make up a system to maximize the power- performance of the system is also a desirable capability. More efficient power conversion will also continue to be developed including voltage regulation close to the loads (such as in microprocessors) that significantly cut the power distribution losses within the system.
However, the biggest challenge is moving the data with minimal latency. The bandwidth requirements have resulted in the off-die data rates continuing to increase even as the processor clock frequencies have stagnated. This change has been enabled by low-power transmitter and receiver designs with continually advancing sophistication in the equalization techniques. The total amount of power for these interfaces is constrained by the amount of energy available as stated above. The parallelization of interconnects is limited by the expectation that the price of the interconnection will also be approximately constant. This results in the current trend of quickly increasing data rates per physical channel.
Packaging technology must increase at a rapid rate to achieve the new performance requirements. As breakpoints are reached, lower loss laminates, smoother copper, higher bandwidth connectors are required. Development of these components in high-volume manufacturing with cost-competitive materials and processes is ongoing.
Optical interconnects are becoming more widely used with VCSEL data rates reaching 50 Gbps and silicon photonics becoming commercially viable. The need for increasing bandwidth to move the data as well as the growing size of data centers defining the distance the high bandwidth interconnect must travel is creating more need for optical communication. At this time 3 meters has the potential to be the breakpoint between copper cables and optical fibers but this is expected to fall to 2 m and 1 m over the next decade. As aggregated data rates surpass 400 Gbps, and increase to 800 Gbps, and even 1600 Gbps, on-board-on-chip photonics will be required.
Internet traffic is one of the important metrics that is being used to show activity in telecommunications (which include trends in end-user connectivity) as well as data communications (which includes datacenters and high-performance computing). Internet Protocol (IP) traffic has typically been used to gauge the amount of data that is being transferred through the internet and this graph is shown below in Figure 3(source: Cisco VNI). In Figure 3, global IP traffic is forecasted to approach 400 exabytes per month by 2022.
The metric is Exabytes per month. An Exabyte is 1018 bytes, which is 1000 Petabytes, or a million Terabytes or a billion Gigabytes of data. The graph shows a strong growth of 26% CAGR (2017-2022), the majority of the traffic is being driven by internet video, and is very fast approaching the metric of Zettabytes per month, which is 1021 bytes, of data per month. Some estimates are discussing the further metric of Yotta which is 1024 bytes of data over the next decade, which is also expected to be driven by internet video but also by emerging web/data, gaming, and, Internet of Things (IoT) devices.
As noted above, it also useful to track end user connectivity trends as user expectations will impact IP traffic growth patterns.
However, “traditional” Internet and user traffic data and forecasts are quickly becoming overshadowed by intra and inter data center traffic patterns which are driven by the rapid adoption on G-AI. While publicly available data is limited, secondary evidence of this growth can also be seen by the growth in computing power and high speed = largely photonic enabled component - link sales
To accommodate the strong growths in data traffic, the fiber-optic infrastructure that allows data to be communicated between network nodes and within datacenters, has to be upgraded. Today, fiber-optic networks are a combination of long, medium and short optical interconnects that range from approximately 3 m to over 1000 km depending on application in the optical network. Photonic components are used to build the fiber-optic infrastructure; they comprise devices like laser diodes, photodetectors, multipliers, modulators and transceivers.
These are known as discrete components, while a mix of these components that are integrated or connected on a single substrate (such as silicon, InP or GaAs) are known as PICs. The summary assessment of the entire photonics-enabled marketplace in 2022 is estimated to be over $2T[iv].
In Figure 8 the forecast of Ethernet-datacom transceivers are shown by data rate for the next decade. As data rates keep increasing on a yearly basis, it can be seen that the main drivers will be 800G applications.
It is expected over the next decade that transceivers will be an excellent platform for the accelerating trends of PICs in both telecom and datacom applications; it is well known that transceiver trends over the past decade have been towards smaller devices i.e. smaller transceiver formats and footprints, with higher densities of photonics components.
The data bandwidth demand is resulting in systems with ever faster interconnect speeds, even as processor speed is staying constant. The size of the data centers creates a challenge for power demand creating an increasing focus on power efficiency of the systems. Data centers have an increasing number of systems residing in an environment of higher temperature and humidity which is also subject to corrosive elements. Despite these challenges, data centers need to reduce their power consumption and operation costs while also managing the total cost of ownership of these systems. It is expected that PIC based technological solutions must achieve not only increased performance, but effective reliability and economic scalability to become competitive in the data center.
Needs 2025
Needs 2030
Needs > 2040
[i] https://www.edn.com/the-fundamentals-of-pam4/ Back to text
[ii] Arista presentation from OFC 2022 for OSFP pluggable optical transceivers Back to text
[iii] https://www.lightwavelogic.com/presentation/polymer-modulators-with-50ghz-performance-for-power-consumption-reduction-at-400-800-and-1600-gbaud-aggregated-datarates Back to text
[iv] SPIE Optics and Photonics Industry Report, 2022
Join the working group and help shape the IPSR chapters.
Help us improve this section for the next reader.