Major changes the new decade may bring to the data center switching market

The start of the new year is forecasting time at Dell’Oro Group. Like clockwork, our analysts update their five-year forecasts for the respective markets they track. We review the prior year’s performance and its implications for the new year. We also review our notes from conferences and meetings, and the insights we gathered from our interviews with key decision makers and distinguished engineers in the industry, including system and component vendors, system integrators, Cloud service providers (SPs), Telco SPs, and large enterprises.

As the start of both a new year and a new decade, this is an especially exciting time to share with you the major trends we’ll be watching in 2020 and beyond:

  1. Macroeconomic conditions will play an important role in shaping the demand environment in 2020. We expect that ongoing trade friction between the U.S. and China, combined with the U.S. presidential election, may create market uncertainty.
  2. Macroeconomic headwinds, among other issues, will impact spending by Tier 2 and 3 Cloud SPs as well as large enterprises, which may rely temporarily on Tier 1 Public Cloud Providers to expand capacity during this period of uncertainty. However, we expect spending from large Tier 2 Cloud SPs as well as enterprises to accelerate once we move past these issues, as it will be more economical for them to build and operate their own data centers.
  3. Due to the delay in high-volume availability of 400 Gbps optics, the 400 Gbps upgrade cycle (outside of Google and Amazon) will not start to materialize until late 2020/early 2021. Facebook is expected to start its speed upgrade cycle in late 2020 (driven by the availability of Broadcom’s Tomahawk 4 chips). In the meantime, Microsoft’s 400 Gbps story is now slated for early 2021, driven by the availability of 400 Gbps ZR optics for data center interconnect.
  4. The 400 Gbps refresh cycle at Google and Amazon did not have an impact on the performance of any of the branded switch vendors, as these two Cloud SPs mostly deploy white box switches in their networks. However, when Facebook and Microsoft initiate their network upgrade cycles, all eyes will be on market-share gainers and market-share losers.
  5. As China tries to develop its home-grown supply chain, we expect to see more advances over the next few years. In particular, we expect to see progress in switch silicon development, which will further fuel the competition in the space. Over the last few years, numerous emerging vendors have entered the merchant switch silicon market, not to mention Cisco’s latest announcement that it is willing to sell its chips to third parties. It will be interesting to watch which players will continue to be in business five years from now and which ones will run out of steam, as the market cannot support the current number of chip suppliers.
  6. White box adoption has mostly been driven by a few large Cloud SPs. However, as this segment becomes increasingly crowded, we expect white box vendors to try to expand to Tier 2 Cloud SPs, large enterprises, and Telco SPs. We expect the next battle ground between white box and branded switch vendors to be large Tier 2 Cloud SPs, Telco SPs, and the high-end portion of the enterprise market. We predict that branded switch vendors will expand their offerings of disaggregated systems as an answer to the threat from white box vendors.
  7. Optics will play an increasingly crucial role in the data center switch market. The availability of high-volume, low-cost optics has been and will remain the enabler of all speed transitions. Additionally, as network speed increases beyond 800 Gbps, pluggable optics will hit density and power issues. When this occurs, the industry will be forced to adopt alternative technologies, such as co-packaged optics (CPO). We expect such a transition to bring major disruptions to the supply chain as it will require a new business model. We further expect to see numerous acquisitions, consolidations, and partnerships among switch chip vendors, switch system vendors, and optical transceiver vendors. It will be interesting to watch which players will thrive and turn the transition into an opportunity to gain share in the market, and which players will fail navigating through the transition.
  8. As adoption of 400 Gbps speeds and higher increases in the coming years, installation of DWDM optical modules into switches instead of DWDM transport systems for Data Center Interconnect (DCI) is expected to increase.
  9. We expect the number of deployable use cases for data center switching products to continue to expand outside the data center, driven by improved merchant silicon as well as the improved capabilities of network operating systems (NOS) to take advantage of these advances in chip technologies.
  10. We expect 100 Gbps SerDes technology to drive new ways for connect servers to Top of Rack (ToR) switches. 100 Gbps SerDes will be associated with a lot of channel loss, which makes it difficult for conventional Direct Attach Copper (DAC) to cover distances longer than 3m and to continue to be used for server-to-ToR connectivity.
  11. We expect that machine learning and artificial intelligence applications may drive new ways to interconnect pools of resources (compute/storage/memory) inside the data center.

The Optical Transport market has continuously evolved, giving consumers around the world one of the most precious assets—fast, affordable bandwidth. Whether it is 5G, home broadband, cloud services, or videos of cats, the one technology that enables the existence of these services is optical. So, it is no wonder that the one equipment that service providers (telecom, cable, cloud, etc.) need to continuously invest in and purchase is Optical Transport gear. Luckily for these buyers, market demand has grown at a rate fast enough for optical system and component manufacturers to continuously invest their R&D money towards developing better optical technology that transports more bits at an even better cost. What if this changes in the future?

Consider this: every generation of optical technology costs more to develop. Meanwhile, Shannon’s limit is around the corner, and fewer optical companies generate enough profit to maintain this pace of innovation. I won’t address this long-term concern here, but considering this, I have listed a few trends to watch in 2020.

  • Selective vertical integration will continue. System houses such as Ciena, Huawei, and Infinera will continue to invest in developing key component technologies to ensure they capture a significant share of the optical systems market and reduce the bill-of-materials (BOM) on highly advanced coherent line cards. When Cisco closes its acquisition of Acacia, the number of Optical Transport vendors that in-source high-end components (coherent DSP, TIA, drivers, and modulators) will increase. In 3Q19, these four vendors had approximately 60 percent share of the WDM market. If we also consider vendors that have in-house coherent DSPs, this share jumps to 70 percent and could potentially increase if additional vendors decide to “make” rather than “buy.”
  • Coherent 800 Gbps-capable line cards will enter the market. We know of three vendors—Ciena, Huawei, and Infinera—that will launch 800 Gbps-capable line cards by the end of 2020. Ciena will be first to market, closely followed by Infinera, and then Huawei. These new line cards will use the latest optical components (90+ Gbaud, photonic integration) and most powerful coherent DSPs with probabilistic constellation shaping that will bring the wavelength performance to near Shannon’s limit.
  • A faster shift away from 100 Gbps wavelengths to 200 Gbps and 400 Gbps wavelengths. The use of 200 Gbps wavelengths has already been rising to maintain a steady price-per-bit decline. With the availability of 800 Gbps-capable line cards, the market will increasingly deploy 400 Gbps wavelengths this year. That is, with 800 Gbps-capability, a line card can be employed at 400 Gbps across longer span lengths, making long haul 400 Gbps at an economical price point a reality.
  • Coherent 400 Gbps in a pluggable form factor is here. There is no denying that coherent optics will shrink into a pluggable form factor. Both Inphi and NeoPhotonics have announced tests and trials of 400G ZR in QSFP-DD and OSFP form factors as well as a 400G ZR+ version in a CFP2-DCO. It is still a little early in the year, but we believe systems using these pluggable 400 Gbps modules will enter the market by the end of 2020. However, significant sales volume may not occur until 1H21.
  • Adoption of IPoDWDM will increase. IPoDWDM isn’t a new concept. It has been available for over a decade, and Cisco has done quite well (relatively speaking) selling IPoDWDM systems. However, this architectural approach never really obtained wide-spread adoption. We think one of the many reasons behind this is that the target market was on core routers stationed in the long-haul network. A better opportunity for IPoDWDM, as evidenced by Inphi’s sales of ColorZ to Microsoft, lies in selling IPoDWDM in metro access applications such as data center interconnect. Therefore, with 400G ZR in a QSFP-DD form factor, we should see a wider adoption of IPoDWDM in metro applications such as data center interconnect and distributed access architecture (DAA).
  • System vendors will move into the components market. This may be a more “why not?” situation. If a system vendor develops a component, why not make it available for others to buy, since selling any components will help offset the company’s R&D costs? Also, at the end of the day, it is a good hedge against IPoDWDM. If you don’t win the system business, why not try to win the optical components portion?

Each of these listed trends are by no means revolutionary. In fact, each has been in the making for many years to sustain one goal—fast, affordable bandwidth.


As we enter a new decade, I would like to share my view on the key trends that will shape the server market at both the cloud and edge. While various use cases of enterprises running workloads in data centers on premise will persist, investments will continue to pour into the major public cloud data service providers (SPs). Workloads will continue to consolidate to the cloud, as cloud data centers scale, gain efficiencies, and deliver transformative services. In the longer-term, we forecast compute nodes could shift from centralized cloud data centers to the distributed edge as new use cases arise that demands lower latency. The following are five technology and market trends in the areas of compute, storage, and network to watch:

  1. Evolution of Server Architecture

Servers continue to densify and increase in complexity and price point. Higher-end processors, novel cooling techniques, accelerated chips, higher-speed interfaces, deeper memory, flash storage implementation, and software-defined architectures are expected to increase the price point of servers. Data centers continue to strive to run more workloads with fewer servers in order to minimize power consumption and footprint. Storage will continue to shift toward server-based software-defined architecture, thus dampening demand for specialized external storage systems.

  1. Software-defined Data Centers

Data centers will continue to become increasingly virtualized. Software-defined architectures, such as hyperconverged and composable infrastructure, will be employed to drive higher degrees of virtualization. Disaggregation of various compute nodes, such as GPU, storage, and compute, will continue to rise, enabling enhanced resource pooling and, hence, driving higher utilization. IT vendors will continue to introduce hybrid/multi-cloud solutions and increase their consumption-based offerings, emulating a cloud-like experience in order to remain relevant.

  1. Cloud Consolidation

The major public cloud SPs – AWS, Microsoft Azure, Google Cloud, and Alibaba Cloud (in Asia Pacific) – will continue to gain share as the majority of small-medium enterprises and certain large enterprises embrace the cloud. Smaller cloud providers and other enterprises will inevitably migrate their IT infrastructure to the public cloud due to its increased flexibility and feature set, improving security, and strong value proposition. The major public cloud SPs continue to scale and drive towards higher efficiencies. On the longer-term, growth among the large cloud SPs are projected to moderate, due to on-going efficiency improvements from the server rack to data center, and consolidation of the cloud data centers.

  1. Emergence of Edge Computing

Centralized cloud data centers will continue to drive the market within the forecast period of 2019 to 2024. At the end of this time frame and beyond, edge computing could be more impactful in driving IT investments because, as new use cases emerge, it has the potential to shift the balance of power from cloud SPs to telecom SPs and equipment vendors. We anticipate that cloud SPs will respond by developing edge capabilities internally and externally, through partnerships or acquisitions, in order to extend their own infrastructure to the edge of the network.

  1. Advances in Server Network Connectivity

From a server network connectivity standpoint, 25 Gbps is expected to dominate the majority of the market and to replace 10 Gbps for a wide range of applications. The large cloud SPs will strive to increase throughput, driving the SerDes technology roadmap, and enabling Ethernet connectivity to 100 Gbps and 200 Gbps. New network architectures, such as Smart NICs and multi-host NICs have the opportunity to drive higher efficiencies and streamline the network for scale-out architectures, provided that the price and power premiums over standard solutions are justified.

This is an exciting time, as increasing demand in cloud computing is driving the latest advances in digital interfaces, AI chip development, and software-defined data centers. Some vendors came out ahead and some were left behind with the transition from the enterprise to the cloud. We will watch closely to see how vendors and service providers will capitalize on the transition to the edge.


At the end of each year, I like to reflect on the key trends that I believe will drive the products, purchasing, and messaging in the year ahead. I review the past year’s meeting notes and marvel, once again, at my good fortune in speaking frequently with intelligent people. Then, I try to read between the lines to figure out if anything I learned in those meetings foreshadows what’s to come.

I then balance my insights and predictions with the noise from CES as it opens a new year of trade shows. With my inbox full of new product announcements, I can’t help but wonder if products showcased at CES will set the tone for the rest of the year. Will the latest WiFi router, gaming console, and series of IoT devices be the next game changers?  Or will they fall flat like so many other consumer products?

This year, I believe that one of the biggest trends we will see is a fundamental shift in how consumers and service providers think about home networking. A confluence of technologies reaching the market at the same time will have a positive impact on the capabilities and management of home networks, including:

    • WiFi 6: For many years, the evolution of WiFi has been focused on improving two key technical attributes: speed and range. WiFi 6, however, is the first iteration to take a holistic view of wireless technology that encompasses improvements in speed and range as well as network intelligence, analytics, and power efficiency. WiFi 6 also has the capacity to dramatically improve how service providers will be able to provision, manage, troubleshoot, and analyze their in-home networking services. It provides options for remote, zero-touch provisioning of devices and services as well as automatic adjustment of WiFi channels to ensure peak performance.
    • 6GHz Spectrum and WiFi 6E: With so many new connected devices competing for available channels and bandwidth on both the 2.4GHz and 5GHz frequency bands, the WiFi Alliance is introducing WiFi 6E, which uses the unlicensed 6GHz band. In 2020, we expect that many countries will provide access to the 6GHz band. This means that a huge chunk of unused spectrum will become available for the growing number of residential and enterprise WiFi devices. More importantly for cellular operators rolling out 5G networks, the 6GHz spectrum band will allow the provision of seamless handoffs to mobile devices in homes and offices where their networks might have had difficulty penetrating walls and treated windows. There has been much discussion about the pending boom in AR (Augmented Reality) and VR (Virtual Reality) applications for a number of years. With the availability of the 6GHz spectrum, those applications can be delivered, in theory, without fear of latency due to channel contention. 6GHz will provide fourteen additional 80MHz channels and seven 160MHz channels. These will be needed for the intense, high-bandwidth applications.
    • Simplified Control: If you were to compare the user interfaces (UIs) of home gateways and routers from just two years ago to those available today, you’d be hard-pressed to find an area that has seen more positive evolution. But 2020 will see even greater transformation in an effort to give subscribers total and intuitive control over their broadband subscriptions. Voice control of broadband services is one of the areas that we expect will see the most growth. Google’s Nest WiFi mesh systems now include voice control and allow users to verbally turn on a guest network, reboot the system, and initiate parental controls and speed tests. Quietly – and just before the end of last year – Amazon announced Alexa-enabled voice control of its eero routers as well as those from ARRIS/Commscope, Asus, Belkin, Netgear, and TP-LINK. The feature is called Alexa WiFi Access. We expect to see this service integrated across a wider range of devices during the year, including integration into service provider-supplied gateways, particularly those from U.S. cable operators.

These technology developments, coupled with the ratcheting up of competition between service providers and consumer electronics companies for home network dominance, will allow consumers to have substantially better control of their WiFi networks in 2020.

Fast-Tracking DOCSIS 4.0, DAA, and 10Gbps

It seems like just yesterday when details emerged surrounding DOCSIS 4.0, which combines two next-generation technology options for cable operators — Extended Spectrum DOCSIS (ESD) and Full Duplex DOCSIS (FDX) — into a single standard with the aim of delivering 10Gbps services to all customers. CableLabs started drafting the specifications last year. Just this week, the company confirmed that the draft version will become available in the first half of 2020.

At the same time, cable operators are expected to launch their first 10Gbps services this year. However, these deployments are not expected to be tied to the DOCSIS 4.0 specification. Instead, they will rely on 10G EPON from remote OLTs located in traditional optical node housings. While focused on Full Duplex DOCSIS to support the mass market delivery of 10Gbps services to existing residential customers, Comcast is also sprinkling in 10G EPON in greenfield deployments, particularly in regions where it competes with fiber-based ISPs. Other cable operators are following a similar path. But instead of Full Duplex DOCSIS, they will rely on ESD. In both cases, outside plant spectrum will be increased to 1.8GHz.

Regardless of which DOCSIS 4.0 technology path a cable operator decides to follow, a precursor to these deployments will be the rollout of distributed access networks. With the DOCSIS 4.0 standard establishing a clear path forward, cable operators can now move ahead with their remote PHY and remote MACPHY deployments to solve immediate headend space and power consumption issues. At the same time, they can feel confident that any DOCSIS 4.0 technology decision they make will start them on the path toward 10Gbps services.

In 2020, we expect cable operators to ramp up their spending on upstream channel capacity in an effort to improve the subscriber experience with services such as online gaming, as well as reducing the time it takes to upload videos to the cloud. A number of operators have already moved to, or are in the process of moving to, mid-split architectures as they pull fiber deeper into their networks. Mid-split architectures allow cable operators to increase upstream capacity from 5-42MHz to 85MHz, providing a theoretical maximum of around 300Mbps of upstream bandwidth. Like DAA, moving to mid-split is another step on the path toward DOCSIS 4.0. With the implementation of 1.8GHz of spectrum, mid-split will allow an upstream path to span up to 684MHz, a nearly 10x improvement over today’s prevailing upstream rates. More importantly, the move to 1.8GHz will allow operators to flexibly operate on six different upstream path splits, resulting in multi-gigabit services.

Other Trends to Watch

In addition to these trends, we expect to see a significant jump in virtualized access platform deployments. The second half of 2019 saw a major ramp in virtual CCAP deployments. Once again, this growth was largely driven by Comcast, as it continues to expand its R-PHY deployments. We expect this trend to continue both within Comcast and among its peers, particularly Cox Communications and Videotron in Canada.

Outside of cable, we expect to see AT&T make headway in its virtual OLT rollout using XGS-PON. In September, the operator said that it expected to have 100% of its core network traffic controlled by SDN. This was step one in its long-term CORD vision. Access platforms, such as OLTs, will receive the virtualization focus in step two. Though we don’t expect to see any pure white box OLTs in AT&T’s production network in 2020, we do expect to see announcements of SDN control of a good portion of the operator’s access network by the end of the year.