Today, Alan Weckel participated in Intel's webinar on how technology is changing from the edge to the cloud. It was clear working on this project that the data center is rapidly innovating to next-generation technologies to keep pace with data growth. How will networks for communication service providers (CoSPs), cloud service providers (CSPs), and enterprises evolve to handle the dramatically increasing data volumes expected in the coming years? Increasing data volumes are being driven today by smartphones, laptops, IoT, and, in the near future, by emerging 5G-enabled services. 650 Group's internal projections indicate that data entering/exiting the data center (north/south) is driven mostly by consumer content (e.g., video). In contrast, a wide range of use cases ranging from enterprise applications, consumer data, and cloud applications drive data between machines.
As part of the webinar, we authored a white paper on how quickly technology is involving in the data center. As we did our end-user interviews during the last few months, we saw many advancements in technology to support the growth of data in the cloud. We are excited to see all the new announcements coming as we close out 2020 and enter 2021.
Please download the white paper by clicking on the link below.
Nokia reiterated its commitment to 25G PON in its two-day briefing with industry researchers this week. It also shared some interesting commentary about is progress with Fixed Wireless Access (FWA) and its consumer Wi-Fi devices. But, what makes Nokia’s 25G announcement so interesting is that there is significant controversy associated with the 25G standardization process; 50G PON is also in the race for standardization, too. It seems that the world will split into two purchasing groups: Chinese and Western. We think the fact that two purchasing groups will emerge is a material negative for the telecommunications industry and is a sign of things to come. Nokia has decided to chart its own path, find partners, and make the best of this controversy. Our view is that Nokia’s 25G PON offerings will see more demand than 50G PON in the upcoming years, and when 50G finally becomes necessary, Nokia can move to support it.
For background, in May 2020, Huawei announced to analysts that it is backing a 50G standards process, in cooperation with the ETSI. Huawei calls its 50G development “F5G,” which stands for Fixed 5G. It demonstrated over a video presentation an FPGA-based prototype, and it and explained that it expects this technology to be adopted first by the mobile infrastructure market for connecting RAN radio systems to baseband systems and for backhaul. Then Huawei expects the market will develop for residential PON, and later for enterprise campus connectivity (to replace Ethernet switches). Huawei explained that in February 2020, it has the support of Chinese operators, ETSI members in Switzerland, a European operator, Altice Portugal, and Chinese operators.
On the other hand, Nokia had developed a chipset that specifically supports both GPON and next generation PON technologies; it is called Quillion and has been available for nine months. Nokia had consistently explained on several occasions in the past several months that during a February 2020 ITU meeting relating to 25G PON, 18 members of the ITU were in favor of initiating the 25G standardization project (including ATT, BT, Korea Telecom, nbn Co, Telecom Italy, SK Telecom, Telus etc). However, there was a minority coalition led by operators and vendors from China that objected to the proposal on the grounds that 25G PON would pre-empt their futuristic vision of 50G PON. This in turn resulted in no consensus being met.
In response, Nokia has worked with operators and suppliers interested in pursuing 25G PON in the near-term, which we interpret as the next 1-2 years. This MSA (multi-source agreement) strategy is used by various groups in the technology industry when there is sufficient buying power to move ahead of (or in this case, without) standards ratification; we see if used frequently by hyperscalers when building their bleeding-edge data center infrastructures. We understand that there are a handful of operators, including Chorus (New Zealand), Chungwa Telecom (Taiwan), and NBN (Australia) and several technology suppliers including AOI, MACOM, MaxLinear, Ciena, Tibit and others. The MSA has a website with more information.
Nokia explains that 25G PON shares the same optical technologies as those used in Ethernet Switches that are common and used by data centers and campus switching environments. Sharing common optical technologies with high volume data center deployments will reduce costs . Our view is that in a few years, data center switching demand for 25G optics will continue to rise, and this is perfect timing for Nokia and others who are going to use 25G for PON because the supply will be there and this technology will be mature and lower cost.
There’s one other thing to consider that pits Nokia against others. It decided to develop its own semiconductors to power its infrastructure PON systems (OLTs). Nokia’s chip system is called Quillion, and its introduction means it won’t be dependent upon OLT chip vendors.
What’s even more interesting about this whole debate is just how future-looking it is. PON has moved through two main generations, GPON (2.5 Gbps), 10 GPON (XGS and XGPON), and now we are talking about two different generations, 25G and 50G. Huawei’s 50G “F5G” approach is a “if you can’t join ‘em, beat em” strategy, where Huawei will leverage its home market telecom operators’ volume and a few others to work outside its home territory. Huawei will leverage this technology to three markets over time: 5G backhaul, residential PON and enterprise networking. On the other hand, Nokia is taking matters into its own hands in that it has developed its own chips. What’s happening now is not uncharted waters, but it is rare for the telecommunications industry to splinter into multiple buying groups – usually standards are developed and followed for the benefit of the industry. This time, in the absence of standards, Nokia has forged on ahead on its own and its headstrong ways are likely to benefit it because many Western operators and now actively seeking to diversify away from Huawei in their procurement of fixed network equipment.
Ethernet Data Center Interconnect (DCI) to Drive Significant Growth in the Ethernet Switch and Next-Generation Router Market
Juniper led the Market in 1H20 and 2019
DCI has a different meaning for different customers and vendors. Ethernet DCI is a critical technology and enabler for Cloud customers to build data centers and transport data between them with lower-cost high-density router/switch platforms with pluggable modules. While DCI has been around for a long time (The Optical Transport Market), Ethernet DCI is different. It uses Ethernet-based Routers and Switches for connectivity between data centers. As we look forward in time, Ethernet DCI will also embrace ZR/ZR+ optics to increase distances in the 400 Gbps and 800 Gbps upgrade cycles. Historically DCI meant routing traffic through Telco SPs.
Today, the Cloud uses its own network and fiber for the majority of traffic until the last-mile to the consumer or enterprise. As an enabler technology, Ethernet DCI is not only the movement of Ethernet platforms into adjacent markets, like Metro Optical Transport, but also new greenfield installations. The ability to move into adjacent markets and new opportunities creates a multi-billion dollar opportunity. It is one of the main drivers for Ethernet-based Switch and Router revenue growth in the Cloud with the 400 and 800 Gbps upgrade cycle. Without this class of system and new optics, the Cloud could not scale, and edge-computing, IoT, and other more modern applications like AI and ML would not be possible.
When we look at 1H20 and 2019 results for Ethernet DCI, Juniper held the number one position in the market with Cisco and Arista rounding out the top three vendors in this segment. Our reporting of Ethernet DCI looks at five different use cases, ranging from Cloud to Colocation to Telco SP. In addition to leading the overall market, our report indicates Juniper also leads in the Cloud/Metro, Colocation/Long Haul, and Telco Cloud use cases with the company’s MX, PTX, and QFX 10K platforms.
In 650 Group's market projections, we expect both next-generation high-density routing platforms and Ethernet Switch platforms to have robust growth. Short-term growth will be driven by strong trends in Content Delivery Networks (CDN) and increased capacity to support Work-From-Home (WFH). As we get into 2021, ZR optics availability will drive additional capacity growth with entering the market in the later half of our forecast. ZR+ allowing Ethernet-based platforms breaking the 1000 km distance. ZR+, under the right conditions (data rate, modulation, and fiber), has the potential to fulfill most connections on each continent. As Telco SPs look towards Cloud architectures and machine-to-machine traffic rapidly goes from inside the data center to spanning multiple facilities, Ethernet DCI will remain a robust market for edge-computing connectivity and the transport of data sets for processing for AI.
Ericsson has served the mobile service provider industry well over the years. Most devices connected to its customers’ networks are mobile phones; this, however, is changing. Internet of Things (IoT) devices are entering the fray and provide an avenue for growth, as is the enterprise market. Additionally, Ericsson’s channels have mostly been to operators, at a time when enterprise growth is expected to provide additional cellular industry growth. Ericsson’s portfolio, until the Cradlepoint acquisition, was not particularly well-positioned to benefit from IoT and enterprise growth vectors.
IoT devices come in all shapes and sizes, and they use a number of different connectivity methods, from cellular to Wi-Fi to Bluetooth to LoRa and many others. In 2020, we expect only 16% of IoT and wirelessly connected devices will connect to cellular systems; the rest connect to more popular (and mostly free) connectivity types. We see cellular connections growing in the future, but as a percentage of all IoT and wirelessly connected devices, we expect it will drop to 13% of all such devices five years from now. The reduction in the fraction of IoT and wireless devices connected to cellular is why the “cellular to other” gateway market (Cradlepoint’s main market) makes sense. There are some use cases where cellular backhaul connections to connect Wi-Fi, Bluetooth, Zigbee and others are vital.
With US-based CBRS and European nations’ private enterprise spectrum opening up the opportunity that enterprises will build their own networks – without needing a mobile operator’s help with sub-leasing licensed spectrum – the folks at Ericsson had a choice to make. The choice was to continue selling to and through mobile operators and hope that mobile operators keep their share of enterprise and IoT growth, or to acquire products and distribution channels to access enterprise growth.
Ericsson’s competitors were partnering with Cradlepoint with some success. Recently, Nokia’s enterprise revenues hit about 10% of revenues, in part because it was selling LTE gear to customers in verticals such as utilities, mining & exploration, and logistics & shipping. Many of these customers were using devices such as Cradlepoint’s. Ericsson is now invited to these ongoing dialogues as these networks expand and change.
We would be remiss if we didn’t mention 5G in relation to Cradlepoint. Some enterprises seek a secondary wireless connection to supplement their primary wired broadband connection. Gear such as Cradlepoint routers can serve this need well. In this sense, we can see why Ericsson uses messages such as “Ericsson accelerates 5G for Enterprise with the Cradlepoint acquisition.”
This acquisition is not without controversy, in our view. The Swedes are acquiring a company located in Boise, Idaho, and as such, managing from afar may present challenges. Cradlepoint sells its devices differently (mainly through channels) from how Ericsson sells its gear (mainly direct); these two distribution methods may conflict. Ericsson sold its cell phone business many years ago because it conflicted with its mobile infrastructure business. Similarly, Cradlepoint gear is focused mainly on enterprises, we see a similar conflict because Cradlepoint’s customer base liked its independence from cellular gear-makers. If Ericsson can manage through these challenges, it may enjoy exposure to IoT, enterprise and 5G gateway growth opportunities.
We attended a well-organized and information-packed virtual conference hosted by Juniper Networks last week. While there were many themes (400G, telco cloud, AI, Mist), the one that really came through was how much the Mist acquisition has reinvigorated the Juniper organization. About a year ago, Juniper made its acquisition of the Wi-Fi startup. It is clear a year later that Juniper’s Enterprise strategy has been “Mistified.” We will share some Mist statistics from conference and our thoughts, as well.
Juniper shared some milestones that now characterize the Juniper Enterprise group:
We asked the company what has worked well with the acquisition, including whether the deal structure may have contributed towards to its success. Juniper executives explained that there was nothing special about how the acquisition was structured beyond the normal incentives for the acquired team. What happened after the acquisition, though, is somewhat unique: in effect, the Enterprise group at Juniper was subject to a reverse merger with Mist, whereby the Mist management team now leads the group. Perhaps the fact that the Mist team is in charge of the show explains the rapid integration of the existing Juniper products into the Mist AI management system. Perhaps, also, the integration went well was because the Juniper products (switches, SD-WAN and Security) were designed to collect and easily share telemetry data for use by the Mist AI system.
Juniper also featured a customer – whose name we cannot share – that is a major fast food chain across the US. It currently has 10,000 Wi-Fi Access Points with another vendor and recently made the decision to replace them with Mist Wi-Fi 6 APs by the end of 2022. We found it interesting how important support is for IoT and the increased importance of outdoor Wi-Fi during the pandemic. The company is deploying four indoor APs and three outdoor APs per restaurant. The customer spokesperson shared that Mist does AI-based, dynamic radio resource management (RRM) very well and he shared numerous screenshots showing power levels that were kept around 11-13 dBm instead of values he recalls seeing closer to 20 on the existing infrastructure. At this point, the customer has installed Mist APs in over 50 restaurants and he says he has not had a single Wi-Fi-related trouble ticket from these locations. He is also looking to implement location services in the future with the company’s vBLE technology.
The company’s cloud-based location analytics capability is designed to serve as a contact tracing and crowd management system to allow Juniper WLAN customers to reduce risks as employees and visitors go to the campus environment. The company introduced this contact tracing capability in May 2020. The company said its cloud-based contact tracing service is something that can operate on top of Juniper infrastructure, as well as its competitors. The company shared details about a large elite US university that is using the service and decided to upgrade to Juniper WLAN infrastructure as well.
Our view is that Juniper is well-positioned in the Enterprise market because by year-end it will have an AI-driven, single-pane operations system that covers Wi-Fi, Ethernet Switching, Security and SD-WAN. This is an enviable position because increasingly customers are making purchasing decisions for various equipment at the same time, and they are looking to reduce ongoing operational costs. By managing a system with a single-pane, customers can correlate customer experience, application responses and network issues in an integrated system, and pinpoint corrective action. This is not to say that Juniper has an exclusive on single-pane managed systems covering all these network systems – Cisco, Fortinet and others lay claim in the same. Juniper is in good company.
Alex Choi, SVP, Head of Strategy and Technology Innovation at Deutsche Telekom, presented at Day 2 joint keynote broadcast for ONF's Spotlight event on 5G and open source. He shared several comments about DT’s strategy that we thought were interesting and showed that the German telco is looking to break away from old ways of doing business.
Activity surrounding Open RAN is hitting a fever pitch. We have been seeing accelerating operator and vendor announcements supporting Open RAN, and now the Open Networking Foundation has announced that it is launching SD-RAN to complement Open RAN. The plan for SD RAN is to open up critical portions of the RAN architecture, allowing both open source and vendor based microservices, called xApps, software connect to the SD RAN architecture’s Radio Ixxx Controller (RIC).
To date, we’ve seen vendors like Parallel Wireless, Mavenir, Altiostar, Samsung and Nokia throw their weight behind Open RAN. Japanese operator Rakuten has been very vocal about its successful commercial launch in April 2020 that uses Open RAN and a virtual computing system to support various RAN functions such as baseband. ONF’s SD RAN project takes things another step, though, by allowing operators and vendors to to leverage open source in the RAN environment.
Getting there presents a challenge. With its announcement, the ONF will support a nRT-RIC and xApps, this is the intelligence that needs to be opened up, according to Timon Sloane, VP for ecosystems and marketing for the ONF based in Menlo Park. He says that functionality from a powerful RIC and xApps can finally deliver the integration and benefits needed for an open approach to work.
Adding some muscle, the open RAN development community, and associated carriers globally, have shown their support for this latest project, a software defined RAN that will put a focus on open systems for 5G and the deeper integration.
The ONF’s SD-RAN project specifically is backed by a coterie of industry players: The O-RAN Alliance, Telecom Infrastructure Project (TIP), and Facebook. Also, global carriers and cloud providers like AT&T, Google, China Mobile, China Unicom, DT and NTT. Lastly, system/chip companies like Intel, Sercomm and Radisys.
The ONF’s proposed µONOS-RIC, is a microservices SDN controller based on ONF’s ONOS platform. 650 Group is bullish on this effort as previous attempts have not come to fruition and the ONF has already had lots of success with its CORD/cloud edge data centers and broadband access with the likes of AT&T DT and Comcast.
Like many companies that sell campus networking gear, Ubiquiti Networks saw a slowdown in 2Q20. Its Enterprise-related revenues grew only 4% Y/Y and were down 16% Q/Q. We reviewed the public material from its disclosures this morning, plus as we do during each of the quarters, we are making checks along the way because we assess Ubiquiti’s market share in many of its markets like Ethernet Switching, Enterprise WLAN, Consumer WLAN, routing and security.
The company experienced production delays in the quarter, primarily as a result of its main manufacturing site being located in southern China. It has established subcontract manufacturing relationships recently where parts are made in Vietnam and Taiwan. The company has been penalized with tariffs because many of its products are made in China, so it has an incentive to get out of the PRC. Its facility lease in China ends in a year, and we expect that Ubiquiti will begin using subcontract manufacturing outside of China increasingly.
Inventory and purchase obligations are at a record high. At the end of 1Q20, inventories had dropped, probably because of shutdowns in China, but inventories grew nearly 40% Q/Q in 2Q20. We believe that the company is expecting revenue growth in the future, based on its high inventory and purchase obligations.
The company attributes its growth to the expansion of distribution channels and expansion of its product line. Since the pandemic shutdowns hit, it appears that the company has not grown its distributor count appreciably. In previous quarters, it had grown its reseller and distributor counts, fueling growth. Coincident with the company’s supply chain difficulties, we have noticed that the company is having trouble getting important new products to volume. For instance, its Amplifi Alien 802.11ax product, while introduced months ago, is unavailable for purchase. We have evidence that some volume was available during 2Q20, though. We see this type of difficulty getting products to volume as related to the sequential growth challenges the company experienced. But, the company has record purchase obligations, so we think it is just a matter of time before it has 802.11ax consumer – and enterprise-class – WLAN products in the market. Our hunch is that by 2H20, the company will have 802.11ax enterprise WLAN products in the market.
Speaking of WLAN, since Ubiquiti is selling primarily 802.11ac products at a time when the market is moving towards the newer generation 802.11ax, this is effectively shrinking Ubiquiti’s addressable market as about 1/3rd of enterprise Access Point revenue is related to 802.11ax. Additionally, the company has significant exposure to smaller customers, which are being hurt more during the shutdowns than larger ones.
Ubiquiti has been a share-taker in the enterprise WLAN market for many years. But, with the short-term challenges it is experiencing (supply chain, distribution, older product portfolio, customer exposure), its share-taking ended in 2Q20. It looks like the company is taking steps to address the supply chain and product refreshes. However, its exposure to smaller customers and its challenges in expanding distribution are more difficult to fix during the pandemic.
AWS (Amazon Web Services) grew nearly 30% Y/Y, remarkable results for a $10B a quarter business. 650 Group enterprise interviews indicate that IaaS is the preferred platform for new application development in the new-normal COVID-19 world.
We do expect at some point enterprises will move some of these workloads back to the premise, but don’t expect a headwind. Still, more of normalization as this premises-based move in 2021 will be occurring right as AI workloads add a new leg of growth to IaaS providers.
AWS Custom ASIC and semi-custom ASIC development include many projects beyond Annapurna’s Smart NIC and Amazon’s investment into satellite connectivity with a $10B investment in project Kuiper for low earth satellites in direct competition with SpaceX’s Starlink will make the company's Cloud platform even more popular. Also, if satellite connectivity is just for media, which we see as unlikely, the way consumers connect their devices over the next decade is going to go through a significant transformation, and this is just the best-case will have a minimal impact.
Apple’s Cloud business remains locked between business models. Apple monetizes the Cloud portion via hardware sales with little direct revenue coming from Cloud services.
While Apple announced a significant networking advancement with Nokia’s new data center switches, a substantial step in the right direction, we have yet to see a significant increase in organic data center spend.
Instead, Apple utilized other large IaaS providers for a lot of Apple’s features, which, to a certain extent, we see a conflicted to Apple’s vertical integration of the company's portfolio. We expect Apple’s organic data center spending to increase significantly over the next few years as the company further vertically integrates and competes directly with Google/Facebook on features and capabilities even though the three companies monetize them differently. For example, computing is just as critical in Augmented Reality and Virtual Reality as the handset itself.