From Europe (3G) to US (4G) to Asia (5G)

We believe that Asia may lead the transition to 5G, similar to what Europe and US did with3G and 4G, respectively. We view countries such as China, South Korea and Japan as mostlikely markets to carry out 5G trials and large scale deployments for telecom operators as well as communications equipment vendors. China Mobile, being the world’s largestwireless carrier by number of subscribers and following the success of its home-grown TD-LTE technology, is actively collaborating with industry participants. Telecom operators in South Korea and Japan have set the earliest targets for their mobile 5G commercial applications – the 2018 Winter Olympics and 2020 Summer Olympics, respectively. Webelieve their compact geography and leadership in 4G also make them suitable markets fora fast upgrade to 5G.

China: In terms of technology advancement, we believe China is becoming more importanton a global scale. From its home grown 3G technology TD-SCDMA which is only adoptedby China Mobile, to its next generation 4G technology TD-LTE which has been adopted by 71 telecom operators globally as of January 2016 including Bharti Airtel, SoftBank,Vodafone etc., China has also been active in the development of 5G technology. China Mobile plans to start 5G commercial trials in 2018 and plans a commercial launch in 2020.

This would put it on par with Verizon, in contrast to 4G where it was more than 2 yearsbehind. CM has established a 5G Joint Innovation Lab (JIL) with 11 partners, including Ericsson, Huawei, Nokia, Qualcomm, ZTE, Datang, Intel, Keysight Technologies, Haier,Hisense, and Beijing Shougang Automation Information Technology. The JIL will build acentral lab in Beijing, and regional labs in the China Mobile International Information Port,Qingdao, Chongqing and other places. China’s CommTech industry has come a long wayover the past three to four decades, from completely depending on foreign vendors andtechnologies to now having some of the biggest telecom operators and CommTech equipment vendors in the world. In October 2015, Huawei and NTT DoCoMo demonstrated mobile internet speeds of 3.6Gbps on a sub-6GHz frequency band outdoors across the cityof Chengdu in Sichuan Province, China. At the MWC in Barcelona in February 2016, ZTE and China Mobile jointly unveiled a 5G high-frequency prototype which operates on a15GHz carrier with a bandwidth of 500 megabits and boasts a hardware structure of alarge-capacity baseband unit and an intelligent remote radio unit.

South Korea: Korea telecom operators plan to launch 5G in 2020 if 5G standards areconfirmed by then. Before the commercial launch, Korea may also showcase thecommercial application of 5G during the 2018 Winter Olympic in Korea. South Korea's Ministry of Science, ICT and Future Planning announced in 2014 that it was committing$1.5 billion to its "5G Creative Mobile Strategy." As arguably the most wired country in theworld, when it comes to network upgrades, Korea's compact geography and existingwireless infrastructure mean that upgrades can happen faster and cheaper, and will reachmore of the population than in geographically spread-out countries like the United States.April 18, 2016 Global: Samsung Electronics started developing 5G technology in 2011 and succeeded indemonstrating 1.2 Gbps data transmission using ultrahigh frequency for the first time inthe world in 2013. In October 2014, Samsung set the first record by achieving a wirelessspeed of 7.5Gbps in tests at its DMC R&D Centre over a 4.35km outdoor race track andusing a 28GHz network. During the Mobile World Congress (MWC) in February 2016,Samsung unveiled the world's first handover technology connecting 5G base stations. At arecent 3GPP meeting in Busan (April 2016), Samsung said it will lead the global standardization of 5G network technologies for the 3GPP RAN1. As a member of the 3GPP,Samsung will announce its plans for integrating diverse IoT services to the 5G network andsecuring compatibility with future 5G technologies that will be further improved by June2017. Based on this research, the company said it will complete the first standardization phase of the 5G network in June 2018.

Japan: Japan’s largest wireless carrier, NTT DoCoMo, plans to launch 5G in time for the2020 Olympics. The company plans to deliver the technology through a variety of “massive MIMO” and 128 polarization elements within the antenna arrays, along with narrow-beam transmission to each user. In October 2015, NTT DoCoMo conducted its first real-worldtests of its upcoming 5G network technology, a 5G data transmission test at a commercial complex in Tokyo, in partnership with Nokia. The test produced 5G speeds in excess of2Gbps. The trial used millimeter-wave signals at 70GHz. In February 2016, researchers from Hiroshima University, Panasonic and Japan's National Institute of Information and Communications Technology have developed a radio transmitter operating in the sub-millimeter terahertz frequency range that is able to carry high speed (100Gbps) dataconnections over multiple channels.

5G as competition for wireline broadband

Fixed wireless broadband access will likely be the first application of 5G, with Verizon planning to deploy it commercially as early as 2017. There are three reasons that we believe 5G may find commercial traction as a fixed wireless broadband service:

1. Improved antenna technology. Recent innovations in antenna technology enablewireless operators to provide 5G services using very high frequency millimeter wavespectrum, which has typically been difficult to use for point-to-multipoint services.These innovations, which we discussed earlier, included MIMO, beamforming andbeam tracking.

Together, these advancements enable 5G antennas to deliver high capacity and highquality signals to fixed customer locations through very high frequency spectrum. Thisis significant because spectrum licenses for very high frequency spectrum typically come in very wide channels (e.g., 100-1000 MHz vs. 5-20 MHz for cellular), which means they can technically enable very high-speed transmissions if their propagation challenges can be overcome.

The same technologies can eventually be used for mobile 5G, but in order to do this over millimeter wave spectrum there will need to be material advances in mobile devices, which are not currently powerful enough to use these capabilities.

2. Fiber densification to support LTE small cells. 5G can be easily deployed as an overlay to the small cell grids that the major wireless carriers are deploying to support their mobile 4G networks. In other words, as operators deploy fiber in dense metro areas to support LTE small cells, they can also attach fixed 5G antennas to these fiberfed LTE nodes. This will provide the dual benefit of getting these 5G antennas close to potential residential customers while providing the high capacity backhaul necessary for a fixed broadband service.

3. Availability of millimeter wave spectrum. The FCC is looking to make 11 GHz of millimeter spectrum available for use in 5G applications in the US. As noted above, the wide channels available at these frequencies enable much higher throughput than we have seen in mobile technologies, which have typically been the basis for past attempts at providing fixed wireless services. With access to much wider channels for fixed services, 5G networks should be able to deliver speeds that are competitive with fiber- and coax-based broadband networks.

Initially, the addressable market for fixed 5G services may be limited. This is a result of fragmented spectrum holdings in millimeter wave frequencies, a need for more fiber density (because the technology works best if the network antennas are within short range of customer locations) and limited availability of commercial equipment. Indeed, because full 5G standards will not be established until 2019-2020, initial fixed wireless applications will be based on pre-standard 5G network gear and devices.

5G is not the only emerging technology that can be used for high-speed fixed wireless broadband. For example, Starry, a private company, is also looking to deploy gigabit speed wireless broadband over millimeter wave spectrum using advanced 802.11ac technology (i.e. Wi-Fi). The key advantage of Starry’s model is that the cost of 802.11-based network gear is much lower than cellular gear owing to the large market for consumer Wi-Fi routers. Facebook is building a similar fixed wireless system called Terragraph, which also utilized millimeter wave spectrum, but is based on the open WiGig standard. So, the market for fixed wireless broadband could become crowded quickly.

From macro cells to small cells; coexisting with WiFi

5G network architecture would likely be characterized by the deployment of small cells, as opposed to the macro cell based architecture in place in today’s networks. While 5G architecture is still fluid, as standardization efforts are underway, it is quite likely that small cells would be critical. A part of the reason for this is necessity (such as for densification), and part of it is due to technological requirements (propagation constraints of high frequency spectrum). Even in fully deployed 4G networks, such as those in the US (AT&T, Verizon, T-Mobile), operators have shifted focus towards deploying small cells for densification of the networks.

A key difference between 5G networks and 4G networks would be the use of higher frequency spectrum (including millimeter wave). In contrast to the crowded low frequency bands, operators would be able to benefit from larger chunks of spectrum in high frequency bands.However, millimeter wave frequencies have poorer propagation characteristics – the higher the frequency of radio waves, the lower the transmission range. This would imply that operators would need to split macro cells into much smaller cells, enabled by the use of small cell equipment. We expect this shift to move a larger portion of the revenue pool from RAN/basestation equipment to small cells.

5G small cells can emerge as a competitive technology to WiFi, given they will have comparable speeds and compete for similar physical space. WiFi offers consumers uncapped internet connectivity (for the most part), in contrast to consumption based cellular services. We would not expect 5G networks to impact the $4.5bn Enterprise WiFi market, as enterprises like to have control over their networks, and as WiFi is the cheaper solution given it leverages their existing wired networks. However, the $300mn Service Provider WiFi market could be partially cannibalized by 5G small cells, in particular the mobile operator portion (as opposed to the cable portion).

One key debate pertaining to WiFi is if 5G networks (more specifically the aggregation of licensed and unlicensed spectrum) will impact the performance of WiFi networks. For example, LTE-U (LTE unlicensed) and LAA (License Assisted Access) would enable LTE connections in the unlicensed 5GHz spectrum band, which is currently used for WiFi networks. Industry participants such as Qualcomm and Ruckus believe that the rules behind LTE-U and LAA can allow them co-exist with WiFi because of “listen-before-talk” protocols that manage interference.

To combat any threat presented by 5G networks, WiFi vendors are pursuing technology advances for WiFi. One example is the WiFi Alliance’s HotSpot 2.0, which enhances the onboarding, security, roaming, and handoff features of WiFi connections, in an effort to create an experience more akin to a cellular service. WiFi standards are also evolving for higher capacity connections (Wave 2 802.11ac) and offer a roadmap towards a broader range of IoT applications (like .11ah for high frequency, high capacity, short range connections and .11ad in low frequency, low capacity, long range connections).

More fiber links from the cell sites to the data centers

Whether it’s high speed cellular connections, fixed line wireless, or disparate IoT endpoints,5G networks will drive more traffic (and likely complexity) on wired and backhaul networks. This will require high capacity and large fiber connections. Therefore, in general, we see 5G connections driving continued investments in fiber network upgrades, particularly around next generation technologies like 100G, 200G, 400G, 1Tb optics. This should prove positive for both optical system vendors (like Ciena, Cisco, and Infinera) as well as optical component suppliers (Lumentum, Finisar). Ciena estimates 5G networks could ultimately drive up to 1000X increase in bandwidth per unit area, based on 100X more connected devices and up to 10Gbps connection rates to mobile devices.

In addition to elevating the capacity of core networks, mobile traffic has a direct impact on backhaul and fronthaul solutions. These are essentially the connections between the radio tower or cell site and the fixed wireline network.

From specialized telecom equipment to servers + software

As detailed in page 17 above, in the radio access portion of 5G networks, service providers would be able to separate the radio and the baseband, with only the radio deployed in physical basestations. The baseband (for signal processing) can be deployed in a data center location (cloud RAN or C-RAN) that aggregates traffic from several basestations. In addition, the C-RAN will host other non-real time functions (subscriber management etc.) that can be virtualized and deployed on servers. This means that with 5G we will see a shift in content away from basestations and other specialized equipment and toward standard servers with NFV.

NFV (network function virtualization) refers to the delivery of networking functions via virtualized software instances on commodity hardware, as opposed to the traditional delivery via specialized telecom equipment (e.g. mobile packet core, session border controllers,application delivery controllers). A lot of the NFV transformation would already be in place ahead of 5G commercialization. For instance, AT&T has set formal targets around virtualization: it intends to virtualize 75% of its network using cloud infrastructure by 2020. Before it began its virtualization journey, AT&T deployed 300 distinct types of telecom equipment in its network.Among the early use cases for virtualization, AT&T targeted GPON optical line terminal (OLT) equipment, and customer premise equipment (CPE). Other service providers who are activelytrialing NFV technologies include Verizon, SK Telecom (Korea), NTT, and Telstra.

The ability of equipment vendors to transition from specialized equipment to NFV will be critical in their ability to maintain relevance in 5G.

On the flip side, we expect demand for servers to increase structurally with 5G as they handle tasks previously addressed with specialized telecom equipment, and as 5G/IoT services proliferate. This will be positive for server vendors, such as Cisco and HP Enterprise,which haven’t historically had much presence in mobile networks. However, these vendors might not benefit if carriers were to adopt white box servers instead, similar to what the major cloud providers have done. In either case, Intel would be a primary beneficiary, given its leadership position in the server processor space, where it has 95%+ market share.We also see 5G and the shift to a more software centric network as an important insertion point for chips based on ARM architecture, both for servers and for networking, given its flexible architecture, clear investment roadmap and low power approach.

Network slicing: Now you can have your own network, too

A number of industry participants, ranging from Cisco to Ericsson to Qualcomm, are architecting solutions to enable “network slicing” as a key capability of 5G networks. A network slice is a virtualized network that can be defined according to a set of requirements, for example by geography, latency, reliability, duration, security, capacity, and/or speed. For example, a carrier such as Verizon that runs a nationwide network can sell various slices of that network to various users, parameterized to fit their requirements: e.g. a utility (for managing smart meters and fault sensors), a healthcare company (for medical device monitoring) and a police department (for mission critical first responder situations).

The enabling technologies for network slicing are cloud and NFV (discussed above). Legacy wireless networks were designed in a more inflexible, vertically integrated way – which made sense given that they largely served one use case (cellphone users) with a relatively predictable growth curve and usage pattern. For next-gen 5G networks, key networking functions will be delivered as virtualized software running in the data center or C-RAN, and thus can be configured in various logical network slices that all share the same underlying physical infrastructure.