September 23, 2020 ■ Cable & Satellite Services ■ Telecom Services: Wireless Gregory Williams, CFA 646 562 1367 gregory.williams@cowen.com Colby Synesael 646 562 1355 colby.synesael@cowen.com Daniel Parr 646 562 1339 daniel.parr@cowen.com Michael Elias 646 562 1358 michael.elias@cowen.com Joseph Kavaloski 646 562 1353 joseph.kavaloski@cowen.com INDUSTRY UPDATE PREPARE FOR LIFTOFF: STARLINK COMING SOON; MINIMAL THREAT TO CABLE BROADBAND THE COWEN INSIGHT With a commercial launch imminent, we provide a deep-dive analysis of Starlink, SpaceX's ambitious satellite-based broadband service. While Starlink has the ability to provide a practical broadband solution for the billions of underserved worldwide over the next decade, our analysis shows it is not well-suited in regions already served by an adequate wireline solution. Ground Control to Major Tom, Commencing Countdown, Engines On... LEO satellite-based internet services have long been dismissed given the multitude of satellites required for an adequate solution. However, advancements in (reusable) rocket technology are being driven by the much broader space-faring ambitions of billionaires Elon Musk and Jeff Bezos, thereby changing the game. Starlink's public beta trial is just a few weeks away, operating >650 satellites today, with additional launches every ~two weeks (60 satellites per launch), for an eventual goal of ~4,440 approved by the FCC (by 2024), and a potential for ~12K satellites. For context, there are only ~3K operational satellites in space today. You've Really Made the Grade... Elon Musk expects to offer 1 Gbps speeds and latency of just 25-35 ms. These targets may just be the beginning. SpaceX filed applications for an additional 30K satellites (a total of ~42K deployed satellites) with the next-gen satellites below 20 ms (ideal for gaming). Furthermore, the business case is less relevant as these satellite-based ISP's are simply byproducts of rocket R&D for larger ambitions (Jeff Bezos' "Blue Origin" to move heavy manufacturing off of our planet, and Elon Musk's goal to colonize Mars), thus no immediate economic mandate for profit. This Is Major Tom to Ground Control, I'm Stepping Through the Door... However, when we consider the threat to Cable broadband, each satellite has capacity of just ~20 Gbps, and our illustrative model shows that by YE 2026, Starlink's 12K satellites can serve ~485K simultaneous 100 Mbps streams over the U.S. at any given time. For context, CHTR and CMCSA have 53.5MM BB subscribers. Furthermore, if we consider Starlink’s plan for a total 42K satellites, combined with Amazon's Project Kuiper (>3K satellites), by the time these next-gen satellites are ready/deployed, data rich apps (8KTV, AR/VR, video conferencing upload) will drive take rates for >600 Mbps plans, plans that only Cable can provide on a large scale (~2026). (Please see full report by opening link) Net/Net: Starlink Ideal for Sparse (Rural) Areas and/or No Internet Today In summary, while Starlink has the ability to provide a practical satellite-based broadband solution for the underserved, the capacity has limitations in most of the U.S. especially considering the growing demand for bandwidth driven by in-home data-rich applications and devices. On the other hand, LEO-based satellites could prove an excellent solution for sparsely populated areas in the U.S. not served by an adequate fixed wireline solution and “the other 4 billion” people worldwide with no access to broadband internet. Please see pages 18 to 21 of this report for important disclosures. COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. EQUITY RESEARCH COWEN EQUITY RESEARCH September 23, 2020 Prepare for Liftoff: Starlink Coming Soon, Though Minimal Threat to Cable Broadband In summary, as our report shows, LEO-based solutions are not well-suited in competitive regions already served by an adequate wireline solution, as specifically, the capacity has limitations in most of the U.S. especially considering the growing demand for bandwidth driven by in-home data-rich applications and devices. On the other hand, LEO-based satellites could prove an excellent solution for sparsely populated areas in the U.S. not served by an adequate fixed wireline solution and “the other 4 billion” people worldwide with no access to broadband internet. How Does Starlink Work? Why Now? LEO Satellites > GEO Satellites. Satellite-based internet has been around for decades. However, most of these solutions we are familiar with utilize Geosynchronous Equatorial Orbit (also known as geostationary, GEO, geosynchronous, or GSO) satellites, which have been in existence for 50+ years, as we consider satellites such as those from Dish/Echostar, DirecTV, ViaSat, InmarSat, Sirius, and most weather satellites. These GEO satellites orbit at 22K miles above the earth (for context, that is nearly “three Earths away”), and orbit in unison with the earth’s rotation, thus each satellite is dedicated to cover a fixed area of the globe. Because of the distance and geosynchronous orbit, a GEO provider could cover the entire globe with <5 satellites in orbit. That said, with so few satellites and the meaningful distance, the speed and latency are largely inadequate (latency > 500 ms), especially as we consider the growing demand and criticality for broadband in the U.S., further augmented by a post-COVID world. As such, GEO solutions are generally relegated as a “last resort” solution. On the other hand, Low Earth Orbit (LEO, non-geostationary, or NGSO) satellites such as those deployed by Starlink are small, cheap satellites orbiting at just ~500 miles (the latest “very low earth orbit” or VLEO satellites at just ~211 miles), barely in space and scraping the upper atmosphere. A LEO solution is made up of a large multitude of satellites (or a “constellation”), orbiting the globe at a rapid pace (LEO’s can complete a full orbit in under one hour). The distance and multitude of satellites (for capacity) can provide a far superior solution in terms of speeds and low-latency. The downside is the said multitude of satellites needed (thousands of satellites), as rocket launch costs have typically been expensive and time prohibitive….however that is beginning to change. Enter the Rocketmen: Elon Musk and Jeff Bezos. The rise of the LEO satellite ISP solution is partly driven by cheaper satellite costs, but largely driven by investment in rocket technology. Specifically, Elon Musk’s SpaceX (funding the Starlink ISP) and Jeff Bezos’ Blue Origin (for Amazon’s Project Kuiper ISP) are investing in reusable rockets driving a far lower marginal cost per launch. Specifically, the advanced heat-shield technology provides far more durability on re-entry for reusability. Elon Musk has a goal of flying reusable rockets twice within 24 hours. As we consider the marginal cost of a 2 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. With the upcoming commercial launch of Starlink, SpaceX’s satellite broadband service, (currently in public beta-testing and a public beta launch in ~November, LINK), and our volume of calls increasing on the subject, we provide our thoughts on Starlink and lowearth-orbit (LEO) satellite broadband solutions in general. While there are other (actual and potential) constellation LEO-satellite providers looking to enter the fray over the next decade (Amazon’s Project Kuiper, OneWeb, Telesat, LeoSat, SES/O3b, Theia), our focus for this report is on Starlink, as the company is ready for commercial availability. While still a private company, with little disclosures (outside of FCC documents, conferences, and Twitter posts), we aggregated all of the relevant information as it relates to Starlink and specifically its place in the U.S. Broadband industry. COWEN EQUITY RESEARCH September 23, 2020 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. reusable rocket vs. a traditional rocket (in which most of the equipment is discarded on each mission), the analogy is akin to imagining the impossibility of ancient exploration “if explorers had to burn their ships each time they reached the New World”. Launch Costs Vastly Improved, but ISP Economics Are Less Relevant, Chasing Bigger Dreams. The lack of disclosures makes a business case analysis challenging. That said, for the purposes of this report (assessing the technology threat to the cable/wireline industry), we note that a business case analysis is especially less relevant considering that these passionate billionaire owners are using personal wealth, and do not have an immediate economic mandate for profit as internet service is more of a byproduct for far larger ambitions. For example, Jeff Bezos is investing in rocket technology for “Blue Origin” with the goal of moving heavy industry off earth and into space, meanwhile Elon Musk is investing in rocket technology to colonize Mars in order for humanity to ensure long-term survival by becoming a multi-planet species. Elon Musk has noted as much, noting Starlink incremental cash flow could help fund the Mars endeavors (LINK), while somewhat factiously saying Starlink’s goal is “not to go bankrupt” as nearly every Satellite ISP has done so in the past. On the other hand, SpaceX President and COO Gwynne Shotwell conveyed more optimistic goals with a potential IPO in “the next several years” stating “right now, we are a private company, but Starlink is the right kind of business that we can go ahead and take public…that particular piece is an element of the business that we are likely to spin out and go public.” (LINK) In any case, according to the latest FCC filings (which we note are consistently being revised), Starlink has approval for 4,409 satellites (LINK), but an initial plan is for 12K satellites on three orbital shells (~1,500 at ~342 miles, ~2,800 at ~684 miles and ~7,500 at ~211 miles, though recent petitions are to move most satellites in the lowest shell for low-latency). The number of satellites is perhaps underappreciated as, for context, we note that there were just ~9K total satellites launched in human history, and just ~3K operating in space today. Starlink currently has >650 operating satellites. Musk has noted a need for ~1K satellites for basic coverage “to be economically viable”, while the FCC requires the first ~4,400 satellites by March 2024. SpaceX has also recently submitted FCC filings to explore a second-generation system calling for ~30K additional next-gen satellites (for lower latency, for a total of ~42K launched satellites). Launch Cadence. SpaceX is currently on a pace of 35-40 launches per year (every two weeks, another launch scheduled in the next few days). Each launch leverages the welltested Falcon 9 rocket, for a payload of 60 satellites per launch. However, SpaceX has indicated potentially leveraging the Falcon Heavy rockets and possibly using the “Starship” when ready, which can carry a payload of ~400 satellites per launch (we assume beginning in 2023 in our illustrative deployment model). The satellites will have a ~4-5 year lifespan, as the low-earth orbit satellites “scrape the atmosphere” and eventually de-orbit back to earth, thus the constellation will need constant replenishing. The current cadence implies Starlink could reach the 12K satellite target by late 2026. How It Works – The Home, the “Bent Pipe” in the Sky, the Ground Station. Keeping in mind that the end goal is for a home/boat/camper to receive internet access, the Starlink LEO constellation system is a simple “bent pipe”, ultimately connecting the end user to an internet on-ramp (i.e. a data center) that could be hundreds of miles away. For an admittedly overly simplistic explanation, the system is essentially made up of three components: (1) the end user terminal, (2) the satellite, and (3) the ground station. The Starlink end user terminal is a small dish (1.5 foot diameter) that sits outside the user’s home. Elon Musk describes the terminal as a pizza box or “UFO on a stick”. It is electrically steered to ensure a connection with the appropriate satellites moving across COWEN.COM 3 COWEN EQUITY RESEARCH September 23, 2020 In order to reach the internet (and for example query a website), the terminal will send its request to one of the satellites crossing the sky. The satellites will use the Ku band (and eventually Ka and V-band spectrum). The satellite, in turn, beams the instructions to a Ground Station, the “brains” of the system. The Ground Station is fiber-fed and ideally in close proximity to a data center for the necessary internet/cloud onramp (we count ~40 Ground Stations in the U.S., building far more over time, and internationally, to reduce ping/latency). Once instructions are received (in this example the website is reached), the process is then reversed as the Ground Station then sends the reciprocating data to the satellite (forward uplink), back to the home terminal (forward downlink). Essentially the system is similar to a typical ISP handshake but instead the satellite provides both the front haul (to the home) and the “bent pipe” back haul to the data center (but unlike an ISP where backhaul is to the telco central office or cable head end a few miles away, this backhaul connection is via the sky and could be hundreds of miles away from the home). 4 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. the sky. Installation is simple with no technician needed, as Musk says “just two instructions in either order: point at sky, plug in.” To that end, the terminal does need to be placed outdoors and requires and unencumbered line of sight to the sky. COWEN EQUITY RESEARCH September 23, 2020 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 1 How Starlink Works Source: FCC filings, Cowen and Company Costs: Lack of Financial Disclosure, but Rocket Technology Is the Game-Changer. The costs of the system remains opaque, but gleaning the various FCC filings and public statements (conferences, twitter posts) suggests far lower costs (especially on a unit basis) compared to traditional satellite systems. A typical satellite launch could cost upwards of $100MM for the satellite and >$100MM for the rocket launch. SpaceX President and COO Gwynne Shotwell noted that analysts projecting costs of $1MM per satellite and $50MM per launch are “wayyyyy off” (LINK), separately noting that their costs are “17 times better or cheaper” than OneWeb. Elon Musk has noted confidence in getting the marginal costs down to $5-6MM per launch. He also noted the cost of each satellite is lower than the per unit launch costs, implying a cost of <$500K each. Furthermore, when the Starship (including the Super Heavy booster) is ready in a few years (we assume 2023), which can carry a payload of 400 satellites, the unit costs are expected to be at least 5x lower given the scale and reusability. The costs for the user terminal may still remain a challenge, perhaps higher than initially anticipated. That is, due to the need for a consistent connection, the terminals need to COWEN.COM 5 COWEN EQUITY RESEARCH September 23, 2020 All said, SpaceX President and COO Gwynne Shotwell noted that completing the constellation system may cost “$10B or more”. For our illustrative analysis, if we assume 60 satellites per launch (eventually up to an average of 175 satellites per launch assuming Starship/Super Heavy boosters in 2023), a satellite unit cost of $500K (eventually $400K), a marginal launch cost of $6MM per launch, and factoring in failure/de-orbiting of satellites, these assumption suggest launching the ~12K satellites by YE2026 drives a total marginal cost of $6.7B going forward. Adding end user terminal costs would add a few hundred million dollars. Performance – Finally a Good Satellite Experience, But Not for Denser Locations Latency Target 25-35 ms. Elon Musk has noted that users can expect “gigabit speeds” and latency of just 20-50 ms. Musk noted in a Tweet that Starlink is targeting latency below 20 ms (ideal for gaming) and perhaps below 10 ms over time (LINK) with the next-gen satellites (at very low earth orbit for lower latency) and additional Ground Station links (requiring fewer hops). FCC filings show a target latency of 25-35 ms (LINK). The low latency thresholds will be pivotal should Starlink bid for government subsidies in the RDOF program (Rural Digital Opportunity Fund, see details below). Speeds Depend on Cell Density. While gigabit speeds are certainly achievable in isolation, such throughput would be challenging in denser markets due to oversubscription risk. Similar to a cell phone architecture, if we consider each satellite akin to a cell phone macro tower, the constellation system can only handle a limited amount of throughput per satellite (“per cell”). Taking our macro-tower analogy further, a cell tower has a cell radius of say 5 miles (on low-band 600/700 MHz spectrum). A space-based “cell” would be enormous, covering hundreds of miles; it could only handle so many data streams at once. According to FCC filings (LINK), “each satellite in the SpaceX System provides aggregate downlink capacity to users ranging from 17 to 23 Gbps”. Elon Musk has corroborated this figure to the media, publicly noting that each launch adds 1 Tbps of capacity (assuming the typical 60 satellites in a Falcon 9 launch equates to ~16 Gbps per satellite). Thus, assuming 100% efficiency (not realistic, but we are simply providing context as a high book-end), and assuming 20 Gbps per satellite implies that each satellite can handle 200 simultaneous streams at 100 Mbps. As we consider the U.S. total addressable market, when all 12K satellites are operational, the satellites could cover ~20% of the U.S. at any given time. This could prove a high-end assumption given the U.S. land total is just 9% of total global surface area, though we also consider Starlink is initially skewed to cover higher latitudes of 4452 degrees North, with the current beta tests covering the Northern U.S. and Canada. As such, the math exercise shows that upon completion, Starlink can serve 485K simultaneous U.S. data streams at 100 Mbps. Equating this number to actual customers depends on Starlink’s policy on oversubscription rate, for example assuming a 3x oversubscription rate could imply a 1.5MM TAM (not all customers will be online simultaneously, far from it, though we believe the post-COVID era will drive more need for internet reliability, thus more conservative oversubscription levels). 6 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. be electrically steered, tracking 2+ satellites simultaneously, adding far more cost and complexity than perhaps initially anticipated. Recent media reports suggest the terminals cost $1,000 each (LINK), however at scale the target is to manufacture the terminals at $100-300 each (LINK). Reducing the cost of the terminal could be pivotal for more notable adoption as we consider the low-income customer that Starlink intends to serve (rural, Africa). COWEN EQUITY RESEARCH September 23, 2020 Figure 3 Estimated Capacity and Simultaneous US Customers Served for the Initial 12K Satellites Estimated Starlink Satellites Launched Cumulative Operating Satellites 3,500 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 2 Estimated Launch Cadence for the Initial 12K Satellites Estimated Capacity and Simultaneous US Customers Served Satellites Launched 3,275 14,000 Total Capacity (Tbps) 700 Consumers (K) Served at 100 Mbps 300 606 12,000 600 10,000 500 8,000 400 1,500 6,000 300 1,000 4,000 200 2,000 100 3,000 2,800 250 514 2,500 2,220 2,000 1,560 500 1,560 1,680 480 0 0 2020 2021 2022 2023 2024 2025 200 399 150 291 252 100 161 50 71 0 2026 0 2020 Source: FCC filings, Cowen and Company estimates 2021 2022 2023 2024 2025 2026 Source: FCC filings, Cowen and Company estimates Upload speeds have not been disclosed, but beta tests show upload at 40% the levels of download speeds, which may become more important in the post-COVID distributed workforce environment (as we consider home video conference calls, FaceTime). Lastly, pricing the product also remains uncertain, though Gwynne Shotwell noted “less than what you are paying now for about five to 10 times the speed you are getting.” Figure 4 Illustrative Starlink US Capacity/Subscribers Launch Date 2020 2021 2022 2023 2024 2025 2026 Satellites Failure/ Launched De-Orbit 480 16 1,560 52 1,560 52 1,680 52 2,220 52 3,275 980 2,800 960 Cumulative Operating Satellites 1,177 2,685 4,193 5,821 7,989 10,284 12,124 Total Capacity (Tbps) 24 54 84 116 160 206 242 US EOP Streams Coverage Served (K) at % (Est.) 100 Mbps 30% 71 30% 161 30% 252 25% 291 20% 320 20% 411 20% 485 Source: Cowen and Company U.S. Broadband Consumption: LEO’s Relegated to Sparsely Populated Areas As we consider Starlink’s plan for ~12K satellites, combined with its potential for nextgen satellites (additional throughput, lower latency, and an additional 30K satellites), and further combined with other LEO operators (Project Kuiper looks to become the #2 provider, filing an application for >3K satellites, though Blue Origin is still 5-10 years behind in rocket technology), the threat to Cable and Telco ISP’s appears to be, at first glance, worrisome. However, we contend that U.S. Broadband consumption, and the speeds that users demand, is continuously growing. Thus, as satellite throughput and technology continues to progress, so too will demand for faster speeds. As such, our analysis below shows that LEO satellites will continuously be a step behind wireline Telco/Cable operators in meeting U.S. consumer demand for broadband. COWEN.COM 7 COWEN EQUITY RESEARCH September 23, 2020 Elon Musk recently admitted as much, noting the niche market that Starlink will serve… “The revenue potential for launching satellites taps out around $3B per year; but I think providing broadband is more like an order of magnitude above that, probably $30B per year…and we’re still like probably below 5% at that point. I want to be clear…it’s not like Starlink is some huge threat to Telcos. I want to be super clear it is not. In fact it will be helpful to Telcos because Starlink will serve the hardest to serve customers that Telcos otherwise have trouble doing with landlines or even with cell radio stations…Starlink will effectively serve the 3 or 4% hardest to reach customers for Telcos or simply those that have no connectivity, or the connectivity is really bad.” -Elon Musk, Satellite 2020 Conference, March 9, 2020 (minute 15:20) The Need for Speed: How Much Will U.S. Consumers Demand? In support of taking Elon Musk’s “not a huge threat” words at face value, we provide analysis to show the magnitude and trajectory of U.S. broadband consumption over the next few years. As a starting point, we look at past trends of U.S. broadband consumption, though we do acknowledge, growth will be driven the data-rich use cases and in-home device complexity that will drive speeds, not historical charts. Nonetheless, we look at trends based on our Cowen proprietary surveys (on plan take rates), “Nielsen’s Law”, Cisco U.S. consumption data, and flagship data plans offered by Cable operators over time. We then look at the underlying said drivers of future demand: data rich applications and in-home IoT coming down the pipeline. We start with admittedly the least scientific method of “Nielsen’s Law”, that is, a highend user’s connection speed increases by 50% every year. We believe the trend, though (very) accurate since 1983, and at 325 Mbps in 2019, will be difficult to sustain over the next decade, as it implies a >20 Gbps connection by 2030E. On the opposite end of the demand curve, we looked at Cisco reports on U.S. average fixed broadband speeds (LINK). For example, Cisco noted the average speed was 43 Mbps in 2017, 57 Mbps in 2018, 70 Mbps in 2019, 93 Mbps in 2020E and expects speeds to increase to 142 Mbps in 2023E. If we extrapolate the historical CAGR (of 22%) going forward, we estimate speeds will increase to 567 Mbps by 2030E. Figure 5 Extrapolating High-End User Speed Demand via Nielsen’s Law Figure 6 Extrapolating U.S. Avg. Speed Using Cisco Data Speed Projections (Mbps): Extrapolating Nielsen's Law Speed Projections (Mbps): Extrapolating Cisco Consumption Data 30,000 600 25,000 500 20,000 400 15,000 300 10,000 200 5,000 100 0 0 2017 2018 2019 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E Source: Cowen and Company Source: Cowen and Company 8 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited.  COWEN EQUITY RESEARCH September 23, 2020 Figure 7 Cowen Survey: Subscribers Taking 100 Mbps+ Speeds 4Q17 80% 1Q18 >100 Mbps Take Rate 3Q18 4Q18 1Q19 2Q19 Figure 8 Cowen Survey: Subscribers Taking 200 Mbps+ Speeds >=200 Mbps Take Rate 3Q19 4Q19 1Q20 2Q20 49% 50% 39% 62% 60% 57% 55% 53% 35% 41% 31% 33% 2Q20 39% 39% 31% 33% 31% 26% 19% 20% 30% 29% 29% 25% 15% 18% 16% 12% 12% 9%10% 8% 7%7%7% 7% 10% 1Q20 25% 37% 20% 4Q19 39% 39% 30% 41% 34% 3Q19 44% 40% 46% 30% 2Q19 45% 63% 62% 60% 59% 60% 1Q19 50% 68%68% 70% 40% 2Q18 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Our Cowen proprietary analysis shows a similar growth trajectory on U.S. Broadband speed demand. Specifically, 68% of Cable survey respondents take speeds >= 100 Mbps, up from 34% from 4Q17, and 44% of Cable survey respondents take speeds of >= 200 Mbps, up from 31% in 1Q19 (when we included the answer “200 Mbps” separately). 12% 10% 5% 5% 6% 1% 5% 0% 0% Total Cable Subs 0% Total Cable Subs Total Telco Subs Total Telco Subs Cable - Telco Delta Cable - Telco Delta Source: Cowen Cable Broadband Surveys 1Q17-2Q20 Source: Cowen Cable Broadband Surveys 1Q19-2Q20 Our survey notes that at the end of 2019, subscriber speeds averaged 271 Mbps. We note that OpenVault highlighted average speeds of 210 Mbps, and Altice noted its customers averaged speeds of 242 Mbps as of 2Q20. As of 2Q20, our survey shows speeds increased to 288 Mbps for Cable subs. Figure 9 Cowen Survey: Weighted Average Broadband Plan Speeds (Mbps) Wtd. Avg. Broadband Plan Speed (Mbps) Total Cable Subs Total Telco Subs Linear (Total Cable Subs) Linear (Total Telco Subs) 310 288 286 290 275 270 250 271 251 242 230 223 221 210 190 272 222 202 180 170 150 1Q19 2Q19 3Q19 4Q19 1Q20 2Q20 Source: Cowen Cable Broadband Surveys 1Q19-2Q20 COWEN.COM 9 COWEN EQUITY RESEARCH September 23, 2020 Source: Altice 2Q20 earnings presentation Assuming internet traffic grows at 25% per annum (also supported by Cogent’s longterm historical observation), extrapolating to speed growth using Cowen’s YE2019 results as a jump-off point implies an average speed of >3 Gbps by 2030E. Lastly, another way of looking at demand is Cable’s “flagship plans” each year. In 2019, the typical lead plan was 100 Mbps by most large cable operators. That has since shifted to 200 Mbps in 2020 (Cable One noted its flagship offer is now 200 Mbps, Charter noted its minimum speed offering now 200 Mbps in most markets). Assuming the lead offer increases 25% per annum implies the Cable flagship offer will reach ~1.4 Gbps by 2030E. Relating this growth back to Starlink, by the time the constellation system is complete in ~2026E (and Project Kuiper underway), Cisco data suggests users will be consuming 257 Mbps, Cowen data suggests respondents would take ~1.2 Gbps, and Cable would be offering ~600 Mbps as its flagship offer by then. Thus, the LEO satellite offers will appear obsolete akin to “DSL” speeds of today. Hence our belief LEO’s will offer competitive speeds, but to a very finite cohort of very rural subscribers. 10 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 10 Altice Avg. Data Speeds Taken by Customers COWEN EQUITY RESEARCH September 23, 2020 Figure 12 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 11 Speed Projections (Mbps): Extrapolating Cable Flagship Plans Speed Projections (Mbps): Extrapolating Cowen Survey Data 3,500 1,600 3,000 1,400 1,200 2,500 1,000 2,000 800 1,500 600 1,000 400 500 200 0 0 2019 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E Source: Cowen Cable Broadband Surveys 1Q17-2Q20 2019 2020E 2021E 2022E 2023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E Source: Cowen Cable Broadband Surveys 1Q17-2Q20 The Drivers of U.S. Broadband Demand – Criticality, In-Home Devices, New Use Cases (4KTV, AR/VR), Low-Latency (Gaming), and Upload (Video Conferencing, Security Cams) Broadband as a Consumer Necessity. Broadband has undoubtedly become a utility for all income classes as more of our lives are online and subsequent applications (video, voice, payments, entertainment, home security, in-home IoT) are migrating to the internet as the de facto underlying network. Broadband has become a necessary utility both upstream and downstream. We’ve argued lower-income cohorts seek broadband alternatives, mainly in the form of wireless (prepaid for subprime customers), WiFi sharing, or public hotspots. While the softer economy could drive a negative impact (slower Cable broadband adds and ARPU growth), there remains a greater need to stay connected as lower-income homes are simply rethinking wallet share and prioritizing broadband. Meanwhile, all cohorts look to either replace non-essential services and/or find services that have an internet-based substitute such as video, cinema, phone, and home security. The COVID pandemic has accelerated this dependency with an increase, and level of permanency, in the work-from-home and play-from-home environment. Cable Investment in Speeds Will Continue, Driving the Apps of the Future. As mentioned, infrastructure investment opens the door for Silicon Valley to drive data-rich use cases, which in turn drive consumer data speed uptake. Cable investment drove the use case evolution from web-browsing, telephony, VPN, to short-form video with the onset of YouTube (in 2005). The evolution became a revolution with long-form OTT video, as we note Netflix began online streaming in 2007. Cable has largely deployed DOCSIS 3.1 in 2018/2019 for speeds of 1 Gbps. A typical rule of thumb for Cable is to upgrade speeds every 18 months. Over the next few years, Cable promises additional speed upgrades with the deployment of DOCSIS 4.0, “10G” (for a pathway to 10 Gbps speeds, LINK), and DAA architecture (see page 45 LINK). Cowen Survey: Video Currently a Primary Driver of Speed Uptake. Specifically, our survey works shows that OTT adoption is directly correlated to adoption of higher speed (higher margin) plans. That is, those taking 1+ Gbps speeds subscribe to 3.79 OTT providers, on average, those taking 100-199 Mbps subscribe to 2.88 providers, and those taking 0-6 Mbps subscribe to just 1.44 providers, on average. COWEN.COM 11 COWEN EQUITY RESEARCH September 23, 2020 How Many OTT Providers Do You Subscribe To? 3.79 (Average) 4.00 3.50 2.88 3.00 2.61 2.99 2.58 2.50 1.94 2.00 1.50 1.44 1.00 0.50 0.00 0-6 Mbps 6-19 Mbps 20-49 Mbps 50-99 Mbps 100-199 Mbps 200-999 Mbps 1+ Gbps Source: Cowen and Company 1Q20 Cable & Broadband Survey The Apps of the Future Could Leave Satellite and Telcos ISPs in the Dust. As we consider data-rich apps of the future, Video will continue to be the long pole in the tent. In the next 5+ years, video will need far more throughput as 8KTV demands 100 Mbps per stream and 6 degrees-of-freedom virtual reality headsets require (at least) a 500 meg circuit. VR has struggled to gain traction, however sub-10 millisecond latency (with a pathway to sub-5 millisecond), to eliminate dizziness, and cloud/edge computing (allowing the CPU/GPU in the network rather than the headset for a better form factor and far lighter weight) could drive adoption. We also note the emergence of light-field displays (holographic-like video). At a recent Cable-Tec conference we attended (LINK), Comcast noted its investment in Light Field Labs and notes applications with “immersive experiences on a multitude of fronts”, hi-resolution in-home cameras (seeing a dramatic increase in “megapixels per home”), and true holographic displays that are “closer than you think”. The provider noted that we could see “another change in display technology in the next 24-36 months”, while a single TV may drive speeds of 600-800 Mbps, leaving Telco and Satellites ISP’s far behind the curve. 12 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 13 Multi-Provider OTT Adoption Generally Increases with Speed COWEN EQUITY RESEARCH September 23, 2020 Figure 15 Speed Requirements for Video Use Cases of Tomorrow Bandwidth Requirements by Video Use Case, in Mbps This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 14 Speed Requirements for Video Use Cases of Today Bandwidth Requirements for Future Video Use Cases in 5+ Years, in Mbps 30.0 1,000 900 25.0 800 20.0 700 600 15.0 500 400 10.0 300 200 5.0 100 0.0 Ultra HD (4K), 1 Stream Google Stadia (HD) Full HD (1080p), 1 stream HD Video Teleconferencing Source: Company data, Cowen and Company 0 Light Field Display 6 DoF Video 8K Video, 1 stream Ultra HD (4K), 1 Stream Source: Company data, Cowen and Company In-Home IoT and Smart Home Management While the primary narrative for long-term Cable broadband growth has been focused on ARPU gains through speed uptake, one area that seems to get overlooked is Cable’s potential role in managing the future complexity of in-home Broadband IoT. Smart Home Management could prove yet another competitive advantage over Telco and Satellite ISPs. For example, the proliferation of in-home IoT is expected to continue unabated, as Cisco forecasts the number of devices per capita in North America will grow from ~8 today to 13 devices by 2021 (Altice has previously noted its Optimum customers were already at ~12 devices, on average as of 1Q19). As such, a new opportunity is emerging in the managing of in-home IoT as we consider smart TVs, smartphones, tablets, voice assistants, and in-home IoT smart plugs, appliances, locks, thermostats, cameras, and other security devices. As we consider this next “battle for the home” which allows for another valuable customer touch point (including data analytics, though privacy is a key sensitivity), vendors such as Amazon (Alexa/Ring), Google (Home), Apple (TV/phones), Arlo, and Nest Labs are in an ongoing turf war to own the ecosystem as customers want to avoid vendor lock-in yet prefer the “one stop shop” single interface for device management. The ISP’s, looking to avoid becoming the dumb pipe to Silicon Valley competition, have the obvious natural extension into this opportunity already owning the connectivity/WiFi, and are often the first call to customer care, regardless of the source of the issue (two-thirds of inbound care calls are WiFi related). To that point, the larger MSO’s have device management platforms such as Comcast’s Xfinity Home App (and the “Works with Xfinity” compatible devices), often included at no additional charge, used as a competitive advantage over telcos and overbuild competitors. Device compatibility/interoperability and device manufacturer registration (often required by the OEMs) still continue to be customer pain points in this battle for the home. Tangentially, we are seeing the emergence of “managed WiFi” for better in-home coverage solutions, which includes RF optimization/education, Wi-Fi extenders, parental controls, and a platform for said device management. While Comcast offers “xFi Pods” (WiFi extenders licensed from Plume), we could see a race for more personalized services. Both “smart/ managed WiFi" providers such as Plume and the MSO’s (which may or may not white label solutions from managed WiFi providers) are certainly seeing the demand for in-home managed WiFi, and with customers are willing to pay for it. COWEN.COM 13 COWEN EQUITY RESEARCH September 23, 2020 Cowen Survey: More Devices = Greater Speeds. Our survey works shows that in-home device adoption is directly correlated to adoption of higher speed plans. That is, those with 11+ devices take speeds of 377 Mbps, on average, those with 6-10 devices take speeds of 329 Mbps, and those with 1-5 devices take speeds of 234 Mbps, on average. Figure 16 Plan Data Speed (Wtd. Avg. Mbps) on In-Home Connected Devices 400 377 350 329 300 234 250 200 150 100 50 0 1-5 Devices 6-10 Devices 11+ Devices Source: Cowen and Company Starlink Applies for RDOF – Could Win Very Rural Areas (If Any), While Telco/Cable Should Take the Lion’s Share The Rural Digital Opportunity Fund (RDOF) is the federal funding program to provide broadband (and broadband upgrades from CAF II) to rural America. Similar to CAF II, RDOF will run a competitive bidding process, targeting up to $20.4B over 10 years. The program will be split into 2 phases: Phase I will allocate up to $16B, increasing the minimum speeds in rural neighborhoods to 25/3 Mbps (bidding on the census block level). Phase II will provide the remaining funding (at least $4.4B) for those areas not won in Phase I. The program looks to provide high speed broadband to ~6MM households. The Phase I auction is set to begin on October 29. See the Report & Order linked here. Phase I will be conducted in a reverse auction process. That is, “the lowest bid wins” as participants will bid on a buildout (per census block) with winners requesting the lowest dollars needed from the government. However, also factored into the equation are weight penalties assigned for those bidding in slower tiers and/or high latency as shown in the grid below: 14 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Providers can charge as much as $13-14 per month with healthy take rates (easily double digits); some operators charge $3.99-7.99 per month and admit to leaving money on the table. Lastly, Cable could find its way into what will be the highly lucrative home healthcare IoT market. WiFi 6 will be key in opening up healthcare and other opportunities as the latest generation will enable partitioning and traffic management, ideal for healthcare monitors, QoS requirements, mission critical use cases, and various classes of service. COWEN EQUITY RESEARCH September 23, 2020 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Figure 17 FCC RDOF Performance Tiers, Latency, and Weight Penalties Source: FCC On September 1, the FCC released those bidders that have applied for the Rural Digital Opportunity Fund Phase I Auction (RDOF). The full list of ~505 applications shows a diverse set of bidders including Telcos, Cable, Cellular carriers (U.S. Cellular), Fixed Wireless players, Fiber-based providers, and even Satellite. To that last point, the FCC may allow satellite providers into the fray as the agency had previously voiced concern over the fact that LEO providers such as SpaceX (while touting high speeds and low latency) do not yet have a commercial business in place. Feeling pressure from the other commissioners, and Starlink presenting to the FCC its ability to deliver at total latency below 50 ms (LINK and LINK), FCC Chairman Ajit Pai left the door open for LEO operators and “will conduct a careful, case-by-case review of applications to ensure that bidders will be able to meet required performance obligations.” (LINK) We’ve written in the past (LINK) that Telco, fixed wireless, and especially Cable operators (namely Charter) could take the large majority of homes given the strategic value and near-net opportunities (though AT&T and Comcast not participating). Meanwhile, Starlink could win, but we think only in very isolated cases. However Starlink’s application was deemed “incomplete” (along with hundreds of applications), as we suspect the deliberation over which latency tier Starlink may qualify for. Incomplete applications are due today, September 23, and the FCC will determine qualified bidders shortly thereafter. If Starlink does not fulfill the requirements, we do not necessarily believe this invalidates the latency claims, but more a decision around politics and/or caution by the FCC considering its $20B, 10-year investment. If Starlink is approved (on the higher latency tier, we believe easily met by leveraging its “very low earth orbit” satellites), the speed requirements could relegate Starlink to only a small handful of the 6MM homes. Other Starlink Use Cases: Wireless Backhaul, High Frequency Trading Key technologies that Starlink is currently testing are (1) the ability for laser optical communication and (2) satellite “cross-links”. That is, under the current system, the satellites only communicate using radio frequency communication. Developing laser communication could further augment throughput and latency. Perhaps more COWEN.COM 15 COWEN EQUITY RESEARCH September 23, 2020 For example, a satellite over Africa can serve residents directly, while a satellite over a dense-urban area such as NYC can cross-link to another satellite hundreds of miles away that has capacity. Furthermore, establishing laser uplinks, downlinks, and crosslinks could open up promising use cases such as wireless backhaul or high-frequency trading (the speed of light is ~40% slower through glass as the light essentially bounces within the fiber strand medium). For example, if a satellite orbits over Ohio, it can be used for residential broadband (simple uplink/downlink in Figure 1), but as that same satellite orbits over NYC (an area where residential broadband is not practical), it can be switched to a high frequency trading client, in which the satellite can laser cross-link in space to a satellite over Japan, which then downlinks to a Ground Station located at/near Nikkei market servers, achieving superior latency. 16 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. importantly, such laser communication could be used as “cross-links”. Currently, Starlink satellites can only support downlinks and uplinks (establishing the “bent pipe” to and from terminals and Ground Stations). However, cross-links allow for the ability for satellites in space to communicate with each other in space. Using laser cross-links, satellites can essentially route and offload traffic to other satellites (and/or to a satellite that has line-of-sight to a Ground Station as a satellite over the ocean serving a maritime customer may not have line-of-sight to a Ground Station). COWEN EQUITY RESEARCH September 23, 2020 BASIC INFORMATION Company Ticker Rating BASIC FINANCIALS Revenue VALUATION EBITDA FCF/Share EV/Sales EV/EBITDA GROWTH FCF Yield Sales Margin (C2020) EBITDA (C2020E) Dividend Net Debt/ Price Price Mkt Cap EV Gross EBITDA Capex/ D&A/ Yield EBITDA 9/22/2020 Target $ MM $ MM C2020E C2021E C2022E C2020E C2021E C2022E C2020E C2021E C2022E C2021E C2022E C2021E C2022E C2021E C2022E C2020E C2021E C2022E C2020E C2021E C2022E Margin Margin Sales Sales C2020E (LQA) $26.26 $633.05 $46.19 $30.13 $1,759.35 $35.00 $622.00 $44.00 $58.00 $1,884.00 15,578 150,039 212,797 17,551 10,594 36,515 224,235 308,827 23,981 11,680 10,074 47,856 101,172 14,439 1,328 10,587 50,796 107,075 15,745 1,422 10,950 53,318 112,434 15,131 1,488 4,334 18,127 29,930 3,073 659 4,615 19,769 30,981 3,072 727 4,879 21,312 33,478 2,771 781 $2.90 $28.16 $3.16 $4.83 $43.68 $3.39 $36.38 $2.55 $4.24 $45.48 $1.51 $44.50 $0.68 ($0.95) $52.61 3.4x 4.4x 2.9x 1.5x 8.2x 4.1x 3.3x 4.2x 2.7x 1.6x 7.9x 3.9x 7.9x 11.3x 10.0x 7.8x 16.1x 10.6x 7.5x 10.5x 9.2x 8.7x 14.9x 10.2x 12.9% 5.7% 5.5% 14.1% 2.6% 8.2% 5.7% 7.0% 1.5% NM 3.0% 4.3% 3.2% 4.6% -7.1% 12.7% 13.7% 5.4% 5.1% 6.1% 5.8% 9.0% 7.1% 6.6% 3.4% 5.0% 5.0% -3.9% 4.6% 2.8% 1.6% 7.5% -12.6% 21.8% 15.9% 6.8% 6.5% 9.1% 3.5% 0.0% 10.3% 5.9% 5.7% 7.8% 8.1% -9.8% 7.5% 3.8% 65.8% 55.8% NM 44.3% 67.4% 58.3% 43.0% 37.9% 29.6% 21.3% 49.6% 36.3% 12.3% 15.2% 9.0% 3.0% 22.1% 12.3% 21.7% 20.6% 12.9% 4.3% 20.1% 15.9% NM NM 2.0% NM 0.5% 1.3% 3.4x 3.3x 10.0x 9.2x 5.7% 4.4% 4.6% 6.1% 4.6% 7.5% 6.5% 7.5% 60.8% 37.9% 12.3% 20.1% 2.1x 2.7x 1.0x 2.8x 2.1x 2.1x 2.7x 1.0x 2.8x 2.1x 6.1x 9.5x 4.1x 7.6x 6.8x 6.0x 8.8x 4.1x 7.4x 6.6x 11.6% 3.9% NM 7.8% 7.8% 12.0% 6.2% 0.6% 8.7% 6.9% -6.7% 47.7% -1.1% -4.5% 8.8% -1.3% 14.2% 2.5% 2.6% 4.5% 1.2% 1.5% 1.0% 1.7% 1.4% -6.7% 50.4% 2.1% -1.7% 11.0% 2.6% 11.7% 1.3% 3.2% 4.7% 1.7% 9.1% -0.4% 2.5% 3.2% NM 60.0% 80.4% 59.6% 66.7% 32.7% 28.9% 23.5% 36.8% 30.5% 9.4% 16.3% 22.2% 14.2% 15.5% 2.4x 2.4x 6.9x 6.7x 7.8% 7.4% -2.8% 2.5% 1.4% 0.2% 2.9% 2.1% 60.0% 30.8% 6.1x 7.3x 2.1x 2.0x 4.4x 5.7x 6.8x 2.1x 2.0x 4.2x 15.7x 9.1x 5.1x 9.1x 9.7x 14.4x 8.6x 5.2x 8.7x 9.2x 3.1% 5.1% 37.2% 44.6% 22.5% 3.5% 6.1% 39.6% 49.8% 24.7% 4.3% -0.2% -3.4% -2.8% -0.5% 4.4% 3.9% -3.0% 0.9% 1.5% 6.0% 5.9% -2.8% 2.0% 2.8% 8.9% 0.2% 0.2% -17.7% -2.1% 6.3% 7.3% -2.3% 5.1% 4.1% 8.6% 6.1% -2.5% 4.0% 4.1% 62.1% NM 58.1% 43.8% 54.7% 4.1x 3.9x 9.1x 8.7x 21.1% 22.9% -1.5% 2.4% 3.9% 0.2% 5.7% 5.1% 1.8x 2.1x 0.8x 1.6x 1.8x 2.2x 0.8x 1.6x 6.7x 5.0x 3.3x 5.0x 6.7x 5.1x 3.2x 5.0x NM 29.7% NM 29.7% NM 8.0% 3.1% 5.5% -0.8% -3.4% 1.1% -1.0% 0.5% -2.0% 2.8% 0.4% -0.9% -1.7% 1.6% -0.3% -0.2% -3.6% 9.4% 1.9% 0.9% -0.6% -2.4% -0.7% Median 1.8x 1.8x 5.0x 5.1x 29.7% 5.5% -0.8% 0.5% -0.9% -0.2% All Cable/Telco Services 6.2x 3.2x 5.6x 3.1x 12.8x 8.4x 11.6x 8.1x 11.2% 14.1% 8.3% 11.0% 9.1% 3.6% 8.1% 3.7% 6.3% 1.9% 2.7x 2.4x 13.3x 11.9x NA NA -1.3% 4.7% 6.8% This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Comparative Analysis – Telecom Services Performance 2019 2020 4.7x 4.3x 2.9x 3.2x 1.7x 3.4x 65.5% 70.2% 32.1% 42.6% 81.5% 58.4% -4.0% 30.5% 2.7% -15.1% 18.2% 6.5% 1.3% 3.2x 65.5% 2.7% 17.1% 16.6% 17.9% 13.1% 16.1% 7.3% NM NM 4.2% 5.7% 2.6x 2.9x 1.4x 2.5x 2.3x 36.9% 23.3% -30.3% 9.2% 9.8% -27.0% 43.3% -14.8% -2.6% -0.3% 15.2% 16.8% 5.7% 2.5x 16.2% -8.7% 38.0% 77.1% 40.6% 21.5% 44.3% 8.7% 25.1% 15.3% 5.2% 13.6% 14.0% 30.7% 25.4% 16.4% 21.6% 4.5% 6.8% NM NM 5.7% 3.2x 6.6x 4.1x 8.8x 5.7x 45.6% NM -60.7% -52.0% -22.4% -6.0% 19.4% 48.2% -53.0% 2.1% 58.1% 39.3% 12.0% 20.9% 5.7% 5.4x -52.0% 6.7% -1.1% -2.2% 2.4% -0.3% NM 56.9% NM 56.9% 26.6% 42.2% 24.7% 31.1% 13.9% 18.0% 23.0% 18.3% NA 22.6% 18.1% 20.4% NM 9.7% NM 9.7% 5.0x 3.7x 1.6x 3.4x 34.6% -12.8% -21.9% 5.8% 43.7% -22.3% -24.0% 5.7% -0.6% -1.1% 56.9% 26.6% 18.0% 20.4% 9.7% 3.7x 5.3% 1.6% 11.4% 4.7% 10.6% 3.9% 13.7% 3.0% 59.9% 59.5% 38.3% 35.9% 34.0% 14.5% 20.9% 18.1% 3.9% 5.0% 3.5x 3.7x 32.6% 17.6% 24.4% 2.3% -8.4% 9.3% 10.9% 34.7% 13.3% 1.8% NA 28.9% 2.6% YTD Cable & Satellite Altice USA Charter Comcast Dish Cable One Mean ATUS ** CHTR ** CMCSA ** DISH ** CABO ** 1 1 2 1 2 Median Wireless AT&T T-Mobile US Cellular Verizon Mean T * TMUS * USM VZ * 2 1 NR 1 $28.51 $112.39 $30.85 $59.82 $34.00 $133.00 NA $62.00 200,607 139,728 2,651 247,655 347,993 205,079 3,908 362,508 169,006 66,482 3,972 125,952 166,753 75,949 4,072 129,170 168,727 77,127 4,113 131,351 55,293 19,231 935 46,397 56,722 21,475 948 47,874 57,670 23,420 944 49,076 $3.23 $2.25 ($0.22) $5.48 $3.30 $4.42 ($1.00) $4.65 $3.41 $6.93 $0.20 $5.18 Median Competitive Cogent Uniti Group Consolidated Comm. GTT Mean CCOI * UNIT ** CNSL ** GTT * 2 1 2 2 $61.84 $9.80 $5.75 $5.34 $77.00 $13.00 $7.00 $11.00 2,916 2,562 409 313 3,606 7,958 2,593 3,454 570 1,056 1,292 1,680 595 1,097 1,253 1,694 631 1,162 1,218 1,728 217 814 525 362 230 874 513 380 250 927 500 395 $1.56 $1.12 $2.28 $1.40 $1.94 $0.50 $2.14 $2.38 $2.16 $0.60 $2.28 $2.66 Median LECs Cincinnati Bell Lumen TDS Mean CBB LUMN ** TDS NR 2 NR $15.05 $10.26 $19.33 $11.00 763 11,107 2,209 2,723 43,643 4,169 1,524 20,735 5,230 1,532 20,312 5,379 1,517 19,976 5,463 405 8,741 1,290 409 8,693 1,259 404 8,503 1,290 N/A $2.85 ($1.21) N/A $3.05 ($0.47) N/A $0.82 $0.59 Market Indices S&P 500 .SPX 3,316 1,360 1,423 1,521 262 286 317 NA NA NA 1 = Outperform, 2 = Market Perform, 3 = Underperform, NR = Not Rated * Covered by Colby Synesael, **Covered by Gregory Williams Source: Company data, Thomson Reuters, Cowen and Company estimates COWEN.COM 17 COWEN EQUITY RESEARCH September 23, 2020 Rating Price* Price Target Ticker Rating Price* Price Target ATUS CABO CMCSA TMUS Outperform Market Perform Market Perform Outperform $26.26 $1,759.35 46.19 $112.39 $35.00 $1,884.00 $44.00 $133.00 T CHTR DISH VZ Market Perform Outperform Outperform Outperform $28.51 $633.05 30.13 $59.82 $34.00 $622.00 $58.00 $62.00 *As of 09/22/2020 VALUATION METHODOLOGY AND RISKS Valuation Methodology Telecom Services: Our valuation methodology consists of an absolute and relative value approach. We arrive at a fair value utilizing a five-year discounted cash flow (DCF) and when appropriate a segmented sum-of-parts (SOP) analysis. Our relative value approach takes into account EV/ EBITDA, P/FCF, and P/E and, when applicable, P/AFFO and dividend yield. Cable & Satellite Services: Our valuation methodology consists of an absolute and relative value approach. We arrive at a fair value utilizing a five-year discounted cash flow (DCF) and when appropriate a segmented sum-of-parts (SOP) analysis. Our relative value approach takes into account EV/ EBITDA, P/FCF, FCF Yield, P/E and, when applicable, P/AFFO and dividend yield. Investment Risks Telecom Services: Risks Include: (1) many companies within Telecom Services are highly regulated where a change in rules could lead to unfavorable conditions; (2) rapidly changing/disruptive technology, new product/service offerings, and evolving industry/technology standards could have an impact on demand and/or pricing; and (3) deterioration in the macro environment both domestically and internationally could lead to a reduction in demand and a consequent impact on valuation multiples. Cable & Satellite Services: Risks Include: (1) rapidly changing/disruptive technology (specifically the threat of OTT video on traditional video services), new product/service offerings, and evolving industry/ technology standards could have an impact on demand and/or pricing; (2) many companies within Cable & Satellite Services are highly regulated where a change in rules could lead to unfavorable conditions; and (3) deterioration in the macro environment could lead to a reduction in demand and a consequent impact on valuation multiples. 18 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. Ticker COWEN EQUITY RESEARCH September 23, 2020 Analyst Certification Each author of this research report hereby certifies that (i) the views expressed in the research report accurately reflect his or her personal views about any and all of the subject securities or issuers, and (ii) no part of his or her compensation was, is, or will be related, directly or indirectly, to the specific recommendations or views expressed in this report. Important Disclosures This report constitutes a compendium report (covers six or more subject companies). As such, Cowen and Company, LLC chooses to provide specific disclosures for the companies mentioned by reference. To access current disclosures for the all companies in this report, clients should refer to https://cowen.bluematrix.com/sellside/Disclosures.action or contact your Cowen and Company, LLC representative for additional information. Cowen and Company, LLC compensates research analysts for activities and services intended to benefit the firm's investor clients. Individual compensation determinations for research analysts, including the author(s) of this report, are based on a variety of factors, including the overall profitability of the firm and the total revenue derived from all sources, including revenues from investment banking, sales and trading or principal trading revenues. Cowen and Company, LLC does not compensate research analysts based on specific investment banking transactions or specific sales and trading or principal trading revenues. Disclaimer Our research reports are simultaneously available to all clients are on our client website. Research reports are for our clients only. Not all research reports are disseminated, e-mailed or made available to third-party aggregators. Cowen and Company, LLC is not responsible for the redistribution of research by third party aggregators. Selected research reports are available in printed form in addition to an electronic form. All published research reports can be obtained on the firm’s client website, https://cowenlibrary.bluematrix.com/client/ library.jsp. The information, opinions, estimates and forecasts are as of the date of this report and subject to change without prior notification. We seek to update our research as appropriate, but various regulations may prevent us from doing so. Research reports are published at irregular intervals as appropriate in the analyst’s judgement. Further information on subject securities may be obtained from our offices. This research report is published solely for information purposes, and is not to be construed as an offer to sell or the solicitation of an offer to buy any security in any state where such an offer or solicitation would be illegal. Other than disclosures relating to Cowen and Company, LLC, the information herein is based on sources we believe to be reliable but is not guaranteed by us and does not purport to be a complete statement or summary of the available data. Any opinions expressed herein are statements of our judgment on this date and are subject to change without notice. The opinions and recommendations herein do not take into account individual client circumstances, objectives or needs and are not intended as recommendations of investment strategy. The recipients of this report must make their own independent decisions regarding any securities subject to this research report. In some cases, securities and other financial instruments may be difficult to value or sell and reliable information about the value or risks related to the security or financial instrument may be difficult to obtain. To the extent that this report discusses any legal proceedings or issues, it has not been prepared to express or intended to express any legal conclusion, opinion or advice. Our salespeople, traders and other professionals may provide oral or written market commentary or trading strategies to our clients that reflect opinions that are contrary to the opinions expressed in our research. Our principal trading area and investing businesses may make investment decisions that are inconsistent with recommendations or views expressed in our research. Cowen and Company, LLC maintains physical, electronic and procedural information barriers to address the flow of information between and among departments within Cowen and Company, LLC in order to prevent and avoid conflicts of interest with respect to analyst recommendations. For important disclosures regarding the companies that are the subject of this research report, please contact Compliance Department, Cowen and Company, LLC, 599 Lexington Avenue, 20th Floor, New York, NY 10022. In addition, the same important disclosures, with the exception of the valuation methods and risks, are available on the Firm's disclosure website at https://cowen.bluematrix.com/sellside/Disclosures.action. Equity Research Price Targets: Cowen and Company, LLC assigns price targets on all companies covered in equity research unless noted otherwise. The equity research price target for an issuer's stock represents the value that the analyst reasonably expects the stock to reach over a performance period of twelve months. Any price targets in equity securities in this report should be considered in the context of all prior published Cowen and Company, LLC equity research reports (including the disclosures in any such equity report or on the Firm's disclosure website), which may or may not include equity research price targets, as well as developments relating to the issuer, its industry and the financial markets. For equity research price target valuation methodology and risks associated with the achievement of any given equity research price target, please see the analyst's equity research report publishing such targets. Cowen Cross-Asset Research: Due to the nature of the fixed income market, the issuers or debt securities of the issuers discussed in “Cowen Cross-Asset Research” reports do not assign ratings and price targets and may not be continuously followed. Accordingly, investors must regard such branded reports as providing stand-alone analysis and reflecting the analyst’s opinion as of the date of the report and should not expect continuing analysis or additional reports relating to such issuers or debt securities of the issuers. From time to time “Cowen Cross-Asset Research” analysts provide investment recommendations on securities that are the subject of this report. These recommendations are intended only as of the time and date of publication and only within the parameters specified in each individual report. “Cowen Cross-Asset Research” investment recommendations are made strictly on a case-by-case basis, and no recommendation is provided as part of an overarching rating system or other set of consistently applied benchmarks. The views expressed in "Cross-Asset Research" report may differ from the views offered in the firm’s equity research reports prepared for our clients. Notice to UK Investors: This publication is produced by Cowen and Company, LLC which is regulated in the United States by FINRA. It is to be communicated only to persons of a kind described in Articles 19 and 49 of the Financial Services and Markets Act 2000 (Financial Promotion) Order 2005. It must not be further transmitted to any other person without our consent. Notice to European Union Investors: Individuals producing recommendations are required to obtain certain licenses by the Financial Regulatory Authority (FINRA). You can review the author’s current licensing status and history, employment history and, if any, reported regulatory, customer dispute, criminal and other matters via “Brokercheck by FINRA” at http:// brokercheck.finra.org/. An individual’s licensing status with FINRA should not be construed as an endorsement by FINRA. General biographical information is also available for each Research Analyst at www.cowen.com. Additionally, the complete preceding 12-month recommendations history related to recommendation in this research report is available at https://cowen.bluematrix.com/sellside/ Disclosures.action The recommendation contained in this report was produced at September 22, 2020, 23:05 ET. and disseminated at September 23, 2020, 05:30 ET. Copyright, User Agreement and other general information related to this report © 2020 Cowen and Company, LLC. All rights reserved. Member NYSE, FINRA and SIPC. This research report is prepared for the exclusive use of Cowen clients and may not be reproduced, displayed, modified, distributed, transmitted or disclosed, in whole or in part, or in any form or manner, to others outside your organization without the express prior written consent of Cowen. Cowen research reports are distributed simultaneously to all clients eligible to receive such research reports. Any unauthorized use or disclosure is COWEN.COM 19 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. ADDENDUM COWEN EQUITY RESEARCH September 23, 2020 Cowen and Company, LLC. New York 646 562 1010 Boston 617 946 3700 San Francisco 415 646 7200 Chicago 312 577 2240 Cleveland 440 331 3531 Atlanta 866 544 7009 Stamford 646 616 3000 Washington, D.C. 202 868 5300 London (affiliate) 44 207 071 7500 COWEN AND COMPANY EQUITY RESEARCH RATING DEFINITIONS Outperform (1): The stock is expected to achieve a total positive return of at least 15% over the next 12 months Market Perform (2): The stock is expected to have a total return that falls between the parameters of an Outperform and Underperform over the next 12 months Underperform (3): Stock is expected to achieve a total negative return of at least 10% over the next 12 months Assumption: The expected total return calculation includes anticipated dividend yield Cowen and Company Equity Research Rating Distribution Distribution of Ratings/Investment Banking Services (IB) as of 06/30/20 Rating Count Ratings Distribution Count IB Services/Past 12 Months Buy (a) 500 63.61% 120 24.00% Hold (b) 277 35.24% 15 5.42% 9 1.15% 0 0.00% Sell (c) (a) Corresponds to "Outperform" rated stocks as defined in Cowen and Company, LLC's equity research rating definitions. (b) Corresponds to "Market Perform" as defined in Cowen and Company, LLC's equity research ratings definitions. (c) Corresponds to "Underperform" as defined in Cowen and Company, LLC's equity research ratings definitions. Cowen and Company Equity Research Rating Distribution Table does not include any company for which the equity research rating is currently suspended or any debt security followed by Cowen Credit Research and Trading. Note: "Buy", "Hold" and "Sell" are not terms that Cowen and Company, LLC uses in its ratings system and should not be construed as investment options. Rather, these ratings terms are used illustratively to comply with FINRA regulation. 20 COWEN.COM This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. prohibited. Receipt and/or review of this research constitutes your agreement not to reproduce, display, modify, distribute, transmit, or disclose to others outside your organization. All Cowen trademarks displayed in this report are owned by Cowen and may not be used without its prior written consent. COWEN EQUITY RESEARCH September 23, 2020 Analyst Profiles Gregory Williams, CFA Colby Synesael Daniel Parr New York New York New York 646 562 1367 646 562 1355 646 562 1339 gregory.williams@cowen.com colby.synesael@cowen.com daniel.parr@cowen.com Greg Williams is a research analyst covering cable and telco services. He has 14 years of relevant experience at AT&T, JPMorgan and Sidoti. Colby Synesael is a senior analyst covering comm. infra. and telecom services. He joined Cowen in 2010 and has been on Wall Street since 2001. Michael Elias Joseph Kavaloski New York New York 646 562 1358 646 562 1353 michael.elias@cowen.com joseph.kavaloski@cowen.com Michael Elias is a member of the telecom services team. He received a BS in engineering management systems from Columbia University. Daniel Parr is a member of the telecom services team. He joined Cowen in 2019 and is a graduate of the University of Delaware. Joseph Kavaloski is a member of the telecom services team. He joined Cowen in 2020 and is a graduate of Fordham University. Reaching Cowen Main U.S. Locations New York Boston Cleveland San Francisco 599 Lexington Avenue New York, NY 10022 646 562 1010 800 221 5616 Two International Place Boston, MA 02110 617 946 3700 800 343 7068 20006 Detroit Road Suite 100 Rocky River, OH 44116 440 331 3531 One Maritime Plaza, 9th Floor San Francisco, CA 94111 415 646 7200 800 858 9316 Atlanta Chicago Stamford Washington, D.C. 3424 Peachtree Road NE Suite 2200 Atlanta, GA 30326 866 544 7009 181 West Madison Street Suite 3135 Chicago, IL 60602 312 577 2240 262 Harbor Drive Stamford, CT 06902 646 616 3000 2900 K Street, NW Suite 520 Washington, DC 20007 202 868 5300 International Location Cowen International Limited London 1 Snowden Street - 11th Floor London EC2A 2DQ United Kingdom 44 20 7071 7500 COWENRESEARCH COWEN COWEN.COM COWEN INC. 21 This report is intended for replaceme@bluematrix.com. Unauthorized redistribution of this report is prohibited. POINTS OF CONTACT