[AMT – American Tower; CCI – Crown Castle] Legacy advantages and incremental returns

In the US, tower companies have enjoyed a 20-year growth tailwind that shows no signs of fatigue. The population of wireless subscribers has grown at a high single digit rate, from fewer than 130mn in 2001 to 440mn in 2018. In turn, network standards have evolved – from 2G (voice, SMS; 2000-2006) to 3G (mobile web browsing; 2005-2011) to 4G (mobile video; 2011 –) to eventually 5G (IoT, automation, immersive experiences) – accommodating 30%-40% annual growth in data traffic, fueling steady demand for network infrastructure (capex from US wireless carriers has run between $20bn-$30bn in each of the last dozen years)

Essential to meeting the irrepressible flood of data are ~150k of these:

Source: steelintheair

That’s a cell tower. These used to be owned and operated by wireless carriers but following two decades of consolidation are today managed by third-parties like American Tower and Crown Castle who, by consolidating tenancies from multiple carriers, can rent out more space on the tower, and do so faster and cheaper than any carrier could on its own[1]. A cell tower receives and propagates radio waves that allow one cell phone to connect to another. It’s basically a vertical structure on a small plot of land with antennas hanging off it. Those antennas are hooked up to coax cables or fiber, threaded up and down the tower, that carry radio signals to and from a base station, which connects to the fiber backhaul of the big telcos. Around the tower, you might find a generator for emergency backup power and a shelter that houses radio and networking equipment.

Sometimes cell towers are camouflaged as palm trees for aesthetic reasons 😊

A tower operator owns the tower and leases

the underlying land while wireless carriers own and operate all the equipment,

cable, and shelters. In the US, carriers

lease space from tower operators under non-cancellable contracts with an initial

term of 5 to 10 years, followed by 5-year renewal periods, embedded with 3%-4%

price escalators. The more stuff a

wireless carrier puts on the tower and the heavier the stuff, the more it

pays. To get a tower up and running, an operator

might invest ~$275k upfront for construction and then incur ~$12k/year in

ongoing costs of ground rent, insurance, real estate taxes, and utilities, plus

~$1.5k annually on upkeep, repairs, and other basic maintenance expenses. A single tenant might pay ~$20k/year for

rent, producing gross profits after MCX of ~$6k, a piddling 2% return on

investment. 

But a typical tower can accommodate ~4

tenants and the costs of operating it are almost entirely fixed. 95%+ of incremental revenue drops to gross

profits, so the economics become compelling as you bring on more tenants. At 3 tenants, each subsequent tenant paying ~$30k

(the initial anchor tenant pays a lower rate), that’s $80k in revenue and maybe

$63k in gross profits ($6k of gross profits from tenant 1 + $60k of incremental

revenue at 95% contribution margins), translating to an ROI of 25%. 

Easy, right? You’ll have to maneuver through some tricky zoning regulations, sure, but to build a tower, all you really need is a 2k sf plot of land and a few hundred thousand in capital. Given the industry’s low entry barriers, the lofty returns that I illustrated above may seem unrealistic. And in a sense, they are. But it depends on whether we’re talking about returns on existing capital or incremental capital and whether we’re talking about the returns generated by towers or the returns realized by capital providers.

As is true of Transdigm’s aircraft parts, the cost of renting tower space is tiny compared to the mission critical, high value service it enables. The average monthly cell phone bill in the US is around 80 bucks, which when multiplied by ~1,200 connections per cell site (the area covered by a tower) translates to $96k/month, or $1.2mn/year. The $20k starting lease expense that a wireless carrier pays is less than 2% of the high margin revenue it generates from an antenna array affixed to a tower (with price escalators and equipment adds, maybe the actual expense is more like ~4% of wireless revenue service revenue). A carrier could relocate but doing so might negatively impact the performance of adjacent cell sites and disrupt service quality, which would require network retuning and invite other engineering aches. When you consider the low cost/value proposition and the relatively high switching costs, it’s perhaps no wonder that across the three publicly traded tower companies – American Tower, Crown Castle, and SBA Communications – annual churn has historically run less ~1%-1.5%. Churn from the big 4 US carries is more like ~0.3%.

There has recently been some commotion

about Tillman Infrastructure – a private developer with $1bn of funding

commitments that has expanded into 36 states since it first began constructing cell

sites in the US in 2017 – poaching carriers with generous lease terms that

supposedly impute returns that incumbents deem inadequate. But Tillman’s portfolio of several hundred

towers is a pimple compared to the 95k that the incumbents manage. And given how hard it is to locate a new

tower proximate to an existing one, those towers will likely be in markets where

AMT and CCI don’t have a strong presence. 

Tillman, like the hundred+ other private developers in the market, is

unlikely to have any discernable impact on incumbents’ existing tower returns. The switching costs are too high, plus Tillman

doesn’t have any core cost advantages that would allow it to offer a better

value proposition than the incumbents at comparable returns.

AMT, CCI, and SBAC – the three largest tower companies in the US, together accounting for ~65% of our wireless towers[2] – have for many years been wary of the meager returns on incremental construction. They’ve instead focused on squeezing more rents from the installed base, sustaining mid/high-single digit organic growth through a combination of contractual price escalators (~3%) and modifications to lease agreements, called amendments. Amendments allow carrier tenants to add more antennas or bigger antennas and have been responsible for something like 80% of the big 3’s new volumes over the last 4 or 5 years (the remaining 20% has come from new co-locations). Sometimes these amendments are incremental (more aggregate equipment to boost a cell site’s capacity) and other times they are cannibalistic (replacing prior generation antennas with newer, usually heavier, ones). 

Amendments come in waves whose undulations

are dictated by the needs of different network standards. While a network standard is in its early days,

carriers are preoccupied with maximizing coverage (the percent of the population

with access to the network), so they start with small, lower priced amendments

and graduate to larger, higher priced ones as the focus shifts to capacity (to

handle heavy network utilization), until eventually, the standard and its

surrounding ecosystem matures to such an extent that new cell sites must be built

to handle the demand. The length of the

amendment phase and the cell splitting phase vary depending on the pace of

adoption. For instance, the transition

from amendments to cell splitting occurred relatively quickly for 3G because

deployments had already happened in Europe and Asia and so the supporting

handset devices were already out there to absorb the coverage. 

The same was not true of 4G. In 2013, 2 years after Verizon introduced 4G LTE in their 700 MHz band, just 10% of devices were 4G capable. Today, we’re still at “just” 60% of devices, so the long tail of 4G amendments continues to stretch into the incipient stages of 5G adoption. Over the last year, as carriers have deployed larger equipment for lower, better propagating spectrum in the 600 MHz –700MHz bands (most 3G networks operate in bands 800 MHz and higher), monthly amendment prices have in turn escalated from ~$500 to ~$800[3]. And if Sprint’s recent massive MIMO[4] rollout is any indication, growing 5G adoption could bring another significant and overlapping amendment cycle as carriers hoist wider and ~2x heavier MIMO antennas to exploit mid-band spectrum (though it is a subject of debate whether towers will be used to the same degree for millimeter wave frequencies as it is whether millimeter wave frequencies will be exploited to any material degree in the next 5 years).

In short, given switching costs and a low

cost/value proposition, incumbent tower operators will continue to realize excellent

returns on existing tower assets through upgrades and capital light

augmentations. 

But discretionary cash flows can’t be reinvested in US organic builds at adequate returns. Instead, to expand their footprints and drive growth, American Tower and Crown Castle have resorted to acquisitions. 

In the US, they have historically acquired smaller developers and peeled away tower assets from wireless carriers. ~60% of American Tower’s current US installed base came from three big acquisitions: SpectraSite (2005, 7.8k towers), Global Tower Partners (2013, 5.4k), and Verizon Wireless (2015, 11.4k). Nearly 70% of Crown Castle’s 40k US towers came through transactions with Global Signal (2007, 10.7k towers), AT&T (2013, 7.1k), and T-Mobile (2012, 9.5k).

These acquisitions were done at lofty

premiums to tangible asset value, meaning that while AMT has generated 20%+

pre-tax returns on gross tangible assets, it has delivered more like 11%-12% on

capital when you include goodwill and intangibles (which, given the

mid/high-teens returns earned overseas, means we’re probably talking about

~9-ish% returns in the US). Paying a

developer 2x what it costs to build a tower gets you a starting yield of ~9%,

assuming the tower is fully leased up with 2-3 tenants. If you finance 50%-60% of that purchase with

debt, as American Tower does, after-tax returns on equity land somewhere in the

high-teens. That return gradually accretes

through price escalators and amendments. 

In its acquisition of Verizon Wireless’ 11k towers in 2015, AMT paid ~$500k/tower[5],

implying a starting yield of 4%-5%. But

the acquired towers averaged just 1.4 tenants/tower generating $2.5k/month in

revenue compared to 2.5 tenants and $6k of monthly revenue for American’s

existing US portfolio, implying capacity for organic growth to take unlevered

returns to ~9%, especially as AT&T and T-Mobile looked to replicate the strength

of Verizon’s coverage. There are other

idiosyncrasies like that, but taken as a whole, I think AMT’s US tower assets

are earning unlevered 8%-9% and levered high-teens.

Nominal returns on capital appear somewhat stronger overseas, where American Tower has been especially aggressive. The company began its international expansion almost immediately after being spun off of American Radio System (as part of CBS’ acquisition of American Radio) in 1998[6], starting in Mexico (1999) before entering Brazil (2000), India (2007), Latin America (2010), and Africa (2011). Its pace of international expansion has dramatically intensified over the last 8 years: 

Today, over ¾ of the company’s towers

are outside the US. In its largest international

markets – India, Brazil, Mexico, and parts of Africa – American Tower is the largest

independent tower player, having scaled to become critical infrastructure for

local wireless carriers. And yet,

despite accounting for just ¼ of AMT’s towers, the US still makes up over half

of revenue and 60% of gross profits due to the lower absolute dollar sums

invested and returned overseas – for instance, the rents in India are

~$600/month (vs. $2k in the US), but the cost to build a tower is just $50k

(vs. $250k) and annual maintenance capex is more like $500 (vs. $2k),

reflecting relatively lower labor and construction costs. Rent escalators are typically tied to

inflation. 

If the Viom acquisition is any guide,

it looks like AMT is targeting unlevered returns of ~10%-13% in India (its

largest market by number of towers), a few points higher than what it generates

in the US. Same in Latin America,

~low-teens. In certain countries within Africa,

where risks are more abundant, the IRR threshold is more like high-teens/low-20s. AMT has been much more willing to grow

organically overseas, adding ~2k towers this year in India, another 600-700 in

each of Latin American and Africa, though the preponderance of international growth

continues to come from acquisitions. 

Organic growth should continue to run ~10%-15% internationally (vs. hsd in the US), as tenancy per tower is lower (1.5 vs. 2.7 in the US) and wireless standards are 5 to 10 years behind[7]. Many countries have just started migrating to 4G, which should translate into the same amendment cycle-driven boosts that have fueled returns on US assets over the last 5 years. The transition to 4G has prompted consolidation in India, AMT’s largest territory, home to 40% of its 170k towers, where the number of major wireless players has consolidated down from 10 to 15 in the mid-2000s to around 4 or 5 today, the winnowing hastened by Reliance Jio’s introduction of a 4G product in 2016[8]. 

AMT tiptoed into India, building a few hundred towers in 2007. But the company got super aggressive shortly thereafter, its expansion efforts mirroring the carrier consolidation that began around 2010. It purchased Essar Group’s portfolio of 4,600 towers before then acquiring a majority stake in Viom Networks (a JV between Tata Teleservices and various other financial parties) in 2015, adding over 40k towers to its then portfolio of 14k[9], in the process becoming the largest independent tower company in India. Several years later, following Vodafone India’s announced merger with Idea Cellular, AMT scooped up tower assets from Vodafone India. While most of the towers are affiliated with the carriers in some way – including 42% Airtel-owned Indus Towers, the largest operator in India with 117k sites – it seems that the wireless value chain in India is evolving similarly to the US, with carriers distancing themselves from tower assets to specialize in higher value services. 

AMT sees emerging markets as analogous to the US market 5 to 10 years ago and is replicating its tower acquisition competency to capitalize on the same evolution of network standards that has propelled wireless growth in the US for so many years. Crown Castle is making a much different bet, content to stay within US borders and deploying large sums of capital into fiber/small cell technology.  

The idea here is that the

proliferating amount data traffic from smartphones, tablets, wearables, and the

pending flood of other smart devices is overwhelming the wireless signals of cell

towers (aka “macro” cells), mandating a complementary solution like Wi-Fi or small

cells, especially in dense urban areas but maybe eventually in thinly populated

suburbs too. You can think of

traditional macro towers as providing an overarching blanket of coverage,

underneath which small cells boost capacity so that you not only have access to

a carrier’s network (coverage) but can also reliably make phone calls or stream

video (capacity) whenever you want. 6 or

7 years ago there was some trepidation about Wi-Fi supplanting macro cells and small

cells have conjured similar concerns. 

But, at least for now, small cells and Wi-Fi are densification solutions

that supplement macro towers in high-traffic urban areas – that Crown Castle is

focusing its small cell effort exclusively in metro markets suggests as much –

while nearly all of AMT’s towers are situated in suburban and rural regions

where over 84% of the US population resides (you don’t see 200 foot cell towers

sprinkled throughout San Francisco and Manhattan).    

Depending on where you live, you may have seen small cells attached to streetlights and utility poles. Like macro cells, they communicate through radio waves and accommodate multiple wireless carriers. But whereas a tower build might require a simple conversation with a homeowners’ association, the approval process for small cells is far more intricate. Because small cells are found in public rights of way and their installation requires busting concrete to access fiber, Crown Castle has had to machete through lengthy permitting discussions with city officials and appease the safety and aesthetic concerns of residents, creating a significant bottleneck to deployment. 

Nonetheless, Crown has laid a huge bet, having spent $13bn+ (5 years of EBITDA) over the last 4-5 years acquiring companies with rights to 10s of thousands of route miles of fiber in Boston, New York, Philadelphia, Chicago, Los Angeles, San Diego, the Bay Area, and metro markets in Florida and Texas. Management thinks that its fiber/small cell business could be as big as towers one day.

[red areas represent CCI’s fiber

footprint]

American Tower has dismissed US fiber investing as a competitive, capital intensive enterprise with lower returns than towers. AMT’s management is more receptive to fiber that can be dual purposed for towers, as is the case in many emerging markets, where only 10%-15% of towers are connected. But that opportunity is not available in the US towers, where ~90% of towers are already fibered.

Crown does not share these concerns

and have been has been connecting the site locations of corporations,

universities, financial institutions, and school districts as a kind of revenue

placeholder until small cell adoption eventually takes hold and justifies the

company’s mountainous deployment of capital. The trajectory of returns will differ from

towers – small cells have higher initial yields but don’t enjoy the 95%+

contribution margins that towers do – but Crown’s management thinks that once

small cells are leased to a 3-tenant tower equivalent[10], they will generate the

same mid/high-teens yields as its towers. 

There are lots of companies installing fiber, though no one of note (outside of the carriers themselves) is as myopically focused on the small cell use case as Crown – Level 3 has eschewed small cells to focus mostly on enterprise customers while Zayo, who spent $7bn rolling up dozens of companies since 2007 before itself being the target of acquisition recently, is going after a wide range of use cases (enterprise, edge data centers, autonomous vehicles, 3G/4G, and some small cells). But where’s the moat? Management claims, “we see the ownership of deep, dense fiber in top metro markets as a competitive advantage in facilitating small cell deployments in a cost-effective and timely manner”…and that’s fine once the assets are in place. But I don’t understand Crown’s advantage in acquiring that fiber in the first place. Infrastructure and private equity funds have been pouring money into fiber assets, boosting valuations by 30% over the last year[11], inviting persistent fears that the market is too crowded. 

A few years ago, Crown’s management

remarked that it was generating unlevered returns of 11% in its 3 largest small

cell markets — New York, LA, and Philadelphia. 

Much of Crown’s presence in those territories was obtained through its

acquisitions of Next G (2012) and Sunesys (2015), at prices that implied 2%-7% anchor

yields. While this suggests that Crown may

have so far correctly bet on a trend, it does not bear evidence of a moat. Moreover, keep in mind that the 11% returns

have been realized in the best performing cities, cities that accounted for

just 30% of deployed capital. Management

is asking investors to believe that the 70% of capital it has put to work in

less receptive markets will eventually deliver comparable yields. In other words, Crown has been acquiring

ahead of small cell growth, buying fiber companies at single digit yields and

hoping to lease those assets up to ~10%+ forward yields through tenancy growth

and price hikes, applying financial leverage to juice returns to high-teens. 

Crown applied the same approach to macro cells in 1999-2000, buying towers in top 25 markets at 3% starting yields, gradually walking them up to 8%-9% as the wireless market matured. We’re at an early enough stage of 5G adoption that half of small cell deployment is still done in-house by carriers (Crown claims that it alone accounts for the other half), but I would expect that, as with macro towers, the co-location model will prove more cost effective. As co-tenancy takes root, the hope is that at some point Crown will have acquired a significant portfolio of small cell assets with sticky wireless tenants generating reliable high-teens levered returns. But what’s to stop competition from running their own fiber and grinding incremental returns to uneconomic levels? I don’t know what unique resources and capabilities that Crown brings to bear.

Again, this has nothing to do with the durability of Crown’s existing tower assets. Those are fine. American Tower and Crown have a mountain of legacy infrastructure that will continue kicking off tons of cash flow because carriers don’t have a strong enough reason to switch – US carriers have talked about negotiating more favorable tower lease agreements, but there is no discernable indication that they done so in a way that harms tower economics to any meaningful degree. Competitors can’t just build adjacent to existing towers and expect to steal carrier tenancies by competing on price. 

In my Ball post, I talked about the difference between a static and self-reinforcing moat. For the most part, I see towers as possessing a static moat. Through price escalators and added tenancies, an operator can lift yields on existing assets, and with more capital it can buy more towers, but neither of those processes feed back into strengthening the character of its moat, except maybe in the following ways:

1/ all else equal, credit rating

agencies assign higher ratings to bigger companies with diversified assets, which

lowers borrowing costs for American and Crown vs. smaller peers 

2/ local carrier oligopolists like Telefonica, America Mobil, Vodafone are more inclined to work with and sell tower assets to credible and experienced infrastructure suppliers with whom they already have relationships 

And so American Tower reinvests the cash flows from its US towers into towers located in India, Africa, and Latin America at mediocre starting yields that get walked higher over time with incremental tenancies and juiced with low cost debt. Crown is basically doing the same thing, but with domestic fiber instead of international towers, though for reasons that are perhaps unjustified, I find Crown’s small cell gambit far more speculative. Its fiber assets are mostly serving enterprises, universities and whatnot (what the company refers to as “Fiber Solutions”) for the time being, but the company disclosed that Fiber Solution demand grew by just 3% last quarter (on 9% churn!) and that lower than expected demand in both small cells and fiber were responsible for the company’s lower full year organic growth guide. Anyway, Fiber Solutions are unlikely to carry the day. Crown will need to see significant small cell traction – which traction is embedded in a larger bet on the use of millimeter waves that requires other technology pieces like RF components and networking software coming together – to justify the prices it has paid for fiber. Maybe everything works out. After all, I could have raised the same doubts about tower acquisitions in the early 2000s. I’m just saying buying fiber and hoping for small cell adoption is more speculative than buying tower assets, as the regulatory hurdles seem lower and the demand prospects appear more solid for the latter (let me know if you disagree).

The tower companies trade at roughly the same multiple as Equinix, another mission critical tech infrastructure REIT enjoying strong growth tailwinds, sticky customers, price escalators, and incumbency benefits that stretch back decades. But the primary distinguishing factor between a connectivity-rich colocation datacenter operator like Equinix and a tower operator like American is that the former’s tenants create value for one another. As I wrote in this EQIX post:

“a customer will choose the

interconnection exchange on which it can peer with many other relevant

customers, partners, and service providers; carriers and service providers, in

virtuous fashion, will connect to the exchange that supports a critical mass of

content providers and enterprises. 

Furthermore, each incremental datacenter that Equinix or Interxion

builds is both strengthened by and reinforces the existing web of connected

participants in current datacenters on campus, creating what are known as ‘communities

of interest’ among related companies”

There are no such network effects in towers. Verizon Wireless’ tenancy does not compel AT&T to sign a lease. Whereas a well-funded private developer like Tillman can start building towers and claim incremental volume, Equinix’s dominance is strongly path dependent. A plucky upstart stands almost no chance spinning up a competing tenant ecosystem. To effectively compete with Equinix, you’d have to go decades back in time, convince the major telcos to land hooks in your datacenter, and from there, spin up virtual cities. Moreover, because large enterprise customers want to reach partners all over the world, global scale is relevant for IX colos[12] in a way that is not true of tower operators, whose carrier customers are local. Verizon doesn’t care that AMT has 75k towers in India. 

Where this might change is if American

and Crown are successful in dual purposing their towers, which are already wired

with fiber and connected to power, as edge datacenters, to meet the latency

demands of IoT, autonomous cars, etc. 

With 180k towers around the world, American seems uniquely positioned to

capture this promising opportunity since enterprises and colos with global footprints

(Equinix, Digital Realty) will want to partner with an infrastructure provider

that also has a global presence. But investments

so far have been small, conversations with potential partners exploratory. American purchased a small colo in Atlanta

last year; Crown invested in edge colo Vapor IO, though it has no ambitions to operate

its own datacenters. I find the edge DC

opportunity really interesting, but there is still some time (5+ years maybe?)

before it translates into a meaningful driver of revenue growth for either company.

Data centers are more operationally intensive to run than towers. Equinix, as an enterprise business that sells into a wide variety of customers, is burdened with more sales and marketing expense than American Tower, who basically leases to 3 or 4 big carriers in each market. These factors give rise to differences in EBITDA margins (62% for AMT vs. just 44% for EQIX), but the gross returns that AMT realizes on a 3-tenant tower are pretty close to those that Equinix realizes on a fully leased datacenter and their gross unlevered pre-tax returns on capital are similar too, in the 10%-12% range. Both companies have assets generating returns that are hard to compete away, because of network effects in the case of Equinix and zoning-related scarcity and cost/value proposition in the case of American Tower.

Analysts have long fretted that American and Crown’s profits could be compromised by carrier consolidation[13], an issue that is front and center again now that regulators have approved the merger between T-Mobile and Sprint. As part of the deal, the two companies will be divesting 20k cell sites to DISH, who will be allowed access to T-Mobile’s network for the next 7 years while it spends $10bn[14] building out its own wireless infrastructure. DISH is also acquiring certain spectrum assets from T-Mobile/Sprint – adding to the billions of spectrum assets that it is currently squatting on – and buying the Boost Mobile and Virgin Mobile businesses, propelling itself into the wireless market with 9mn subscribers.

The reason the government blessed this deal at all is because the new market structure, which replaces an enfeebled Sprint with a viable #4 (Dish) and a fortified #3 (T-Mobile + Sprint) – is better positioned to develop 5G networks. So now Dish, which prior to the merger was building an IoT network, will be building out 5G as it works to meet its commitment to the FCC of covering 70% of the US population with the new standard by June 2023. 

The new T-Mo will be decommissioning a

bunch of towers post-merger, which would seem to present revenue risk for Crown Castle and American, who derive 1/3 and

16% of total revenue from Sprint and T-Mobile. 

On towers where they overlap, S and TMUS account for just 5%-6% of Crown’s

revenue and more like 3%-4% of American’s. 

In reality, to meet its aggressive synergy targets, T-Mobile will probably

have to shut down more than just the overlapping Sprint sites…though probably not

all of them since it still needs the coverage to compete effectively with

Verizon and AT&T. Let’s just say

somewhere between 3% and 8% of American’s revenue and 5% to 16% for Crown is

potentially at risk. 

But it will take some time, maybe 5 to 10 years, for this revenue to slide off due to the long-term rental agreements already in place and the practical timing constraints of transitioning subs to a new network (it’s worth noting that when Sprint acquired Nextel back in 2005 and shut down the latter’s iDEN network, it renegotiated its lease agreements with American Tower to dilute the churn impact, and I imagine a similar arrangement would be negotiated here). Moreover, to ensure consistent coverage, replacement antennas will have to go up before any towers are decommissioned. Plus, Dish will likely assume some of those abandoned leases. So, taking all this into account, I don’t really see the T-Mobile/Sprint merger as a material negative for either American or Crown.

It may even be a positive development. Changes in market structure attract lots of attention because they give can sometimes give rise to temporary blips in churn, but what we don’t see is the alternative history – the next generation networks that wouldn’t be built and the associated capex that wouldn’t occur in the absence of well-capitalized carriers that emerge from consolidation. Last year, merger activity caused AMT’s churn in India to spike to 20% and organic revenues to decline double-digits. But consolidation-related disruption is a necessary growing pain for the 3G/4G network buildouts that will drive AMT’s double-digit organic growth in India for many years. In the US, the number of major wireless carriers has declined from 7 to 4 from over the last 15 years, and yet, both AMT and CCI have evolved into stronger and more profitable entities. The new T-Mo plans to spend more together than TMUS and S did separately ($40bn over the first 3 years, more than T-Mobile and Sprint have together spent in any 3 year period as separate companies). In short, the tower builds and amendment cycles that accompany robust underlying demand for wireless services tends to more than offset the deleterious impact from changes in market structure. Whether there are 5 carriers or 3 serving the market, the aggregate demand for data will continue to grow.

American and Crown trade at 33x and 26x adjusted FFO (approximately, net income + depreciation + one-timers – maintenance capex), respectively….which I think is, like, you know, fine for a couple of 6%-7% organic revenue growers with a long track record of delivering ~mid-teens returns on equity, though I have more confidence that American can deploy capital at comparably attractive returns through international expansion than I do that Crown Castle can through US fiber builds[15]. At 30x year 7 AFFO and after assuming 2% share count dilution, an AMT investment compounds at maybe ~10%, including dividends from its current price ($254).

Footnotes

[1] If you run the math, a typical

carrier saves ~$200k in present value per tower over a 20-year term by

co-locating on an existing site relative to building its own.

From American Tower’s June 2019 Investor Presentation:

[2] Per wirelessestimator.com, the top 130 tower companies account for 122k

towers, with close to 60% of those companies managing portfolios of fewer than

100 towers. AMT, CCI, and SBAC combined

have 95k towers in the US. 

[3] From 2013 to 2015, when the 4G

amendment phase was in full swing, SBAC was seeing some amendments hit $2k or

more per month.

[4] massive-input, massive-output (MIMO) combines a bunch more antennas at the base station with other technologies and design changes – like spatial multiplexing, where data between the tower and the smartphone runs through multiple virtual pipelines, allowing more data to flow vs. a single pipeline; and beamforming, which concentrates signals in a specific direction – to boost capacity and hasten data rates.

[5] I calculate the purchase price as

the sum of (a) the upfront $5.1bn prepaid lease + (b) the present value of the

$5bn aggregate option payment in year 28, using a 9% discount rate + (c) $220mn

of start-up capex

[6] In 1995, American Tower was

founded as a subsidiary of American Radio Systems to manage and develop

broadcast tower sites.

[7] in 2015, the wireless sector in

India was where the US was in 2005, with most of the population on 2G. 

[8] among the major wireless

consolidations: Vodafone’s August 2018 merger with Idea Cellular and Bharti

Airtel’s purchase of Tata Teleservice’s mobile business. The pending sale of Reliance Communication’s

telecom assets to Reliance Jio was recently terminated after the companies failed to secure

regulatory approval.

[9] American bought out Tata’s

interest in the following years, bringing its ownership of the JV up to

90%. 

[10] Because a carrier that wants to

further densify can claim a second spot, Crown believes that over the long-term

the likelihood of 3+ tenancies is higher with small cells than it is with

towers.

[11] Per CoBank (link)

[12] Through Equinix’s Global ECX Fabric, any customer can reach any other customer located in any Equinix datacenter in the world.     

[13] In the US, AMT and CCI derived

88% and 73% of 2018 site revenue from just 4 carriers: AT&T, Verizon

Wireless, Sprint, and T-Mobile. In

India, 4 carriers account for 95% of AMT’s revenue; in each of EMEA and Latin

America, 2-3 carriers are responsible for over half. 

[14] MoffettNathanson analyst Craig

Moffett thinks the notion that Dish can build an entirely new network for $10bn

to be “silly” considering that Verizon spends $15bn just maintaining its existing

network[a].

[a] Unburdened by the legacy infrastructure

underpinning previous wireless standards, Dish is pulling a Rakuten and deploying

a modern network that uses off-the-shelf hardware and leverages public clouds,

which it claims will result in significant capital savings relative to a

traditional build. 

[15] fwiw, AMT’s pre-tax returns on

capital and equity have historically exceeded those of Crown.

Disclaimer: At the time this report

was posted, Forage Capital held a position in AMT. This may have changed

at any time since.