A strong U.S. market is driving macro tower demand for all the tower companies we cover, and American Tower's second-quarter results show it is no exception. We don't expect the higher level of tower revenue growth will materially wane any time soon, as increasing mobile data use and 5G deployment will require carriers to maintain high levels of spending. We think those factors dwarf potential headwinds from U.S. carrier consolidation, which will likely lead to incrementally higher churn over time but may be largely offset by spending from Dish. Nonetheless, with few surprises in American Tower's report, we expect only a modest increase in our $160 fair value estimate. Although American Tower is our favorite and we expect relative long-term outperformance versus peers SBA and Crown Castle, we currently see all tower companies as materially overvalued.
American Tower's U.S. growth was again stronger than SBA's, but unlike SBA, American Tower's growth rate on existing towers decelerated. Organic growth on existing U.S. towers was 7.5% year over year, down from about 8% growth each of the past two quarters. We attribute the deceleration primarily to tougher comparisons American Tower now faces, as the second quarter last year was when the firm first took a big step up in growth. We forecast the growth rate to average 7% over the next several years.
As expected, international performance was again affected by churn resulting from carrier consolidation in India, which led to a 3.6% decline in existing tower revenue, an improvement over last quarter. Indian churn should come down substantially by the fourth quarter, which is when we expect international revenue to return to growth. Adjusted for the abnormal churn, international revenue on existing towers grew 8%, similar to our long-term projections.
We think American Tower's strategy to diversify its tower portfolio globally leaves it best positioned among the three U.S. tower companies, as it is primed to benefit from the continually increasing demand in mobile data worldwide.
The tower business in general is very attractive. Wireless carriers enter long-term leases that include rent escalators (annual increases of about 3% in the United States and generally tied to inflation internationally), which gives the firm a highly visible and stable revenue stream, and towers have significant operating leverage. Adding tenants to existing towers and equipment upgrades by tenants both provide sharp revenue increases while adding little incremental cost. As data use grows and networks get more stretched, locating equipment on more towers and upgrading equipment are primary solutions for carriers. Consumers' ever-increasing need for mobile data and little geographic overlap with competitors result in normalized churn of about 1%.
We think American Tower is well served being the market leader in the U.S, which is the most profitable market. American Tower generates more than half its total revenue and has about one fourth of its towers in the country. In addition to being the most profitable geography, the U.S. growth trajectory remains strong. Mobile data use has been growing 30%-40% per year and is expected to continue at a similar clip through 2021. We think the looming upgrade to 5G networks will extend the required carrier investment for years to come.
American Tower's international growth has been depressed recently due to consolidation of mobile carriers in India, which has led to high churn. We expect conditions to largely improve after 2019. Long term, we think prospects are bright in American Tower's international markets, which are dominated by India, Brazil, and Mexico. Many international markets are a decade behind the U.S. and are now building their 4G networks. Indian data usage has been growing 100% per year, and carriers have been substantially increasing spending to improve networks. Latin American countries now have over 200 million people in the middle class, making them potential smartphone users.
We assign American Tower a narrow economic moat due to the efficient scale that underlies the tower industry and switching costs that make tenants loath to seek alternative providers once they have installed their equipment on towers. American Tower's returns returns have exceeded 10% in each of the last 10 years, above the 8% weighted average cost of capital we estimate for the firm. Given the critical nature towers play in allowing the public access to mobile data, the lack of near-term substitutes to towers, and the barriers to alternative tower providers, we expect excess returns to continue for the next decade as well, resulting in our narrow moat rating.
The overriding feature of the tower business is the extreme operating leverage inherent in it. Costs to the tower provider are mostly fixed per tower, meaning costs to operate a tower are virtually the same whether a tower has one tenant or several tenants. Tower operating expenses typically consist primarily of the rent expense the companies pay to lease the land, which is subject to multiyear leases. The tower business is very capital intensive, and returns are typically not good unless or until the tower provider can secure multiple tenants on the tower, a feat made more difficult for a competitor by the limited pool of potential customers. In the U.S., for example, over 90% of American Tower's 2018 revenue was generated by only the big four wireless service providers, or carriers. Customer concentration is similar throughout the industry, and it applies in American Tower's other markets as well, though not to the same extent. Customer concentration levels in American Tower's other geographies range between 60% and 80% and should increase as carrier consolidation in those markets continues.
American Tower states that an average U.S. tower costs about $275,000 to build; rents for the initial tenant are about $20,000 per year; and the company's operating expenses with one tenant on the tower are $12,000, resulting in only a 3% return. By American Tower's estimates, the return jumps to 13% once it has added the second tenant and can reach 24% with a third. Globally, American Tower averaged 1.9 tenants per tower at the end of 2017. Consequently, even before considering other hurdles an upstart competitor would have to overcome to compete on American Tower's turf, it would have to contemplate sinking tremendous amounts of capital into building towers, locking itself into $1,000 per month land lease payments over a term of 5-10 years, and, given a likely need to offer lower average prices to carriers to entice them to switch, would need to average close to two tenants per tower in hopes of generating a slim excess return on its capital. As a reference, Crown Castle has stated that it has historically added tenants per tower at a rate of about one every 10 years.
We think the existing efficient scale alone is a deterrent to competitors, but they would have other obstacles to procuring carriers' business. First, because carriers own and are responsible for the equipment they deploy at and on the towers, they bear the brunt of removing it when they leave the towers. Industry peer Crown Castle has stated that it costs carriers about $40,000 to remove equipment from a tower in the U.S. With American Tower's U.S. tenants paying between $20,000 and $30,000 per year in rent, it means they would have to recoup one and a half to two years' worth of rent payments to break even. A rival would need to undercut American Tower's prices by that much more to make up for the loss carriers would be facing in deciding to switch, further impeding a competitor's prospects of generating excess returns. Yet direct costs are not the only switching costs carriers face. Carriers' most valuable assets are their networks, and any disruption to the network, whether temporarily during a tower switch or more permanently as a result of a new setup and location, risks customer dissatisfaction with their service. Given the negligible switching costs to carriers' customers, we don't think network risk is one that carriers would take lightly or put in jeopardy for minimal cost savings, adding even more to the amount a rival would likely need undercut prices to coax carriers into making the gamble. History suggests that carriers hesitate to leave towers that they are on, as churn for American Tower is typically 1%-2%, inclusive of churn related to industry consolidation.
Even if a competitor wanted to take on an incumbent, it could struggle to obtain the permits necessary to build a tower. Towers are subject to zoning restrictions, and zoning authorities and community residents often oppose tower construction in their communities. The opposition has obviously not been such an impediment that it precludes towers from ever gaining approval, but we think additional towers in areas that already have sufficient network coverage would face significantly greater difficulty obtaining approval than those that are built to improve a network, thus making tower construction for the sole purpose of providing competition a much more difficult regulatory endeavor.
While we think barriers to competition for the existing public tower companies are overwhelming in the near term, a couple of factors reduce our visibility over a 20-year span, keeping us from labeling American Tower's moat as wide. First, the tower companies typically lease significantly more land than they own. At the end of 2018, American Tower owned only about 10% of the land under its towers. Its leases typically have initial terms of five to 10 years and give it the option for at least one renewal period. Under existing leases, American Tower controls the land under more than 46% of its towers until 2028 or later, so we don't think it is a significant near-term concern. However, as leases need to be renegotiated, we have less confidence that the tower companies have leverage over land owners or advantages versus other potential land tenants. We think competitors are unlikely to build competing towers in tower vicinities; we think their better competitive move would be to secure available land where towers already exist. While incumbents may be best positioned to hold on to their land, they may be forced to do it on less favorable terms, which could affect returns.
The other factor that could affect towers over the long term is technology changes. We expect 5G to become the standard in the U.S. at some point during the next decade, and with 5G will come greater use of small cells, meaning carriers will rely more on smaller structures, such as light poles, and potentially less on macro towers like American Tower operates. Again, we are less concerned that this will be a major threat in the near term for American Tower. First, about 45% of the company's property revenue comes from outside the U.S., and many of those markets lag the U.S. technologically and will continue building 4G networks over the next decade. Second, because small cells have much smaller coverage areas, many more of them need to be deployed to blanket the same area that towers do. As a result, we think even if they were ultimately to be a replacement, which we doubt, it will take a long time horizon to get there. Small cells are significantly more expensive to deploy over a given area of coverage, meaning carriers are likely to continue to prefer macro towers when feasible. Since American Tower's locations are skewed toward rural and suburban locations rather than dense, urban markets, we expect theirs would either be the last towers to be disrupted by small cells, or, perhaps more likely, macro towers will continue to be the preferred structures in those areas even with a 5G standard. It is too early to say exactly what the 5G network will require and how big a threat it is to the macro tower model. We don't expect it will negatively affect American Tower substantially over the next decade, but we cannot yet speculate on what it or subsequent technologies might mean to the tower model 20 years from now.
We are raising our fair value estimate for American Tower to $163 from $160, due primarily to the time value of money. Our valuation implies P/AFFO and adjusted EV/EBITDA multiples of 21 and 20, respectively, on our 2019 forecasts, both roughly in line with where the stock has traded for much of the past decade.
We project revenue growth to average about 7% annually over the next five years while adjusted EBITDA margin expands 400 basis points from 61% in 2018, more comparable to levels seen at the beginning of the decade. Still, we expect gross margin to expand less, as the lower margin international business becomes a bigger portion of total revenue.
We project U.S. revenue to grow 6%-7% annually over our five-year forecast, with colocation and amendment growth remaining close to 5% per year as carriers continue investing in their networks to accommodate greater data usage; escalators holding steady at 3%; and churn gradually getting to below 1% as contracts for recently defunct carriers continue to get worked off. We assume relatively few towers will be built and acquired in the U.S., contributing to a modest decline in the company's capital expenditures and to a 300-basis-point rise in segment gross margin.
We expect investment to stay elevated in American Tower's international geographies, where we forecast the tower portfolio to grow by about 3,000 to 4,000 per year. We expect collective international growth on existing towers to be 8%-9% annually from 2020-23 (after the effects from Indian carrier consolidation moderate), and we expect new tower billings to add about 2% per year after 2019. The new towers and a bigger revenue shift toward the lower-margin Indian business should lead to a slight drop in gross margin over the next few years, but we expect it to be back to 2018 levels by 2023.
We see American Tower as a medium uncertainty stock. Though the magnitude of growth will likely vary, we are confident mobile data usage will continue growing throughout the world, and American Tower is well positioned to benefit.
Beyond simple variance in demand growth, we see a few secular risks, but we don't think any are cause for significant concern.
Greater use of small cells by carriers, which we expect to be more likely with 5G, could affect tower demand. American Tower's doesn't participate in the U.S. outdoor small cell market, so any move away from towers would be detrimental. However, we believe towers will continue to be carriers' most cost-effective option when they are available. Further mitigating the small cell risk is American Tower's geographical footprint, which is predominantly suburban and rural. Less densely populated areas are far less conducive to small cells, so if small cells become a viable alternative for some towers, we expect it to be mostly just in urban areas for the foreseeable future.
Another long-term risk is that American Tower perpetually controls only 31% of its U.S. land, leaving it largely subject to landlords and leases. We think competitors are hard-pressed to compete with alternative towers, so they could instead bid for American Tower's land as leases expire. However, American Tower's diversified landlord base (90% of its leases are with landlords who own only one site) makes big losses less likely. Also, American Tower owns its towers, so changes in land control could be complicated. Although American Tower owns very little international land, those geographies pose less concern, because land is generally cheaper, and rents are directly passed on to carriers.
Health of carriers is important, but we find it less significant. Though consolidation or carrier bankruptcies would cause volatility, we think investment ultimately ends up at a similar place to support the networks required to meet consumers' data demands.
We label American Tower as having Standard stewardship. The board is made up of 10 directors, nine of whom are independent, and each is up for re-election annually, which we see as shareholder-friendly. However, seven directors have been on the board for more than 10 years, which causes us some concern about their ability to be truly independent from management. The board is chaired by Jim Taiclet, who has served in that role since 2004 and has also been CEO since 2003. While we think a splitting of the chairman and CEO roles is preferable in the name of good corporate governance, we think strategic decisions for American Tower have a long record of being sound.
In our view, management has been smart to diversify its investments over a wide range of countries to participate in data growth in developing countries. A significant presence in India, Brazil, and Mexico gives it multiple markets in which to profit, while limiting exposure in case of slower market development. The company regularly spends multiple billion dollars per year in acquisitions (mostly towers) in the U.S. and abroad, yet it consistently posts excess returns on its invested capital (every year since at least 2008) and has rarely written down goodwill. We also project it will continue to return over $1 billion in capital to shareholders each year with its dividend, which, as a REIT, must continue to grow along with earnings growth.
We think compensation of management is well aligned with the best interests of shareholders. Only 10%-15% of executive compensation is in the form of base salary, and about 75% of target compensation is in the form of American Tower stock. Executive bonuses are 80% based on company, rather than individual, performance. Further, executive officers and directors are required own stock worth 3-6 times their base salaries (annual retainers for directors). Each of the company's seven executive officers has been with American Tower for at least nine years.
