Cell towers are among the best real estate an ordinary investor can own, but the price swings hard on interest rates and the biggest losses have been self-inflicted by management, so the return lives less in the steel than in what you pay and who runs it.
Key takeaways
- Market size: the global cell-tower market sits at about $50bn in 2025, forecast to $75bn by 2034; there are more than 4.4 million telecom towers worldwide.
- The demand engine: mobile data traffic rising 146 → 328 exabytes a month by 2031, with 5G’s share of traffic climbing from 48 per cent toward 85 per cent.
- The economics: one tenant earns ~3 per cent return on investment; three tenants earn ~24 per cent, on leases with ~3 per cent annual escalators.
- The main vehicles: US-listed REITs American Tower (AMT), Crown Castle (CCI) and SBA Communications (SBAC); non-US pure-plays including Cellnex (BME: CLNX) in Europe, Indus Towers (NSE: INDUSTOWER) in India, Helios Towers (LSE: HTWS) and IHS Holding (NYSE: IHS) in Africa and emerging markets; a digital-infrastructure ETF at a 0.50 per cent fee; and a London-listed global infrastructure fund, the iShares Global Infrastructure UCITS ETF (LSE: INFR) at 0.70 per cent, priced in sterling.
- The cautionary tale: Crown Castle agreed to sell its fibre and small-cell business for $8.5bn after investing more than $17bn in it, a multi-billion-dollar reminder that the risk is capital allocation, not the towers.
The 60-second version
Every call you make, every video you stream on the train, every map that finds you in a strange city: the signal has to touch a piece of steel somewhere before it reaches you. That piece of steel is a cell tower, a tall structure whose entire job is to hold antennas high enough that a mobile network can reach the ground around it. For most of the mobile era these were things the phone companies built and owned themselves, a cost of doing business nobody thought of as an asset. Then a quieter idea took hold: that the tower and the network running on it are two different businesses, and the more boring one, the landlord, might be the better one to own.
The reason is a peculiar bit of arithmetic. A tower costs roughly $275,000 to build in the United States, and almost all of that cost is fixed the day it goes up. Put one mobile operator’s antennas on it and you earn a thin return. Put a second operator’s antennas on the same steel, which adds almost no cost, and the extra rent falls straight to the bottom line. The margin on a single-tenant tower runs about 40 per cent; at three tenants it runs about 83 per cent, according to Starlight Capital. You build the fixed asset once and then rent it repeatedly, which is most of what there is to understand about the business.
Sitting underneath that arithmetic is a demand curve that is hard to argue with. Global mobile data traffic is forecast to rise from 146 exabytes a month at the end of 2025 to 328 by 2031, more than double, on Ericsson‘s numbers, and every one of those bytes needs a radio to carry it. There is a complication. A tower is a long-duration, rate-sensitive asset that fell 43 per cent when interest rates rose sharply, and the biggest risk turns out not to be the towers at all but what management does with the cash they throw off.
I. What it is
A cell tower is a tall structure built for one purpose: holding radio antennas high above the ground so a mobile network can cover the area around it. It is one of those objects you have seen ten thousand times and never once looked at. It might be a steel lattice on a hill, a monopole beside a motorway, or a disguised “tree” on the edge of a car park. The engineering is almost beside the point. What matters for an investor is who owns the steel and who pays to hang things on it.
Here the industry splits into two businesses that used to be one. The mobile network operator, the phone company you pay every month, such as a Vodafone or an AT&T, runs the network: the radios, the spectrum, the customers. The tower company owns the physical structure and the land rights beneath it, and rents space on the tower to the operators. In the old model the operator owned both. In the modern model, which now dominates the industry, the operator sells its towers to a specialist and leases the space back, freeing capital to spend on the network itself. That single decision, separating the passive real estate from the active network, is what created the asset class.
The unit that matters is the tenant: one mobile operator’s set of antennas and ground equipment on a given tower. A tower with one operator’s kit on it has one tenant; add a second operator and it has two; a third makes three. The number of tenants per tower, the “tenancy ratio”, drives the returns more than anything else, because of the cost structure described earlier and worked through in Section VII.
For an investor, the framing is the same as any good property thesis. When you buy shares in a tower REIT, a REIT being a real estate investment trust, a listed company that owns income-producing property and passes most of its profit to shareholders, you are buying the landlord, not the network. You own the steel, the land rights and the leases, rather than a bet on which mobile operator wins the market. You own the structure all of them have to rent regardless of which one wins, which is the picks-and-shovels position: the tower gets paid whichever network carries the traffic, as long as the traffic keeps growing. Section XIII deals with what happens if the number of tenants stops growing.
II. Market history and growth
The mobile tower is nearly as old as mobile telephony, but the asset class, towers owned by specialists rather than by phone companies, is a creation of the last quarter-century. For most of the industry’s history, an operator that wanted coverage built its own towers, and a competitor that wanted coverage in the same place built a second tower next to it. That was wasteful in the extreme: two structures, two land leases, two sets of steel, to do a job one structure could do. The insight that built this sector was that the tower is a shared, passive asset, closer to a toll road than to a technology product, and that a company doing nothing but owning and sharing towers could earn better returns than the operators renting from it.
The scale today is substantial without being enormous by the standards of, say, data centres. The global cell-site-tower market sits at about $50bn in 2025 and is forecast to reach roughly $75bn by 2034, a compound annual growth rate of about 5.5 per cent, on Market Reports World‘s numbers. The United States, the most mature single market, ran at about $8.7bn in 2024, rising to $9.1bn in 2025 and forecast toward $14.8bn by 2033, roughly 6 per cent a year. These are not hypergrowth numbers. They are the steady, compounding numbers of an infrastructure asset, which is precisely the point.
The physical footprint is what gives the sector its heft. There are more than 4.4 million telecom towers worldwide, supporting over 8.5 billion mobile connections, and the top 100 tower companies alone run more than 1.7 million shared sites, per Dgtl Infra. “Shared” is the word doing the work there. A shared site is one built to carry multiple operators, which is the whole economic model made physical.
The growth driver that does not show up in the headline market size, but matters more than any of it, is densification, the industry’s word for packing more radio capacity into the same geography as data demand rises. One number captures it: the number of towers required per 1,000 subscribers rose 12 per cent between 2021 and 2024. Read that slowly. The same population needs meaningfully more towers than it did three years earlier, not because there are more people but because each person is pulling far more data through the air. That is the mechanism by which a market growing subscribers at low single digits generates tower demand well above that rate.
III. Demand drivers
Market-size forecasts are cheap. What makes the tower thesis credible is that the demand is measurable in the raw traffic flowing across mobile networks, and that traffic is growing on a curve that is difficult to bend.
Start with the headline. Global mobile data traffic reached 146 exabytes a month at the end of 2025 and is forecast to hit 328 by 2031, roughly a 2.2-fold rise, on Ericsson‘s numbers. Fold in fixed wireless access, home broadband delivered over a mobile network rather than a cable, and the total rises from 203 to 515 exabytes a month, about 2.5 times. An exabyte is a billion gigabytes. The precise number matters less than the shape of the line, which points relentlessly up. Every byte on that curve has to be carried by a radio, and radios hang on towers.
The second driver is the generational shift to 5G, the fifth generation of mobile network technology, faster and higher-capacity than the 4G it replaces, but with shorter range, which means it needs more sites to cover the same ground. During 2025 alone, 5G’s share of mobile traffic jumped from 34 per cent to 48 per cent, and is projected to reach 85 per cent by 2031. That transition is a tailwind for towers twice over: it drives the traffic that fills existing sites, and its shorter range drives the densification that fills new ones. The deployment is already global and well past the pilot stage, with more than 210 operators across 95 countries having launched commercial 5G, on Cohen & Steers‘ count.
The third driver tells you where the growth is skewed: emerging markets. In India, per-smartphone data use had reached 37 gigabytes a month by the end of 2025 and is heading for 70 by 2031, against a global average rising from 22 to 42 gigabytes, on Ericsson data reported by TelecomLead. A country where each phone already burns nearly twice the global average, still growing fast, is a country that needs towers built at a pace the mature markets left behind years ago. That skew is both the sector’s biggest growth opportunity and, as Section X shows, one of its sharpest traps, because the same emerging markets carry weak-tenant and weak-currency risk that has cost at least one major operator dearly.
The demand case is unusually clean because all three drivers push the same way and every byte they generate has to pass through a tower to reach a phone.
IV. The players
The firms and people who dominate an asset class are a good way into it. The big owners of towers fall into a few groups: the American listed giants, the international pure-plays, and the private infrastructure capital circling the sector.
The American listed giants. The US public market is dominated by three names. American Tower (NYSE: AMT) is the largest tower REIT in the world, with a market capitalisation of about $95bn and roughly 225,000 communications sites globally. It is run by Steven Vondran as President and Chief Executive. Crown Castle (NYSE: CCI) is the second American giant, at about $46bn and more than 80,000 towers, now entirely US-focused after divesting its fibre business, a divestiture that is itself the cautionary tale of Section X. SBA Communications (NASDAQ: SBAC) is the third, smaller but tightly run, at about $20bn. It posted FY2024 revenue of $2.68bn and adjusted funds from operations of $13.37 per share, up 2.2 per cent, and raised its quarterly dividend 13 per cent, on Zacks‘ report. “Adjusted funds from operations” (AFFO) is the REIT world’s preferred profit measure: cash earnings after the property depreciation that accounting demands but that a well-maintained tower does not really suffer.
The international pure-plays. Outside America the sector is younger and, in places, larger by site count. Indus Towers (NSE: INDUSTOWER) in India runs 259,622 towers on FY24 revenue of ₹301,710m, about $3.6bn, and its network has since surpassed 219,736 macro towers by 2025, more sites than American Tower on a fraction of the revenue, which tells you everything about the difference between a mature-market and an emerging-market tower. Cellnex (BME: CLNX) is Europe’s largest independent, Madrid-listed and part of the IBEX 35, with more than 100,000 towers across Europe, a market capitalisation of about €20bn, and FY2025 revenue of €3,995m. It is run by Chief Executive Marco Patuano, whose strategic emphasis is on cutting debt. IHS Holding (NYSE: IHS) operates more than 37,000 sites across seven emerging markets spanning Africa and Latin America, including Nigeria, Brazil, Colombia and South Africa, the highest-growth, highest-risk end of the spectrum. Helios Towers (LSE: HTWS), London-listed with a market capitalisation of about £2bn, runs more than 14,000 tower sites across nine African markets and Oman. And Vantage Towers, the Vodafone-spun European operator, runs 87,824 sites across ten European countries as of March 2025, though it is no longer directly investable on-exchange, having been delisted from Frankfurt in May 2023 after the KKR and GIP consortium took control.
The private capital. The most telling recent buyers of tower assets have been private infrastructure funds rather than public investors, and the price they have paid is the clearest signal of how the sector is valued when it changes hands privately. A KKR and Global Infrastructure Partners consortium’s move on Vantage Towers, detailed in Section X, put a premium on tower assets that the public market rarely matches. The private-infrastructure bid is a recurring feature of this sector, and a reason listed valuations tend to hold a floor.
Here is American Tower’s chief executive, on the company’s fourth-quarter 2024 earnings call, framing the business:
“We have an exceptional portfolio of assets, unmatched operating capabilities, and we’re in what I believe to be one of the most durable businesses, capitalized by ever-increasing mobile network demand.”
— Steven Vondran, President & CEO, American Tower, Q4 2024 earnings call, 25 February 2025
On the same call he added that the secular demand should require a doubling in wireless network capacity between now and 2030. A doubling of the thing your towers carry, over five years, is the bull case stated by the person accountable for delivering into it, which is worth weighing against the fact that he is also the person with the most reason to state it warmly.
V. Geography
Towers are a global asset, but the economics change so completely across regions that “cell towers” almost means a different investment depending on where the steel stands. The variables that move are tenant credit quality, currency, tenancy ratios and growth rate, and they pull in opposite directions.
North America is the deepest and most mature market. The three American REITs, American Tower, Crown Castle and SBA, anchor it, and the US market alone is worth about $9.1bn in 2025. Its advantage is tenant quality: a handful of large, well-capitalised operators sign long leases in a stable currency. Its disadvantage is that most of the easy densification is done, so growth is steadier and slower, the roughly 6 per cent annual figure, not a frontier-market rate.
Europe clusters around a consolidating set of independents, of which Cellnex, with 111,688 sites, is the largest, followed by Vodafone’s Vantage Towers at 87,824 sites. The European story right now is less about land-grab expansion and more about balance sheets. The operators that once spent aggressively to buy towers are now deleveraging, a theme Cellnex’s chief executive has made explicit. For an investor that means the listed European names, chiefly Cellnex on the Madrid exchange, offer a maturing, cash-returning profile rather than a growth one, with the euro rather than the dollar as the underlying currency.
Asia-Pacific, led by India, is where the raw growth lives. Indus Towers alone runs 259,622 towers and is listed in Mumbai on the NSE under INDUSTOWER, and Indian per-phone data use is heading toward 70 gigabytes a month by 2031. The growth is real, but so is the risk: emerging-market operators are weaker credits, and the rupee can erode a foreign investor’s return regardless of how the towers themselves perform.
Africa and Latin America are the frontier, home to IHS Holding’s 37,590 towers across seven emerging markets and Helios Towers’ 14,000-plus sites. Helios lists in London, IHS in New York, so a sterling or dollar investor can reach this frontier through a familiar exchange even though the underlying towers sit in Nigeria, Tanzania or Brazil. These markets have the steepest coverage-growth curves, huge young rapidly connecting populations, and the highest political and currency risk. The tower economics can be excellent on paper and painful once translated back into hard currency.
The investment lesson from geography is that the tenancy ratio and the tenant’s creditworthiness matter more than the growth rate. A slower-growing tower rented to a rock-solid operator in a stable currency can be worth more than a fast-growing one rented to a stretched operator in a currency that halves. An investor in the listed American or European REITs gets the first; an investor reaching for emerging-market growth through Helios, IHS or Indus is taking the second, whether they price it or not.
Here is Cellnex’s chief executive, on the company’s fourth-quarter 2024 results call, on where the cash from asset sales is going:
“the possible cash coming from the two disposals will be allocated … to debt reduction because, of course, some EBITDA will be reduced in the disposals. And so we don’t want to worsen our credit ratio and the remaining part will be allocated on further return to shareholders.”
— Marco Patuano, CEO, Cellnex Telecom, Q4 2024 results call, 26 February 2025
That answer captures the European tower theme. The operators are paying down debt and returning what is left rather than trying to buy up more towers.
VI. How to actually invest
There are five practical routes into this asset class, each trading off access, cost and control against the others. Only two of them are US-only; the rest let you buy the same thesis in euros, sterling or rupees.
| Route | Example | What you own | Minimum | Typical fee | Liquidity |
|---|---|---|---|---|---|
| US tower REIT | American Tower (AMT), Crown Castle (CCI), SBA (SBAC) | Shares in a US-listed landlord owning tens of thousands of towers | One share | Brokerage commission only | Daily, on-exchange |
| European pure-play | Cellnex (BME: CLNX) | Shares in Europe’s largest independent towerco, in euros | One share | Brokerage commission only | Daily, on-exchange (Madrid) |
| Emerging-market pure-play | Indus Towers (NSE), Helios Towers (LSE: HTWS), IHS Holding (NYSE: IHS) | Shares in a regional tower operator across India, Africa or Latin America | One share | Brokerage commission only | Daily, on-exchange |
| Digital-infrastructure or global-infra ETF | Global X VPN (US), iShares Global Infrastructure UCITS (LSE: INFR) | A basket of towers, data centres and digital-infra names | One share | 0.50 to 0.70 per cent a year | Daily, on-exchange |
| Private infrastructure fund | KKR / GIP-style vehicles (see Vantage Towers deal) | A stake in a privately held tower portfolio | Institutional (often seven figures) | Management + performance fee | Locked, multi-year |
The listed REIT is the default route and the one most investors will use. In the US you buy a share in American Tower, Crown Castle or SBA Communications the way you would any other stock, hold the underlying real estate through the corporate wrapper, and receive the return largely as an ongoing distribution. The minimum is a single share and the only fee is your broker’s commission. The trade-off is that you take whatever the public market thinks a tower is worth on the day, which, as Section VIII shows, can swing hard with interest rates.
The non-US pure-play is the same mechanism aimed at a different region and a different currency. Cellnex gives you European consolidation in euros on the Madrid exchange; Indus Towers gives you Indian growth in rupees; Helios Towers gives you African frontier exposure in sterling on the London exchange; and IHS Holding gives you an Africa-and-Latin-America portfolio listed in New York. You are choosing a geography and, with it, a currency and a tenant-credit profile. A UK or European investor can own the tower thesis without ever touching a US-domiciled REIT.
The ETF wrapper buys the theme rather than a single name, and here too there is a non-US route. In the US, the Global X VPN ETF tracks the Solactive Data Center REITs & Digital Infrastructure Index, a basket spanning data centres, cell towers and digital-infrastructure hardware, at a total expense ratio of 0.50 per cent. For a sterling investor, the London-listed iShares Global Infrastructure UCITS ETF (INFR) tracks the Macquarie Global Infrastructure 100 Index at a 0.70 per cent expense ratio, holding towers alongside utilities, pipelines and transport infrastructure. The advantage of either is diversification and one decision; the cost is that towers are only a slice of what you own, so it is an infrastructure bet more than a pure-tower one.
The private infrastructure fund is the institutional route, and it is where the highest prices for tower assets are struck. The Vantage Towers transaction in Section X is the example. It offers direct ownership of a portfolio at a scale individuals cannot reach, but at the cost of high minimums, multi-year lock-ups, and layered fees. For almost everyone, the listed routes are how this asset class is actually bought, and those listed routes now exist in dollars, euros, sterling and rupees alike.
VII. Unit economics
Why anyone builds a tower comes down to the arithmetic of stacking tenants on it, and that arithmetic is worth reading slowly.
Start with the cost. A tower costs about $275,000 to build in the United States, at a 7 per cent capitalisation rate. A capitalisation rate, “cap rate” in the trade, is the annual income a property produces divided by its price, so a 7 per cent cap rate on a $275,000 tower implies roughly $19,250 of annual net income from the first tenant. The critical feature is that the $275,000 is nearly all fixed the day the tower goes up: the land lease, the foundation, the steel and the power connection do not change when you add a second tenant to the structure.
The tenant-stacking ladder carries the whole investment case, and it looks like this.
| Tenants on the tower | Gross margin | Return on investment |
|---|---|---|
| 1 tenant | ~40 per cent | ~3 per cent |
| 2 tenants | ~74 per cent | ~13 per cent |
| 3 tenants | ~83 per cent | ~24 per cent |
Read down that return column. One tenant returns about 3 per cent, a mediocre number that would not justify building anything. Add a second tenant and the return jumps to about 13 per cent. Add a third and it reaches about 24 per cent. This happens because the second and third tenants add almost no cost and almost no new capital, so the extra rent they pay falls straight to margin. You built the fixed asset once, and every additional tenant is close to pure profit on an unchanged cost base. That is why the tenancy ratio, not the number of towers, is the number that determines whether a tower company is a good business or a mediocre one.
Two features make the rent itself durable. Leases run 5 to 10 years with annual escalators of about 3 per cent, the escalator being a contractual rent increase written into the lease, which gives the owner built-in inflation protection whether or not a new tenant ever arrives. And the assets are valued richly by the market precisely because that cash flow is so dependable: individual US towers trade at 30 to 34 times cash flow, and tower REITs at 25 to 35 times AFFO. Those are high multiples for real estate, and they are the market pricing the reliability of the rent, which, as Section VIII shows, is also why the sector is so sensitive to interest rates.
To see the model at scale rather than on a single tower, look at American Tower’s full year 2024: total revenue of $10,127m, up 1.1 per cent; attributable AFFO per share of $10.54, up about 7 per cent; and organic tenant billings growth of 5 per cent. Notice the gap between revenue growth of 1 per cent and AFFO-per-share growth of 7 per cent. That gap is the escalators and the tenant stacking at work across a whole portfolio, with modest top-line growth converting into materially faster growth in the cash that reaches shareholders.
VIII. Macro sensitivity
A tower’s cash flow is stable, but its share price is not, because a tower is a long-duration, low-cap-rate asset, and those are the assets most sensitive to interest rates. That sensitivity explains the sector’s worst stretch and the opportunity that came out of it.
The mechanism is straightforward. A tower is prized for a stream of contracted, escalating rent stretching far into the future. When interest rates rise, the value today of that far-off future income falls, and the further out the cash flow, the harder it is discounted. Long-duration assets therefore fall hardest when rates rise, and rise most when rates fall. From January 2022 to October 2023, real interest rates rose 322 basis points, and tower REITs returned −43 per cent over that window. The towers were carrying more traffic and earning more rent the entire time. The asset itself was fine, and rates alone cut its price nearly in half.
That reset shows up in the valuation multiples. Tower REITs entered 2022 at 34.4 times AFFO, against 27.5 times for the average REIT, priced as premium growth assets. By March 2023 they traded at roughly a 10 per cent discount to net asset value, with implied cap rates reset to about 5 per cent. Crown Castle’s dividend yield reached an all-time high of 6.3 per cent in August 2023, against a pre-2022 high of 4.4 per cent, a yield that high on a quality asset being the market shouting that the price has fallen far, whether from fear or opportunity.
Here is how the asset behaves across four rate-and-growth regimes.
| Regime | Rates | Growth | What happens to towers |
|---|---|---|---|
| Falling rates, healthy growth | Down | Up | Best case. Long-duration assets re-rate up; escalators and tenant stacking still compound. Multiples expand. |
| Rising rates, healthy growth | Up | Up | The 2022-23 case. Cash flow keeps rising but prices fall hard, −43 per cent, as discount rates dominate. Painful for holders, an entry point for buyers. |
| Falling rates, weak growth | Down | Down | Supportive. The rate tailwind helps valuations even as new-tenant additions slow; escalators still protect the base rent. |
| Rising rates, weak growth | Up | Down | Worst case. Discount rates rise while tenant additions stall, the double squeeze. Escalators are the floor that stops it becoming a rout. |
The pattern to hold onto is that the towers’ cash flow barely flinched through the worst of it, while the price moved violently. For a long-term owner that gap is either the risk or the opportunity depending entirely on the price paid going in. Rates move the share price a great deal in the short term and do very little to the rent over the long term.
IX. Tax
This section is general, jurisdiction-neutral, and explicitly not tax advice. Tax on any tower investment depends on where you live, what account you hold it in, and rules that change, so treat the following as structure, not instruction, and take advice specific to your own situation.
The structural fact that shapes the tax treatment is that the major US tower operators are organised as REITs. A REIT is a tax structure as much as a business: by design it distributes the bulk of its taxable income to shareholders and is generally not taxed at the entity level on the income it distributes, per Nareit. The practical consequence for an investor is that the return arrives largely as ongoing distributions rather than as profit retained and compounded inside the company. You are, in effect, paid out along the way rather than left to compound untaxed within the wrapper. Many of the non-US operators, from Cellnex to the Indian and African towercos, sit in different corporate and tax structures again, which is one more reason the wrapper matters.
That shapes how a tower holding is likely to sit in a portfolio. Because the return comes mostly as a distribution stream, the investor-level tax treatment of those distributions matters a great deal, and it varies widely by jurisdiction and by account type. In some jurisdictions REIT distributions are taxed differently from ordinary company dividends; in some tax-sheltered accounts the distribution can be received far more efficiently than in a taxable one. Cross-border holders face a further layer: withholding tax on distributions from a company domiciled in another country, which may or may not be recoverable under a treaty. A UK investor buying a US REIT, or a European buying an Indian towerco, meets exactly this layer.
None of this is a reason for or against the asset. It is a reason to decide which wrapper you hold it in before you buy, because with a distribution-heavy asset the account you choose can matter as much to your after-tax return as the price you pay. In short, the return arrives largely as ongoing distributions rather than retained compounding, and how those distributions are taxed is a question only your own jurisdiction and adviser can answer.
X. Case studies
Three real transactions, two disciplined and one destructive, show what happens when tower assets change hands and capital gets allocated.
The win: Vantage Towers, November 2022. Vodafone sold co-control of its European tower arm, Vantage Towers, to a consortium of KKR and Global Infrastructure Partners. The deal put an equity valuation of €16.2bn on the business, at €32.00 a share, a 19 per cent premium, equating to 26 times FY22 EBITDAaL, with Vodafone able to net up to €7.1bn in cash. (“EBITDAaL” is earnings before interest, tax, depreciation and amortisation after lease costs, a common tower-sector profit measure.) What the deal shows is the private-infrastructure premium. When a disciplined operator sells a quality tower portfolio, well-capitalised infrastructure buyers pay multiples the public market rarely offers, which is why listed tower valuations tend to hold a floor.
The discipline: American Tower exits India, closed 12 September 2024. American Tower sold 100 per cent of its India operations to Brookfield’s Data Infrastructure Trust for about $2.5bn, but booked a $1.2bn loss on the sale, most of it a $1.1bn cumulative-translation reclassification, the accounting recognition of years of currency erosion. India had the towers and the growth, but it also had weaker-credit tenants and a currency that ground down a dollar investor’s return. This is the emerging-market trap made concrete. The towers performed, and the investment still lost money, because the tenant and the currency were the risk, not the steel. Walking away was the disciplined move.
The cautionary tale: Crown Castle’s fibre write-down, March 2025. This is the one to sit with, because the damage was entirely self-inflicted. Crown Castle agreed to sell its whole fibre and small-cell segment to EQT and Zayo for $8.5bn, having invested more than $17bn in that business, plus an expected loss of about $900m in FY2025 on classifying it as held-for-sale. More than $17bn went in and $8.5bn came out. That is billions of shareholder capital destroyed, not by a downturn in towers but by a management decision to chase a “digital infrastructure” adjacency that did not compound the way the tower does. It makes the point that keeps recurring across this asset class: the towers are not the risk. Capital allocation is. A great asset in the hands of managers who redeploy its cash flow poorly can still be a poor investment.
The pattern across the three is consistent. The value in this sector is in the tower and its stacked, escalating rent. The losses came from currency (India) and from management straying off the tower into a business with worse economics (Crown Castle’s fibre). The wins came from disciplined ownership and disciplined exit. The asset itself behaves, and the risk sits in the decisions made around it.
XI. The core constraint
For data centres, the one thing that governs whether an investment works is electrical power. For towers it is not power, or land, or even demand, but capital allocation.
That is an unusual answer, so it is worth defending. Demand is not the constraint, because traffic is set to more than double by 2031 and the densification is structural. The asset is not the constraint either, because a tower with three tenants earns about 24 per cent on invested capital on contracted, escalating rent, which is as good as infrastructure economics get. What varies, wildly, is what the people running the tower companies do with the enormous, dependable cash flow those towers throw off.
The evidence is in Section X. American Tower’s India loss and Crown Castle’s fibre write-down were not tower problems; they were decisions to deploy tower cash flow into a weak currency and a weak-economics adjacency respectively. The $17bn invested for an $8.5bn sale is the constraint made visible. The same cash flow that compounds beautifully when reinvested into more towers, or returned to shareholders, can be set on fire when reinvested into the wrong thing.
This reframes what an investor is actually evaluating. Buying a tower REIT is only partly a bet on towers. It is at least as much a bet on the capital discipline of the people running it: whether they add tenants to existing towers, buy well-priced portfolios, return excess cash, and resist the temptation to become something other than a tower company. The asset is close to as good as real estate gets. The constraint is the human judgement sitting on top of it.
XII. Inside the asset
Step back from the share price and look at what you actually own when you own a tower, because the physical and contractual details are where the durability comes from.
A tower is, physically, a very simple thing: a steel structure, a foundation, a land lease or freehold beneath it, a power connection, and a set of equipment cabinets at the base. It has no moving parts to speak of, no rapid obsolescence, and a working life measured in decades. It does not care which generation of mobile technology runs on it; 4G, 5G and whatever follows all hang antennas on the same steel. That is the first source of durability: the asset does not become obsolete when the technology on it does.
The contract is where the economics live. Each tenant signs a lease of 5 to 10 years with an annual escalator of about 3 per cent. Two things follow. First, the revenue is contracted and long-dated, which is why the market values it so highly. Second, the escalator means the rent rises every year whether or not a new tenant ever arrives, a built-in contractual growth rate on top of whatever tenant stacking adds. The base case for a tower is that its rent grows about 3 per cent a year forever; the upside case is that a new tenant arrives and adds an increment that is almost pure margin.
The switching costs cement it. Once an operator has installed its antennas and equipment on a tower, tuned its network around that site, and integrated it into its coverage map, moving to a different tower is expensive and disruptive. Tenants are sticky not out of loyalty but out of cost. That stickiness is why tower leases renew, why the escalators hold, and why the cash flow that so worried nobody through the 2022-23 rate shock kept growing while the share price fell 43 per cent.
Put the pieces together, an asset that does not obsolete, a lease that escalates, and a tenant that cannot easily leave, and you have the anatomy of an infrastructure asset. The steel is the dull part; the contract is where the value actually sits.
XIII. The central dilemma
The thing that could break the thesis is specific: the number of tenants per tower could stop growing, or even shrink, and the whole model depends on it rising.
The mechanism of the risk is the reward’s mechanism running in reverse. Tenant stacking works because a second and third operator hang their kit on an existing tower. But in a mature market there are only so many operators, and when those operators merge, a tower can lose a tenant overnight, two networks becoming one meaning one of the two sets of antennas comes down. In the United States, carrier consolidation has already removed tenants from towers this way. A tower that fell from three tenants to two does not just lose one-third of its rent; because of the margin ladder, it slides back down the return curve from 24 per cent toward 13 per cent. The same arithmetic that makes stacking so powerful on the way up makes de-stacking painful on the way down.
There is a second edge to the dilemma. The operators, the tenants, would rather not pay ever-rising rent forever, and they have alternatives they can lean on: building their own sites, using new radio technologies that need fewer traditional towers, or negotiating harder at renewal. The tower company’s pricing power rests on the tenant having no good alternative. Most of the time it does not. But the tension is permanent, and it is why the 3 per cent escalators are periodically renegotiated in large deals rather than being an untouchable law of nature.
The dilemma comes down to this. The demand for data is overwhelming and rising, with traffic more than doubling by 2031. But the demand for tenants, for distinct operators willing to pay for space on distinct towers, depends on there being enough independent operators, which consolidation slowly erodes. The bytes are close to guaranteed; the tenants are not. An investor has to believe that data growth and densification add tenants and drive rents faster than consolidation removes them. In mature markets that is a genuine question. In growing markets, where operators are still expanding coverage, it is much less of one. Where you buy determines how heavily this dilemma weighs.
XIV. The next frontier
The tower thesis today is mostly a densification-and-5G story. What comes after it points in two directions at once, one expansionary and one threatening.
The expansionary direction is the continued build-out of the network into ever-smaller cells. As 5G matures and 6G eventually arrives, the shorter ranges and higher frequencies of newer radio technology demand more sites, not fewer; the densification that lifted towers-per-1,000-subscribers 12 per cent between 2021 and 2024 is a preview, not a peak. Fixed wireless access widens it further: delivering home broadband over a mobile network, which helps push the total-traffic forecast to 515 exabytes a month by 2031, loads the same towers even harder. More traffic, shorter ranges and new use cases all point toward more demand for elevated radio real estate.
The threatening direction is the “digital infrastructure” temptation, and this is where the frontier turns into a trap the sector has already fallen into. As towers matured, operators reached for adjacent growth: small cells, fibre, edge computing, distributed antenna systems. Some of these genuinely complement the tower. But they do not share the tower’s economics; they are more capital-intensive, more competitive, and lower-margin. Crown Castle’s $17bn-in, $8.5bn-out fibre experience is the frontier’s cautionary marker: the pull toward “the next thing” destroyed billions precisely because the next thing did not compound like the tower.
The other genuine frontier is geographic. The emerging markets, India at 37 gigabytes per phone heading to 70, Africa and Latin America still building basic coverage, are where the raw tower growth lives, and where the listed pure-plays such as Helios Towers and IHS Holding give an ordinary investor direct access. But as American Tower’s India exit showed, the frontier’s growth comes bundled with currency and tenant-credit risk that can consume the return.
Read across both, and the boring core, more tenants on more towers in stable markets, is the durable opportunity, while both exciting frontiers, new adjacencies and new geographies, carry the risks that have already cost the sector most of its real money. For a disciplined tower investor, the smart move may be to resist the frontier altogether and keep buying the boring core.
XV. Lessons from history
Towers are young as an asset class, but they have already run through a full cycle, and the lessons are unusually clear.
The cash flow matters more than the price swings. Through the 2022-23 rate shock, tower REITs fell 43 per cent while their rent kept rising on 3 per cent escalators and growing tenant billings. Anyone who confused the falling price with a falling business sold the asset at exactly the wrong time. The lesson is to watch the rent rather than the ticker, and to understand that the two can move in opposite directions for a year or more.
A premium multiple is a warning, not only a compliment. Towers entered 2022 at 34.4 times AFFO, priced for everything to go right. A high multiple is the market saying an asset is wonderful, but it also loads in a long fall if rates or growth disappoint, and both were true here: the business was excellent and the price still halved. Paying up for quality can be defensible; paying any price at all for it is not.
The losses come from wandering off the tower. The sector’s two biggest recent wounds, American Tower’s $1.2bn India loss and Crown Castle’s fibre destruction, came from geography and adjacency, not from the core business. The danger in a great asset class is rarely the asset itself; it is the temptation to become more than the asset.
The private market keeps setting a floor. Even through the public sell-off, private infrastructure capital kept paying up for towers; the Vantage Towers deal at 26 times EBITDAaL is the marker. When public prices fell to a discount to net asset value, the gap between what the public and private markets would pay was itself a signal, that the public price had detached from what the assets were worth to a long-term owner.
The through-line of every lesson is the same one. This is a superb underlying asset that the market prices violently and that management can damage from the inside. History rewards the investor who separates those three things.
XVI. The case for it
The bull case for cell towers is unusual in that every part of it can be pinned to a figure that already exists rather than a forecast anyone has to trust.
A demand curve that is hard to bend. Mobile data traffic is forecast to rise from 146 to 328 exabytes a month by 2031, with 5G’s share of it climbing toward 85 per cent. This is not a speculative use case waiting to appear; it is the observed, funded, already-happening growth of the thing every tower carries. American Tower’s chief executive frames it as a required doubling of wireless network capacity by 2030.
Economics that compound with almost no extra capital. The tenant-stacking ladder, 3 per cent ROI at one tenant, 24 per cent at three on an unchanged cost base, is a rare feature in real estate: growth that requires almost no new investment. Add the 3 per cent annual escalators and the base rent rises contractually even before a new tenant arrives.
Barriers to entry that protect the incumbent. You cannot easily build a competing tower next to an existing one; planning, land rights, and the fact that operators prefer to share a site all favour the tower already standing. The 4.4 million towers already in the ground, with the top 100 companies running 1.7 million shared sites, are an installed base a newcomer cannot replicate.
A private market that keeps validating the value. When operators sell tower portfolios, infrastructure funds pay premium multiples, 26 times EBITDAaL for Vantage Towers, which puts a floor under listed valuations and confirms that patient, professional capital wants this asset.
Taken together, that is a genuinely scarce, obsolescence-resistant asset, carrying overwhelming and rising demand, with economics that compound cheaply and a private bid underneath the price.
XVII. The risks
Two of the risks here have already cost the sector billions, and neither of them is the tower itself.
Interest-rate sensitivity. This is the biggest short-term risk. Towers are long-duration, low-cap-rate assets, and they fell 43 per cent when real rates rose 322 basis points in 2022-23, even as the underlying business kept growing. An investor who cannot stomach the price of the asset halving while its earnings rise should not own it in a rising-rate environment.
Capital allocation, the self-inflicted risk. The sector’s worst losses were decisions, not markets. Crown Castle put more than $17bn into fibre and is selling it for $8.5bn. No downturn did that; management did. Because tower companies generate so much cash, the risk that they redeploy it poorly is permanent, and it is the risk that recurs more than any other across the sector’s history.
Emerging-market and currency risk. The fastest growth sits in the riskiest currencies. American Tower’s India exit cost $1.2bn, most of it currency translation; the towers performed and the dollars still shrank. Reaching for emerging-market growth through the Indian, African or Latin American pure-plays means accepting that the currency can undo the operating result.
Tenant consolidation. In mature markets, carrier mergers can strip a tenant from a tower, sliding it back down the margin ladder from 24 per cent toward 13 per cent. The margin ladder that rewards stacking punishes de-stacking just as sharply.
Valuation. The assets trade at 25 to 35 times AFFO, rich multiples that leave little margin for error. Buy at the top of that range and a good business can still be a poor investment; the 34.4 times entry point into 2022 is the proof.
The risks are not evenly matched to the rewards. The demand side is about as solid as any thesis gets. The vulnerability is on the price side, through rates and valuation in the short term, and on the human side, through capital allocation and consolidation over the long term. It is rarely the towers that let an investor down; it is the rates and the managers.
XVIII. The Alternative Fortune verdict
Cell towers are, at the asset level, close to as good as real estate gets: a scarce, obsolescence-resistant structure carrying demand that is forecast to more than double by 2031, earning contracted rent that escalates about 3 per cent a year and compounds toward 24 per cent returns as tenants stack with almost no new capital. That is the case, and it is a strong one.
Two facts have to sit alongside that case. First, the price of this excellent asset is violently rate-sensitive; it fell 43 per cent in the last rate shock while the business grew, so the entry price and the rate environment matter enormously to the outcome. Second, and more important, the sector’s biggest losses have been self-inflicted, from the $1.2bn walking out of India to billions more in Crown Castle’s fibre misadventure. This is a wonderful asset that the market misprices and that management can wound from the inside.
Where the edge actually is. The edge is not in discovering the demand, since everyone can read the same traffic forecast. It lies in two things the market undervalues. The first is temperament: being willing to own the asset while its price falls and its rent rises, because that gap between price and business is where towers are bought well. The second is discrimination on capital allocation, favouring the operators and regions where management sticks to towers, stacks tenants, and returns cash, over those chasing adjacencies or emerging-market growth at any currency risk. The asset itself is not the variable; the price paid and the people running it are.
Questions to ask, by vehicle:
- US tower REIT (AMT, CCI, SBAC): What is the tenancy ratio and is it rising or falling? What is management doing with the cash flow: more towers, buybacks, dividends, or a new adjacency? What AFFO multiple am I paying relative to the sector’s 25-35x range, and what does that imply if rates rise?
- Non-US pure-play (Cellnex, Indus, Helios, IHS): How creditworthy are the tenant operators, and how concentrated? What currency am I really taking on, euro, rupee, sterling-quoted-but-African-underlying, and could it erode the operating return the way the rupee did for American Tower? Is this a growth story (India, Africa) or a deleveraging one (Europe)?
- Infrastructure ETF (VPN in the US, INFR in London): What share of the basket is actually towers versus data centres, utilities and hardware? Am I comfortable paying 0.50 to 0.70 per cent a year for a diversified infrastructure bet rather than a pure-tower one?
- Private infrastructure fund: What multiple is the fund paying versus the 26× struck for Vantage Towers? What are the lock-up, the fees, and the leverage on the portfolio?
The tower is one of the better infrastructure assets an ordinary investor can own, in dollars, euros, sterling or rupees, but you are buying it at a price the market swings hard, from managers whose capital discipline varies wildly, so the return lives less in the steel than in what you pay for it and who is looking after it.
This deep dive sits under Alternative Fortune’s real estate & infrastructure category.