You would think it would be easy to count telecom towers, right? Big lump of concrete with a hundred feet of steel sticking out of it – fairly conspicuous, right? Given the blood, sweat and tears invested to navigate one of the world’s most notorious permitting regimes, not to mention the investment of US$100,000 or so, you’d imagine we’d know how many towers were in Brazil, right? Yet asset registers in Brazil are as inaccurate as anywhere and, try as we might, TowerXchange just cannot make a Brazilian tower count add up to the 70,000 towers everyone seems to think are in the country.
I thought it might be fun to share with our readers the challenge of publishing an accurate tower count, whilst at the same time shedding what light we can on one of the most investible tower markets in the world: Brazil.
Our tower counts are based on dozens of telephone and face to face conversations with the most knowledgeable people in Brazilian towers. Based on the latest figures in the public domain, we wouldn’t know much about Brazil. But we overlay multiple informed stakeholders’ opinions and estimates of the structure and size of the tower market, we apply some weighting, and come up with an informed estimate. The simple fact is that no-body really knows how many towers there are in Brazil, but aggregating and averaging informed estimates gives us a good insight.
I should point our that it is extremely difficult to separate macro towers from rooftop structures, but what really matters for this analysis is the number of shareable structures. So if a few robust rooftops get into the calculation, so be it.
What we DO know about Brazil’s towers
We know American Tower has been buying a lot of Brazilian towers. So has SBA Communications. As publicly listed companies, they’re pretty transparent – as long as you don’t mind your data being up to date only to the last published quarterly results, where you can find their tower counts in every market.
American Tower’s recent acquisitions of BR Towers and from TIM brings their count to 18,851 towers in Brazil – some of which are in the process of being transferred, others of which they’ve operated and marketed since as long ago as 2002. SBA Communications own around 7,000 towers and rooftops in Brazil, topped up by their most recent acquisition from Oi.
What we DON’T know about Brazil’s towers
AMT and SBA are where the transparency ends in Brazil. There are a host of private tower companies, headed by Grupo Torresur who, when they last told us their tower count this time last year, had a little over 6,000. Our sources suggest GTS’s count is now closer to 6,185.
Then there’s Brazil’s host of smaller independent developer towercos, what we call middle market towercos because for most, their raison d’etre is to be acquired by AMT or SBA! T4U and QMC each have around 500 towers and rooftops. Cell Site Solutions (CSS) has approximately 350, Brazil Tower Company (BTC) 300, Highline do Brasil 200, Z Sites 200, Rede Sul 175, Centennial 100 sites, Torre Online 51, Telecom Torres and Skysites around 40 each. Phoenix Tower International are in Brazil, but their BTS sites are most likely still works in progress at time of press. In addition, Brazil has around a dozen small regional towercos each with less than 20 towers.
Counting a build-to-suit centric towercos’ towers in Brazil is a challenging exercise – some are relatively dormant, others are building as quickly as they can get sites permitted, so it’s a movable feast. However, we’re confident we’ve identified and quantified the assets at most of Brazil’s middle market towercos. Brazil’s independent developers account for a total of ~2,600 towers, plus GTS’s 6,185.
This gives us an aggregated estimate of 34,636 towerco-owned sites in Brazil, broken down in figure 1.
Figure 1: Estimated towers owned by towercos in Brazil
The value of Brazil’s 2,600 independent developer towers
Those 2,600 independent developer-owned sites might look like a relatively insignificant slice of the market, but consider this; the most active and aggressive companies in this category are growing their tower counts by 100% or more per year. Brazil’s independent developers deliver more than half the build to suit programmes in the country. Their towers are newer than the carriers’. They are designed for multiple tenants (usually three or four) from the outset. They are being aggressively leased up – tenancy ratios of 1.3 upward are not uncommon already, but they may not come to market until tenancy ratios are nearer two than one. Smart independent developers ensure their leases are readily transferrable, and ensure their site maps don’t overlap with AMT and SBA.
In summary, many of those 2,600 independent developer towers in Brazil are built in unique locations, and they are built to sell. If their tenancy ratio gets nearer two than one, they will attract valuations well above the $200k+ cost per tower that is the current norm for an operator-captive tower.
How many operator-captive towers are left in Brazil?
If you thought counting towerco-owned towers was complicated, try counting those retained by carriers! As a legacy of the era when networks were a genuine competitive differentiator, carriers worldwide are reluctant to share tower counts. However, we can count the towers each has sold since 2010 – see figure 2.
Figure 2: Announced number of towers sold by Brazilian carriers since 2010
Figure 3: Brazil cell-site growth
Brazil’s other leading three carriers have divested a significant proportion of their towers, retaining only a few sites considered too strategic, plus a modest inventory to fulfill swap rights where contractually necessary (we’ll explain swap rights later).
Vivo reportedly has sold around 80% of their towers, with the majority of those remaining considered too strategic to sell, plus a few to cover swaps. Again, our sources give us a range of estimates of Vivo’s remaining captive towers, the mid point of which is 1,750.
TIM had around 8,200 towers and rooftops prior to the recent sale of 6,480 to American Tower, leaving TIM retaining ~1,720 for operational and legal reasons.
Struggling #4 carrier Oi has already sold over 11,000 towers, potentially leaving them with just ~1,000 towers to cover swaps. Note that Oi has a 4G infrastructure sharing agreement with TIM. Therefore any prospective consolidation affecting the two companies would have a reduced impact on tower tenancies.
Nextel sold the majority of their towers in the deals that sent 4,730 towers to AMT. TowerXchange have been unable to substantiate rumors that the potential acquisition of a further 900 towers from Nextel was mentioned in the Q3 2014 analyst conference call – there’s no mention of it in the call transcript. Such a deal is yet to be consummated so we’ll treat it with a pinch of salt. Nextel may have had a few sites built since, so we’ll cautiously estimate 500 towers on their balance sheet.
Then there’s a long tail of smaller and non-traditional operators that will have built and co-located on towers, including Sky Brasil, Algar Telecom, Sercomtel and ON Telecom. We estimate they have 500 towers and rooftops between them.
Brazil’s build to suit market
Capacity and permitting constraints are limiting Brazil’s carriers to adding ~5,000 tower per annum, almost exclusively executed by towercos through build to suit contracts. SBA Communications seem to be building as fast as anyone, adding around 250 sites per annum compared to a little over 100 each at GTS and AMT. However, some of the build to suit-focused independent developers may be building even faster still.
Why is there such a discrepancy between the 70,000 towers people think there are in Brazil and the 48,606 this analysis reveals?
There are several possible explanations for the discrepancy between the number of towers our analysis can ascribe in Brazil, and the 70,000 tower count most commentators referenced, ourselves included, until now.
1. Anatel’s data is widely available, and it suggests there are over 70,000 cell sites in Brazil. However, as previously mentioned, cell-sites include co-locations and ‘special structures’, and thus produces a significantly higher number than a pure count of shareable structures.
2. Perhaps we have yet to identify a few more middle market towercos in Brazil – in the course of this research, we identified four prev