Africa Mobile Networks’ plan to reach 20,000 sites

Africa Mobile Networks (AMN) will have 2,000 sites live by the middle of 2020, most in rural areas that nobody else was willing or able to serve. By vertically integrating passive and active telecoms infrastructure, AMN are able to drive down costs to a remarkable degree. Innovative revenue share models have enabled them to scale to several hundred sites over the last four years, and a move to add opex-only models will enable them to scale further. TowerXchange sat down with CEO Mike Darcy to discuss the details of the model and AMN’s ambitions to reach 20,000 sites next decade, and to transform rural coverage.

TowerXchange: We last interviewed you four years ago, shortly after you’d broken ground on your first sites. You are now on your way to 2,000 sites by mid-2020. Tell us about the growth trajectory of AMN over the last few years.

The plan is basically the same, in terms of big picture. Our big breakthrough was in 2015, when we scored our first contract with Orange, we now have a second and are working with Orange in DRC and Cameroon, and a third is imminent. We also signed a framework agreement with MTN and are now working with them in Cameroon, Zambia, Guinea Conakry, Guinea Bissau, Nigeria, Cameroon, Sudan, South Sudan, and Liberia. We have nearly 1,000 sites live, and we’re fully funded up to 2,000 sites which we plan to hit by mid-2020.

TowerXchange: MTN recently closed an ambitious RFP for rural roll-out, what is your relationship with MTN and where are you working together? How does your bid differ from others that won?

This isn’t the first time that MTN have issued an RFP around rural connectivity. We succeeded last time and were the most successful of the vendors selected. We were the only ones with the appetite and solution which could deliver the thousands of sites MTN wanted. 

Rural connectivity has become an even larger strategic concern for MTN so they opened the rapid rural rollout RFP and encouraged us to apply again. The larger firms in the sector were either unsuccessful or did not bid at all, so you have seen smaller players being awarded alongside Africa Mobile Networks. 

The way AMN has reached scale was through a commercial model based on a revenue share with operators. This model has proven really successful in allowing us to expand to sites we think are commercially viable and pitch to mobile operators. We offer mobile operators the chance to expand their networks without capex or opex risk. It is AMN that builds, operates and maintains the sites and connects them to the operator’s core network. So we offer a full turnkey service in return for a share of revenues they wouldn’t otherwise generate.

How we will operate in the future is that we are moving to sign contracts which are on an opex-model, at locations specified by the operator, and then combined with additional revenue share sites in locations identified by AMN. The opex model moves us closer to a conventional towerco model, in which operator is taking the operational risk and pay AMN a fixed fee per month, rather than a revenue share. 

The reason we are switching towards an opex model over a revenue share model is to be able to service more areas. For AMN to take on a site it must meet strict criteria. Each operator will have list of rural sites they want to build, and not all will meet our criteria. So we can still take on the capex risk of erecting a new site, but if it is not in one of our preferred areas we will only proceed on an opex-basis with an operator. 

We submitted a bid to MTN based around this new and improved opex model. We don’t know if AMN and MTN will agree another framework agreement, but it has gone down well with most opcos. 

We are now active in our first ten countries, and so far AMN has taken all the risk. I think we are sufficiently well established now that we can begin to do things differently through an opex model. We have shown that these sites are viable. Now we are moving forward with the first opex contracts with a minimum of 120 sites on an opex basis and then we can layer more revenue share sites on top, where we think they are viable. 

TowerXchange: MTN is you principal customer, what can you tell us about the other operators you currently work with, and plan to work with in the future?

We don’t want to be beholden to one operator but at the moment one operator is responsible for 70% or so of our business, so we would like to diversify this more. We began by working with Orange in 2015 and we are working to add more countries with that operator. We are also working on adding other tier-1 operators, especially in Nigeria, which is our most important market in sub-Saharan Africa.

TowerXchange: Do you see AMN’s proposition as competitive or complementary to the Energy Services Company (ESCO) model?

We are a compliment to MNOs’ existing strategy of outsourcing tower power management to ESCOs, but there’s no overlap between the propositions. 

TowerXchange: What areas are you covering? What makes them uncoverable by normal mobile network roll-outs?

AMN are at the top of our game, for what we do, but there are many, many other players who would like to do what we do. Where AMN have differentiated is in our ability to drive the capex out of investment in base station and minimise opex costs. We simply have a lower cost base than anyone else. And we continue to innovate. So I don’t think the sites we cover are uncoverable.

We have tools which allow us to survey and process a whole country and identify all human settlements from satellite imagery. We also have a database of existing towers. We combine the two resources and find out where the people are covered and filter in those which are more than 10km away from a tower with a reasonable population. 

Nigeria is interesting. It has 69,000 settlements and 50,000 of those have less than 500 people. You cannot make money out of 500 people as that will only translates into 150 subscribers. But once we go through our exercise we identify 4-5,000 candidate sites which are big enough and far enough away from existing sites. 

We then do physical site surveys and those sites may pass or fail. If they pass, we begin RF planning, predicted coverage et cetera. Then we begin site acquisition and build. 

I don’t know why someone else isn’t there already, but they are not. We have just started our volume roll out in Nigeria, and have 501 successful site surveys which are at various stages of site acquisition. That will cover a population of just under 2mn in settlements of 1,000-5,000 people, and an average population of 3,700 under site. The ARPUs at these sites will range from US$1.20-US$5. 

TowerXchange: What in-country capabilities have you had to develop to service these sites for your customers?

It costs us roughly US$1mn to get started in each new country – to set-up a local operations, buy the vehicles, agree VSAT access, install our equipment in our client’s core networks - before anything else is live. We build base stations in increments of 60 because that is how many fit in a 40ft container coming from Shanghai. 

We buy all our heavy components from China; steel, solar, batteries, cabinets, antennae, VSAT. Our BTS are not put together in China, we pay factory door prices, ship to Africa and assemble there. The lattice tower is designed ourselves and we order the steel from factories at US$800 or so per metric ton, and again we build it ourselves using our own guys. Outsourcing is expensive and we can’t control quality or timing, so we are 100% vertically integrated. 

We are active in ten countries and all our people are local. We don’t use expats. Let’s take Nigeria as an example, where we just started volume roll out. I visited the office and warehouse in October. We have vehicles to deliver and equipment for close to 400 sites already there. We have teams of engineers. Everything is there to go in a cookie cutter process to get our sites up quite quickly. We only employ twelve people. Seven to eight field engineers, some admin staff and a country manager. 

One of our field teams can install about 15 sites in a month, including travel. Once our team arrives on site, the service can come online within 2-3 hours. So it is a really rapid service.

We put a huge amount of effort into remote network management because our sites are so inaccessible. We need that ability to manage and fix sites remotely and all our sites are very much designed with this in mind. Our opex is focused on minimising truck rolls. AMN have built a network monitoring tool in-house which updates every one minute with the status of every site; battery status, temperature, traffic, VSAT link, package loss. We can reboot remotely if necessary, and we can access the CCTV remotely too, to keep an eye on the site. 

Our in country teams visit every site between every 2-3 years. We are currently running at a 27 month average between visits to a site, and that could go up as our sites age. We can fix a lot of issues remotely, but a hardware failure necessitates a site visit.

TowerXchange: Is there a typical AMN site typology? 

We have now standardised on two different site types. We do have some 40m towers in network, but we have now standardised on something smaller. 

Our Monopole is 9m to the bottom of the antennas, and 12m to top. I think you’d call it a 9.5m tower. The monopole design means it can be built in a day, at very low capex. It can support 2G or 2G/3G, but has a limited capacity. 70% of our sites can be served with a monopole. Our sites tend to be isolated from the main network, but there are some areas where two or more towers overlap. 

Our lattice tower is 20m tall. These take a week to erect, and we are deploying those with three sector antennas. Because they’re bigger they can take a lot more equipment, so they could become a multi-tenant site. 

We upgrade our monopoles to these 20m lattice sites when a revenue share site proves itself to be high volume. And for our opex-model, as described earlier, we deploy our larger tower more often. If operators are paying a fixed opex charge, they tend to want a standard, larger tower.

TowerXchange: What is the energy set up at AMN sites? Do you have a one-size-fits-all solution, or do you have a few different flavours of solution depending on the site?

All our sites are solar+battery with uptime of 99.8%+ around the whole year. 

We have moved from lead acid to lithium ion. Lead acid batteries are heavier and have to be on the ground, switching to lithium allows us to move the battery into the cabinet on the monopole. If a site is working we leave it, but as the lead acid battery reaches its end of life we switch to lithium ion. 

We hit 99.8% by sizing the site correctly in order to achieve the right level of availability. We receive data from 1,000 sites across Africa minute by minute, so we have the data to do things really well. Some places are more demanding, like Guinea Conakry or Cameroon in the rainy season, but we can still maintain availability targets.

Unlike other operators, we do not have a serious theft problem. We still install our Solar panels high up and out of reach. Occasionally ground based batteries are taken, but it is not a major operational headache. I think a major difference is that where AMN are going, there is no other connectivity and it is very precious to the people there. It is difficult to overstate how important it is. This means people don’t want to steal from us and the local community polices it. We have a caretaker and security guard at every site who helps keeps the solar panels clean and generally keeps an eye on the tower, but they are not there to fight off intruders or put themselves in harm’s way. 

TowerXchange: Does AMN exclusively use VSAT for backhaul, or are some of your sites able to use microwave?

All our sites so far use VSAT for backhaul. We want our offering to be ubiquitous and not reliant only on microwave backhaul. Intelsat are an investor and so we use them for VSAT backhaul, but we also work with other satellite operators too. But now we are adding a MW option also, for cases where a line-of-sight connection to a backbone POP tower exists. 

TowerXchange: Do you have plans to follow Helios Towers and IPO, or need to raise further capital in other ways?

We are continually raising capital, as both equity and debt. We have invested over US$50mn so far and as I said, we are fully funded up to the 2,000 sites we have planned up to mid-2020. We are planning for another 3,000 sites by sometime in 2021. But we won’t need to raise as much as the US$50mn we have so far to fund those sites. All our fundraising is done at group level and allocated to which ever project requires it. But the fundraising at group level is driven by the contracts we sign with the opcos in country.

TowerXchange: What are the next steps for AMN? 

Building, building, building! We plan to reach 20,000 towers across all of Africa and to be present in 30+ countries. And we think we can supplant a lot of those active in Africa already. With our opex model and 20m towers, operators now appreciate the benefits of our turnkey solution, so we can disrupt and replace what they pay to a towerco, to an OEM, and to their O&M provider. 

Mobile operators have realised that they have loss making towers which can be replaced with an AMN tower. Many towerco contracts have termination clauses which enable the expiration of leases on 3-4% of sites per year without penalty. So operators are replacing big towers with AMN towers, not the other way round. We are providing service at an appropriate level at a much lower cost.

Take our monopole sites on our new opex model. We charge US$690 for a ten year lease, but we also offer a 50% discount for the first two years. After five years the lease rates starts to reduce. For a 20m lattice, prices start at $1,200 with the same discount and tapering structure after five years. And 3G and 4G can be added also. This is for passive and active infrastructure at prices which are highly competitive.  

AMN remains profitable; our EBITDA margins are around 50%, and our business is very sustainable. Our revenues per tower are much lower than for a conventional towerco, which is why we are targeting 20,000 towers, because at those volumes the economics start to look very attractive. So we are not worried about other operators such as towercos moving in and taking over our rural sites once they become more developed and conventionally viable. We think Africa’s other towercos should be worried about us replacing them!