In a recent feature, I explored how operators are reducing their capex by turning to alternatives for RAN, including Infrastructure as a Service and Virtual RAN.
This was in part inspired by South African operator Cell C’s switch to a fully ‘outsourced’ network model; in the process of researching the model and how it might be viable in South Africa’s market, I spoke to Mark Walker, VP of Sub-Saharan Africa at IDC.
While we began by discussing Cell C’s shift to a new model, as the conversation progressed we looked more broadly at how operators in Africa in general and South Africa in particular must constantly adapt, appraising their value proposition to remain relevant.
Why has Cell C switched to this VRAN model now?
Cell C was third after MTN and Vodacom into that space. Their prior performance has been based on sharing the Vodacom and MTN infrastructure, an MVNO type model. They carried a lot of the other retail MVNOs, those players as well. In the last year, they've undergone some significant changes, new management. They got bought out. Why have they gone with this core network model? The sense that I get is they're focused. The MVNO thing, I think, was not really effective for them…there's also a lot of competition in the South African market between MTN Vodacom and Telkom. The market’s only so big, and South Africa is often used as a test case for the bigger international networks. You see what will work, what won't work, given various demographic and economic situations within South Africa, which is kind of contained, so you can see what happens. They've had to think: what is the operating cost? What is their value proposition to the market? How do they take this forward, going with a core network approach? I think it's still very early days.
Tower and leaseback agreements have been pretty commonplace in emerging markets for quite a while. Do you think we'll see more operators like Cell C not doing infrastructure anymore? Is operating a virtual RAN likely to become viable?
That makes sense to me - infrastructure is very much the operator play, because the expectation is cheaper and faster always, yet capex is going through the roof, and the return on that investment is not real. Every time they're investing in 5G, 6G, it’s a massive capital expenditure, and then that ends up with OPEX as well…the average revenue per user doesn't necessarily reflect an uptick because of that additional capex investment. So it might be a bit of a challenge there.
How viable are network/infrastructure sharing models?
To lay infrastructure, especially in Sub-Saharan Africa, is not trivial. There've been multiple projects in fibre…they’ve been reusing old copper for ages. For South Africa, fibre was first: the sequence of events was for very high density, commercial enterprise locations, geographic locations - city centres, big manufacturing hubs where there's a high demand for this. After that, the High Net Worth areas - gated communities, high density suburbs - and it's moved on now to the less affluent areas, even to the point of players rolling out into informal settlements now, not so much using fibre, but using wireless connectivity.
Competitively, the intensity’s higher and higher. Going forward, my feeling is that the only way for the virtual or fixed players is to share. Companies are divesting and selling off; Telkom are selling off their tower business, MTN is also looking at [whether] they want towers. Does this fall into the telcos’ strategic remit? Sharing is going to be part of it - the question is, what are we going to be sharing, and how? Towers is an easy one - why have dual areas? But then there are some regulatory issues that also hinder the play; for example, many of the gated communities have exclusive agreements with single suppliers. But sharing is going to be the name of the game by and large, unless the cost of deployment is low enough that it doesn't matter.
Infrastructure as a service relies on cloud first of all. Who plays in cloud the best? Data centres - so unless a telco owns a data centre or partners with the likes of Telehouse or Equinix, where does it leave them? What do they offer, especially if they don't own their data centre infrastructure? Equinix is not going to give up a chunk of their business because the telco says so. They'll have to find a mutual agreement.
People are discussing 6G now but the use cases are not very dissimilar to those touted for 5G that failed to materialise. What became clear during the 5G generation was that people were not willing to pay anymore for that upgrade - I'd imagine we'll be seeing that again with 6G. The leap is not going to be as significant as it was with 4G, which was enough for the basic consumer functions. ROI will be delivered from the enterprise side: how are operators dealing with that?
Telcos effectively are not getting their piece of the pie: they're providing all the infrastructure, all the connectivity services, the BSS and OSS services, and yet where is it reflecting on the EBITDA? Or even their bottom line? There's got to be a balance. There have been initiatives about paying your way - basically forcing players like Netflix and over the top (OTT) players to pay telcos for the use of the networks, which they are already doing inadvertently, and that’s where the money is coming from. Telco is in a tough place right now; there's massive client expectation, both consumer and enterprise, but the cost base is staying the same. If I look at the MTN results recently, they purported great growth over the last couple years, but slowing down very quickly. Ralph Mupita's saying big things, but internally, they’ll have to figure out about sharing. How do they share with the likes of Cell C in a in a virtual type environment, whether it’s infrastructure, network, or core - how does that actually work? How does it benefit both parties? Otherwise you're never going to have an agreement.
Shifting towards infrastructure sharing increases reliance on a smaller number of players. How does that affect the competitive landscape?
If you're sharing, that's between you and whoever you're sharing with, so you both stand to benefit if whoever's buying your services is willing to either buy more, or buy at a higher price. Those are the dimensions that you would look at before you enter into any sharing agreement. It's the same as the argument with AI at the moment: how much is AI driving productivity and efficiency? Efficiency is just producing more units in a shorter time. Will there come a time when there are units being produced more efficiently than the market can purchase? Then what - you're producing a lot more, a lot faster? Is there a price differential to increase demand? The cheaper the price, demand will pick up, theoretically - but will that happen? Sharing agreements have to be based on perceived demand for technological solutions and then worked backwards. It's not a case of “we'll build it, and they'll come”. It's a more risk-based approach, trying to assess that level of demand, which is also difficult at the moment because of AI. Theoretically everybody's going to need more bandwidth, more connectivity, more data centres - but are they willing to pay for it?
What’s demand like in Africa?
Over the next couple of years, Africa’s got a very young population with spending power that's increasing at a decent rate, and a lot of uncharted territory. There’s a lot of focus on data centre placement and connectivity. How do we roll it out? What kind of services are demanded by these players? What type of technology can service these players, given the constraints, either in physical geography or latency? That's going to be the deciding factor, how those questions are going to be answered. Hubs like South Africa are pretty developed. It's still the rural versus urban, there's a divide there. What is the impact of low Earth orbit satellite? That's a big challenge for them.
If you look at Vodacom for instance, I'm hoping that with Shameel Joosub having a seat on the Vodafone board directly, it's not intermediated by a local Vodafone representative that's tasked with looking after Africa - Shameel is from Vodacom, and he knows what's going on. My sense is that with that direct voice, there will be a better focus on what is going on in the African markets, [whether] South Africa [or] Safaricom in Kenya. These are two big players.
We haven't looked at Orange, or Airtel yet...Etisalat’s coming in from the Middle East into West Africa. The thing there, is going to be speed to market. We know the mobile phone stats, smartphone versus feature phone - that's happening very quickly, with producers like Tecno coming up with sophisticated phones for a reasonably affordable price, especially from an African perspective; they can't afford a $2,000 iPhone 17. That's going make a big difference in the market.
Where does this end up?
From the investment point of view, they are going to need a lot of speed to market. The technology exists, but the connectivity may not. There are initiatives - obviously the subsea cables providing connectivity, and I think we'll see a lot more access points landing around the coast. But then connecting the interior is the issue. Are they going to use more satellite? Is it going to be fibre that they're going to have to deploy and roll out? It's easy to say ‘deploy the technology’, but this is where engineers and [operators] fall into a trap.
We don't appreciate how big Africa is and we don't appreciate how diverse it is. I've been living and doing research up here since the late 1990s, and I've seen how it works, and the difference is that there's a lot of regulatory issues, there's a lot of local cultural issues that have to be navigated. You can't just go and say, we're going to deploy fibre across the Democratic Republic of Congo. There's going to be regulatory issues, there's going to be political issues, and before you even turn one sod, you're going to have to do a whole lot of work, which includes lobbying, before you can even get there.
Satellite represents the fastest way to get there, but [recently] Starlink said they're not taking on any new orders in Lagos, for instance, because they’re oversubscribed. We're seeing this happening in Lagos, in Accra, you see it in even in Nairobi, where Starlink is now becoming the preferred mode. That's the uplinks, the latency is acceptable by and large, but now they're running into problems where there's too much traffic; the bandwidth can’t accommodate all the traffic. So unless they launch more satellites…if you've seen a map of satellites rotating around the world recently, there's not all that much space.
Speed and affordability are going to be of the essence; applicable and relevant APIs and use cases are going to be very relevant. The big focus is obviously on consumer bases, leveraging of mobile and financial services. Going further, there will be concentrations in various areas of manufacturing. You can look at the annual reports by the telco - but that’s what they think the market is going to look like. What I really pay attention to is: what is the market demand - the enterprise demand by sector? It’s very important to understand that.
The way I see it, telcos have to start in a vacuum. Moving forward, they're going to see what it is that makes them money. Are they just going to be connectivity players? Become completely commoditized, so they're making fractions of cents on the dollar, and it's a volume business model? Or are they going niche, going into partnership ideas? On top of that, AI requires a lot more compute, a lot more storage and a lot more connectivity.
