Who is building Africa's internet infrastructure?
Mobile phones, cell towers, subsea cables, data centres and the web of physical infrastructure powering the continent's digital economy
Good morning friends and happy Wednesday!
After a long pregnancy (and Asabaako festival in between), this article finally sees the light of day.
It is my first long-format experiment, where I try to decompose a topic, make it intelligible, and highlight its hard-to-find fundamental data.
A special thanks go to Nomsa Muswai, for making time for my questions. Nomsa is a South Africa-based Network Engineer at Meta, and co-founder of the Africa Subsea+ Ecosystem Forum, an amazing initiative aimed at supporting the development of the Subsea Ecosystem in Africa.
Let’s jump into the backstory now.
The backstory: does infrastructure matter? 👷🏽♂️👷🏽♂️
The big news of last month?
Norrsken22 recently closed a new 200M fund to invest in growth-stage startups in Africa. As it is a big pot of money - especially in the current VC downturn - it garnered a lot of media buzz.
In an episode of The Flip, Natalie Kolbe - Managing Partner at Norrsken22 - summarises the strategy of the fund, explaining why they are committing this amount of money at this moment in time.
“There are three megatrends that we're looking to invest behind on the African continent”.
The explanation that followed is the reason behind this article.
Demographics
“In the next 25 years Africa will add about 800 million people to the working age population (...) And they are young. And they're doing everything on a mobile phone. So they feel very comfortable with digital products and services, (…) and these are the kinds of businesses that we're looking to invest behind that are exactly that, that are providing digital products and services”
Digital Infrastructure
“If you have a look at Africa, at what is available, most of Africa still runs on 3G and less. And we can all see there are most apps and the loading of web pages takes a long time, etcetera. It's frustrating to work on 3G. That is changing. There's a lot of investment that is going into digital infrastructure. And as that moves to a 4G, 5G environment, you're going to see a lot of uptake and adoption of digital products and services, again, which is going to drive the businesses that we're investing behind.”
Urbanization
“A lot of Africans are moving to the cities, and that creates a density around the city where you can scale. So as the populations are moving into the cities, they need financial services, they're needing healthcare, they're needing access to markets. And again, if you're able to provide that in a very quick and scalable way through these digital products and services.”
I loved her answer.
It was my first time hearing an industry insider explain - in plain English - the macro-trends shaping the continent. And although I knew all of them beforehand, her confidence got me excited, so I decided to delve into each trend to personally assess their status and find quantitative signals of their evolution.
Guess what?
While data on demographics and urbanization has been pretty easy to find, and equally easy to understand, discerning the state of digital infrastructure happened to be trickier:
what indicators should I look at, and where can I find them? But even before that: what exactly is digital infrastructure?
There is a gap on the topic. A lot of investments in tech products ultimately depend on the ability of people to consume digital services, i.e. connecting to the internet. And yet all the noise is captured by fund closings and startup raising, with little news on the underlying infrastructure that allows people to connect to the internet.
My perplexity grew as soon as I stumbled upon a 2019 well-cited report by the United Nations Broadband Commission for Sustainable Development.
I cite:
“According to the World Bank, achieving universal, good-quality internet access across Africa will require investments of $100 billion, 80 percent of which is needed for core infrastructure to establish and maintain broadband networks (…). This includes 250,000 new 4G base stations, at least 250,000 kilometres of fibre, and migration to 5G, which, in turn, could dramatically increase the need for data centres. While investors are increasingly aware of and interested in opportunities within digital infrastructure across Africa, investment levels today are insufficient to meet the growing demand”
Wow!
What if Natalie Kolbe was wrong, and infrastructure investments are not progressing quickly enough to sustain the growth of the businesses they are backing?
Who’s doing all this work? And how quickly they’re getting it done?
I had no clue and this is what we want to find out.
The invisible network: why infrastructure isn’t cool? 😔😔
One of the reasons they rarely talk about infrastructure in the media, is that it’s hard to make it “cool”.
Digital infrastructure is:
Mainly B2B: no consumers directly buy “pieces” of the digital infrastructure, there is no direct link between them and its components. Only large companies (telcos, internet service providers, cloud computing providers) interact with it, so it is hard to understand their utility if you’re not in the sector.
Technical: telecommunication protocols, data packets, switches and routers, servers, fiber-optic cables. All these engineering and technical terms and “I just use Instagram I don’t really need all this info”
Requires slow, patient capital: you don’t close a $50M Series A to build the next fiber optic cable, so no headlines on that front. You usually form a consortium of companies run by boomers that need to spend a lot of time negotiating with local governments, taking on debt, etc… It looks more like diplomacy than disrupting for the sake of it.
My thesis here is that: infrastructure can be cool, once you understand its importance.
Let’s imagine we are friends with Kodjo, a 20-year-old Ghanaian music producer. Kodjo is into Amapiano, and he’s planning on reaching out to several South African artists to feature them in his new project. Cool.
He managed to get a couple of phone numbers, so the way Kodjo is participating in the digital economy is by sending them messages on WhatsApp.
Breaking it down:
He needs a device (i.e. a smartphone)
He needs a relevant application (i.e. WhatsApp)
He needs to send and receive information (i.e. ”Are you down for a collab?”)
He needs (or better, Whatsapp needs) memory to store the application and its data (i.e. making sure the artists can still open the message in a week)
Let’s see how each of these activities is related to a piece of digital infrastructure:
He needs a device: device manufacturers, like laptops, smartphones, and tablets;
He needs a relevant application: all of the software companies building applications supported by the operating system running on his device;
He needs to send and receive information: a network that propagates data (strings of 0s and 1s representing his text message), i.e. cables, cell towers, IXPs, terrestrial fiber network, satellites;
He needs memory to store the application and its data: data centers, where servers are located.
If we want Natalie Kolbe’s prophecy to materialize, we can say that Africa will have a mature digital economy once:
everyone can use and afford a device;
plenty of applications exist to do all sorts of things (from watching a movie to booking your next haircut);
connection to those applications is fast, reliable, and low latency;
and where ideally every company can stock data and applications locally, in a cost-effective way.
If it’s yet unclear, don’t worry, we are about to see step by step, how a simple message on WhatsApp travels across tonnes and kilometers of physical infrastructure.
The Mobile Revolution: how the internet reaches the masses 📱📱
Let’s follow Kodjo’s journey into each step of the digital infrastructure.
To send a message to fellow South African artists, Kodjo needs a device. And not n’importe quel device: a device capable of connecting to the internet.
It might sound trivial, but it is not: devices are not as widely available as we might think. In Sub-Saharan Africa, only 45% of the population has a mobile device; and only 25% have a device and is connected to the internet.
This is both a problem and an opportunity: as 75% of web traffic comes from mobile phones, mobile phone ownership is the de facto gateway to the digital economy in the continent.
For decades (and up to now), the “typical” device looked like this: it is called a “feature phone”.
Feature phones still dominate the African market, even though smart feature phones and entry-level smartphones are threatening their position. As of Q4 2022, a total of 22.7M feature phones have been shipped to the continent, against 17.6M smartphones.
However, in the next 5-10 years, devices in the continent will increasingly look like this.
The first might look similar to the Nokia in the previous picture, but it’s not. It is an MTN Smart. It is a feature phone that supports an operating system (KaiOS), allowing applications like Facebook and YouTube to be used: a smart feature phone, a real innovation. As they can be produced at lower costs, but still support most of the applications of a smartphone, there is a major opportunity for large-scale adoption and other players are stepping into the game.
Some examples?
Vodacom launched a smart feature phone in Tanzania in 2019; Nokia has set up a strategy to upgrade its feature phones into 4G smart feature phones to compete with entry-level smartphones.
But feature phones are just a part of the puzzle.
In the smartphone segment, the market is way more concentrated, as large Asian manufacturers take the largest share of the pie.
Think of Transsion Holdings.
The company is the owner of 3 brands: Tecno, Infinix, and iTel. Combined, the three brands have 48 percent of the smartphone market in Africa in 2023. Following there is Samsung (22%), and way behind, Xiaomi and Oppo.
The company has been praised for its strategy of being smarter than competitors in understanding the specific needs of African customers. For instance, they have allegedly been the first in:
allowing dual SIM cards (increasing coverage);
adjusting camera for darker skin colors (better pictures);
embedding keyboards in Amharic, Swahili, and Hausa (local languages adaptability)
The goal of everyone here? Deliver the highest functionalities for the lowest prices.
However, adoption won’t be easy nor automatic. Although increasingly affordable, mobile phones can still take up a big portion of the income of an average customer.
In countries like Liberia, Burkina Faso, or Congo (RDC), buying a feature phone can take up to 50% of the average monthly income. And we are talking about devices that cost $25.
In Nigeria, MTN feature phone sells at $18.37, representing 11.5% of average monthly income.
At these price levels, even a feature phone can be considered a luxury good.
And even if there have been attempts at establishing African phone manufacturers, that of device affordability is a story that will likely be written by Chinese companies.
Who doesn’t remember the famous “first smartphone entirely produced in Africa”? The Mara Phones, produced in Rwanda, made headlines in 2019, but we haven’t heard news of it since then. The same applies to AfriOne, which localized smartphone production plant in Nigeria and released model Gravity Z1.
Although these attempts are cool, their market share is below 1%.
They might play a role in the next 5-10 years, but they are unlikely to modify the market.
One way where African innovation can play a role, however, is by making devices more affordable via smart asset financing models. I am talking about companies like M-Kopa, who are applying pay-as-you-go models where smartphones can be paid in installments. With one million smartphone customers reached between 2020 and 2021, 75% first-time 4G smartphone users and 40% first-time smartphone owners, these numbers show the real impact a company can have beyond phone manufacturing.
The company has already a great track record of making expensive assets available to large pools of consumers (like solar panels, TVs, and fridges), and appears to be well-suited for its new venture, with a combination of established customer relationships, software, and hardware (using a secure locking system embedded in the hardware, enabling the company to “lock” the smartphone if payment milestones are not met).
If the trajectory continues, it will represent a game-changer for smartphone ownership in the continent.
Let’s stop here and recap.
What devices are the internet’s port of entry in the continent?
Smart feature phones and entry-level smartphones, especially those supporting 4G connectivity.
Who’s building for device affordability?
Chinese companies.
Is there room for African innovation?
Not much in manufacturing, but rather in asset-financing models like M-Kopa.
Let’s move forward.
Applications: not the topic today 🚫
Once Kodjo has a device, he needs a relevant application to send a message.
Here we enter the startup world, the ultra-covered segment by the media. Almost every company has some type of mobile application to deal with distribution and customer relationships.
I will not delve into it too much as I want to shed light on the part of the infrastructure that is not talked about.
I will just say that two factors underpinning the quantity and quality of available applications are:
availability of talent (developers) and
investments levels (risk-capital).
The relationship between them has been explored in my first article.
Metals and cables: the world of the physical Internet 🗼🗼
Once Kodjo has WhatsApp, what is happening?
Whenever we use our phone like browsing the internet or sending a text, what we are ultimately doing is sending data. More precisely, we are sending a series of bits, which are strings of 0s or 1s. Let’s say you are sending a text message to your friend “How are you?”.
This is approximately 200 bits. Meaning your text is translated into 200 0s and 1s.
Given that 1 Mega-bit is 1 million bits, if you are connected to a 4G network, with a maximum capacity of 50 Mbps (megabits per second), ideally, every second you would be able to send 250,000 worth of “How are you”.
Interacting with the internet is nothing but constantly sending and receiving bits (0s and 1s), at a certain capacity, all across the globe at speed-light.
Now, it is important to know which routes these bits take to travel.
A lot depends on:
roads available;
traffic;
different owners of roads
All these is what we call internet infrastructure: where the data packets transit and how they reach my phone.
And it is a rather physical infrastructure.
In the case of Kodjo, we know that:
He sends a text from his phone in Accra, Ghana;
bits will travel via radio waves to a tower cell operated (or leased) by the mobile network operator (i.e. Vodafone);
the cell tower propagates the bits to the larger internet network via a physical connection to the terrestrial fiber network;
the bits will reach a submarine cable;
they will travel along the submarine until a landing point;
from the landing point, they will reach a data center where the application is stored;
the message will be redirected to a terrestrial fiber network;
who reaches the cell tower;
who reaches the cell phone of a South African artist.
Once the South African artist responds, the message will drive all the way back to Kodjo’s phone.
The reasons behind slow internet, high data costs, and low-latency connection are intrinsically related to all these pieces of infrastructure and how they interact with each other.
So if we want to understand it, we need to pass by:
cell towers
fiber terrestrial network
subsea cables
data centers
Let’s go.
Cell towers: the real backbone of Africa’s internet users 🔌 🔌
Back to Kodjo.
He is sending a WhatsApp message.
What is happening?
His SIM card sends radio waves to the closest cell tower antennas; these antennas are responsible for transmitting and receiving wireless signals to and from mobile devices.
Remember, the wave contains nothing but a packet of 0s and 1s, with additional info concerning where the message is going, like the IP address of the receiver.
How far is the closest cell tower, and how many are out there?
According to some market research, in Africa, there are +180,000 cell towers, and their number is expected to grow at a projected CAGR of 4.6% in the next 5 years. As 85% of the population in Sub-saharan Africa lives in an area covered by mobile broadband internet access, towers already cover a large part of the continent’s territory.
Good!
The reason behind their widespread presence is that their construction started as early as in the mid-90s when the first mobile revolution happened (the 2G revolution, which dates back to 1991; it was the first time voice calls, but also SMS and MMS could be transferred over long distances via radio-waves, even if the speed was limited at 64kbps).
As explained by a 2022 ITU report:
“The dominance of mobile broadband can be attributed to the low cost of upgrading existing mobile cellular networks to offer broadband compared with the cost of extending fixed networks”.
Translated: once you’ve built cell towers to support basic phone calls and SMS, it is easier to upgrade them to serve mobile internet.
This army of cell towers scattered throughout the continent, which constitutes the real backbone infrastructure of connectivity, is owned and operated mainly by two types of companies:
telecom operators
tower companies
Historically, it’s telecom operators who have been responsible for investing in the construction of cell towers.
Why? At the time, reaching customers equaled building towers. As your towers covered new areas previously unserved, you had built a competitive advantage over other telecom operators.
It started, of course, with older generation 2G. Over time, they have been upgraded to newer generations, like 3G and 4G, but the basic underlying infrastructure remains the same, and it looks like this.
You’d find a lot of them just by walking in Accra (I can see one from my co-working space).
It is important to repeat it: the main competitive advantage for a telecom operator was coverage. It was critical for them to build and maintain towers, in order to sell airtime and data transfer services to customers living in areas that weren’t covered. And of course, they wouldn’t let competitors use their towers, or they would charge them high prices for that. Building towers was one of the main drivers of the growth of telecommunications companies in the 90s and in the early 2000s.
The largest pan-African telcos are:
MTN
Vodacom
Airtel Africa
Orange
You then have national champions, who have a nearly-monopolistic position in their home markets, like:
Safaricom (Kenya)
Etisalat Egypt (Egypt)
Maroc Telecom (Morocco)
Ethio Telecom (Ethiopia)
These companies still own and operate a good deal of these towers.
But they are not anymore the dominant player.
Things have changed for two reasons:
While cell towers represent a big cost for the operators, they are not a big competitive differentiator anymore;
Governments and regulators push for shared infrastructure (meaning you can’t be greedy with your towers anymore, preventing competitors from using them).
This is why a lot of these companies are slowly divesting their physical infrastructure.
Telcos nowadays fancy an asset-light model, where they want to avoid the overhead headaches of maintaining physical towers and prefer focusing on serving end-customers.
They are slowly transforming into what some people call “Mobile Virtual Network Operators”, i.e. wireless communications services providers that do not own the wireless network infrastructure over which they provide services to its customers.
In what is now a global trend, and has been baptized as “the great sell-off”, MNOs continue to sell off their tower assets.
In Africa today, less than half of cell towers are owned by Mobile Network Operators.
The question that arises then is: who’s buying these towers? Well, tower companies.
I didn’t even know of the existence of this specific market segment. I’ll just say that if you think the telcos market is concentrated, it’s because you haven’t seen towercos.
The market is dominated by giants like:
Helios Towers
American Tower
IHS Towers
Together, these 3 companies own almost 40% of the total cell towers in Africa.
And towercos have now a majority share in the infrastructure of key markets like NIgeria, DRC and South Africa.
So in 10-15 years, we have come from telcos building infrastructure to sell their services to underserved populations, to physical infrastructure shifting hands to companies (towercos) exclusively focused on owning and operating these hard infrastructural assets and leasing them to whoever needs broadband.
And we are seeing deals that prove it.
One example? Airtel Africa’s selling over 1200 of its towers in Malawi and Madagascar to Helios Towers, for a compounded value of $107M. Another example? IHS Towers purchased MTN’s South African 5,700 towers for a staggering $413M.
But we also see interest by private equity investors like Verod Capital in backing regional players, like investments flowing into Nigerian Pan African Towers, aiming at strengthening the company’s position in key markets like Nigeria.
So to recap:
What are cell towers?
They are the backbone of internet connectivity and cover the almost entirety of the territory.
Who built them?
Telcos built them, but they don’t want them anymore.
Who owns them?
Increasingly tower companies entering the market.
However, the great sell-off comes with a cost.
If you are telco, you not only want towers; you want towers supporting 3G, 4G, and eventually 5G technologies. These upgrades come at a cost. Will there be the right incentive structure for these upgrades once ownership has shifted hands?
This is the reason why, for example, in markets where some companies are ridiculously dominant, like MTN in Ghana, there seems to be a more cautious approach to divesting. They rather have control over their physical network of towers to make sure they timely roll out 5 upgrades.
“MTN is investing US$1bn in network infrastructure by 2025, with plans to deploy 4G on around 1,000 new sites by the end of 2023 as well as expanding its fibre network to increase capacity, reaching 2G, 3G and 4G coverage levels of 99.5% and 99.3%”.
MTN is well aware that maintaining such a dominant position requires maintaining quality of service, which in turn demands infrastructure investment. And they are willing to make it with a rather hands-on approach.
Fiber terrestrial network: the hard part of the internet plumbing 🪖 🪖
Kodjo’s message has reached the antennas of the cell tower owned by his network operator (i.e. Vodafone).
Now what?
The data packet (a string of 0s and 1s) must be propagated to the larger internet network in order to reach the shores of South Africa.
This is where fiber terrestrial networks come into play.
What is it?
It is tubes that are either buried underground or installed alongside roads or power lines (i.e. hanging from utility poles), and they look like this.
Here is a map of the terrestrial fiber network in Africa.
As of June 2022, the continent had 1.184 million kilometers of installed capacity.
South Africa has by far the most extended terrestrial network. Ongoing projects are about to cover important portions of territory, especially in countries like Côte d’Ivoire and DRC.
Do you want to know who is building one of the biggest projects of terrestrial infrastructure (with more than 12 Tbits capacity)? Nokia. Yes, you heard right. Nokia partners with Liquid Intelligent Technologies to deliver a terrestrial fiber corridor as a first-of-its-kind in Africa in terms of distance and capacity.
Other big deals are not missing. The most juicy is probably a $320M partnership deal between Bayobab and Africa50 to develop Project East2West, a terrestrial fiber optic cable network connecting the eastern shores of Africa to those on the continent’s west.
Some of the largest players in the sector are:
Liquid Intelligent Technologies (owns over 100,000km of terrestrial fiber)
Bayobab (formerly MTN GlobalConnect, has over 112,000km of fiber)
BCS Group (owns 80,000km of terrestrial fiber)
WIOCC (75,000km of terrestrial fiber, mainly focused in southern Africa)
Telesonic (former Airtle, owns 75,000km of fiber)
What can we say about these companies? Two main observations.
First, telcos here are once again leading actors in the space, with major terrestrial fiber companies being the product of spin-outs of network operators like MTN and Airtel.
The reason is simple.
What telcos sell to consumers like Kodjo (the possibility to use the internet), is ultimately physical infrastructure: the cable highways that propagate data packets to their final destination. The “stock” of products these mobile network operators sell, is the internet bandwidth they purchase. If there is not enough bandwidth, because there is not enough infrastructure, they have to build the infrastructure, otherwise, they will not be able to service clients.
However, in some regions, regulatory frameworks encourage or require the separation of infrastructure and service provision to promote competition and prevent monopolistic practices. This is why as in the case of cell towers, companies are establishing separate companies that will engage more fairly with a broader set of actors. It is the above-mentioned case of Bayobab (MTN) and Telesonic (Airtel).
Secondly, given the high complexity of these projects (you bore holes on kilometers of land), they often require the active involvement of the regulator.
This is why you more often see state companies involved in these projects. It is the case of Côte d’Ivoire, where a large part of the terrestrial network is being directly commissioned by the Minister of Telecommunications.
A bit more extreme is the case of Cameroon, with Camtel’s dominant position. As expected, not everyone is happy with it.
According to ITWeb Africa:
“The market dominance of Camtel in the ownership of critical infrastructure continues to threaten competition and the viability of investment in additional network expansion by other operators. Camtel maintains a monopoly over access to the country’s international and terrestrial fibre networks and appears to enjoy the protection of the regulator”.
Who is the client of these companies?
Whoever needs connectivity at a B2B level.
For instance, MTN purchases bandwidth from owned and operated by Liquid Intelligent Technologies, at a certain price.
Subsea cables: the oxygen of global connectivity 🤿 🤿
Kodjo’s message has reached the coast of Ghana. Now it has to find its way to South Africa. How?
Via a subsea cable.
To grasp the importance of subsea cables, think that 91.7% of the total bandwidth (16.630 Tbps) used in Sub-Saharan Africa is supplied directly by submarine cables. This means that the majority of databits sent and received by African internet consumers passes through optic-fiber highways stretched under the sea.
The growth of this infrastructure component has been simply astonishing: the total subsea cable capacity to Africa has grown ~7x since 2016.
If we look at the 3 most recent projects, they alone surpass the total installed capacity of all previous cables:
PEACE (15,000km long, with a capacity of up to 96Tbps)
Equiano (Google’s owned cable, 15,000km long with 144Tbps capacity)
2Africa (Meta’s jewel, 45,000km long, making it the longest subsea cable ever deployed, with 180 Tbps capacity)
Now, the question arises: who is building these cables? And why?
The recent surge in subsea cable investments has been mainly driven by what in jargon is called “hyperscalers”.
Hyperscalers is a general term describing:
Cloud Service Providers (like Amazon Web Services);
Over-the-top media companies (like Netflix);
Big tech (like Facebook)
What do these companies have in common?
They use A LOT of internet traffic. Some argue that 2/3 of global internet traffic comes from their services.
Social networks, cloud computing, and streaming are all expensive applications in terms of data consumption. To continue providing their services at quality, and reach new customers, they are building their own infrastructure.
Nomsa (Network Engineer at Mera) explains it well:
“(For Meta) the amount of infrastructure required is insane (..) There is not a single operator in the world that can give us all the capacity we need, so in that case we need to create our own. It is mostly an issue of scale (…)
Also, the present cables are reaching their end of life (…) After 10-15 years you already see decreased capacity in existing cables, so you need to reinvest on them. We need capacity to push traffic, and we push petabits of traffic. It is mostly an issue of scale, but in some cases it is a matter of Quality of Service. We invest in infrastructure to improve redundacy and give our clients reliability and best quality of service.”
Globally, demand for bandwidth is expected to double every two years. This growth in bandwidth is being driven by an increasing number of users, coupled with greater data consumption per user. Basically, more people on the internet using more data-expensive applications.
So if these companies want to:
be able to grow, while
maintaining scale, and
preserving the quality of service
they need to adopt a hands-on approach and invest in subsea cables.
Otherwise, you would soon see Facebook breaking every other day, which is not good for their reputation.
Hyperscalers are important, but they are not the only players.
The case of the 2Africa subsea cable is a good example to shed light on different types of actors that have incentives to invest in the space. You have:
big tech (Meta)
Telcos spin-offs (Bayobab, formerly MTN GlobalConnect)
Traditional Telecoms (Orange, Telecom Egypt, Vodafone)
State-owned, huge telcos: China Mobile International, STC Group (Saudi Arabia)
connectivity wholesalers (WIOCC)
I wouldn’t be surprised to see tickets from purely infrastructure investment firms.
Here is a map of all submarine cables in Africa.
Who do they sell to? Who uses their capacity?
It is, as expected, a B2B play. Telecommunications carriers, mobile network operators, cloud service providers (CSPs), and over-the-top (OTT) media service companies, are major users of bandwidth provided by subsea cables. But you also have multi-tenant data center operators, financial services companies, government agencies, and large enterprises.
We will see why in the next section.
Data Centres: the hardware backbone of the digital economy 💽 💽
Now that Kodjo’s message has finally reached the shores of Cape Town, where does it go next?
It goes to a server.
Welcome to the last component of the digital infrastructure: data centers.
Now, you may wonder why the message has to pass through a data center, instead of going directly to the mobile phone of the receiver.
The reason is that every interaction we have with an application has to pass through:
where the application is stored,
where the application stores its data.
And this is usually a server, hosted in a data center.
An example of why we interact with WhatsApp first, before the message is sent to the receiver:
WhatsApp servers determine the most efficient path to deliver the message to the recipient's device, including handling messages when the recipient is offline;
The servers play a role in managing the encryption keys (though they don't have access to the keys themselves) of the message;
WhatsApp servers have an active role in various features attached to the act of sending a message, such as message status (sent, delivered, read);
By routing messages through its servers, WhatsApp can implement measures to detect and prevent spam and abuse.
Ultimately, the message is stored for a specific amount of time on Whatsapp servers.
(In Kodjo’s example, we are acting as if Whatsapp servers are in South Africa. They are not).
What is a data center?
A data center is a physical facility that houses:
a lot of servers;
cooling equipment (to cool down servers);
access to reliable electricity (allowing servers to run);
all types of connections to the internet network (to manage the inflows and outflows of data);
a bunch of other stuff like IXPs (not the topic here).
Data centers are the invisible physical computers powering the global digital economy. None of the applications we use could exist without data centers.
They enable the processing and storage of large amounts of data, host websites, support cloud computing services, blablabla. This is why they consume a lot of bandwidth: a lot of internet traffic necessarily passes through them.
Currently, Africa accounts for only 1 percent of global data center capacity. And 2/3 of that capacity is located in South Africa.
This means the majority of applications and data are stored outside of the continent, even if these applications are owned by Africans and the data is of African consumers.
Good news?
Data centers might not be a lot, but their number is growing. Carrier-neutral data centers went from 20 to 67 in the period 2020-2022.
In 2020, the African data center market was valued at US$2 billion and was expected to reach US$5 billion by 2026, growing at a compounded annual growth rate of 15% from 2021 to 2026, according to Intelligent CIO.
I think we are already beyond that.
In the last edition of the Africa Interconnection Report, around US$6 billion of announced investments in data centers have been identified. In descending order: Vantage (US$1 billion), Africa Data Centres (US$500 million), WIOCC’s Open Access Data Centres (US$500 million) and Colo West (US$259 million).
And this is old data. Investments are progressing.
One recent example? After the 2021 acquisition of MainOne for $320M, Equinix alone is set to invest a further $390M in African data centers in the next 5 years.
The question that now arises is: who is building data centers?
Here are some of the key players:
Africa Data Centres: part of Cassava Technologies, they are operating 9 data centers in 6 countries, with 54MW of total capacity by 2025)
PAIX: born in 2016, they have 2 data centers, one in Kenya and one in Ghana
Teraco: South African company, they have 3 data centers in the country
MainOne: acquired by industry leader Equinix, they have 3 data centers in Nigeria, Ghana, and Cote d’Ivoire
Open Access Data Centres: operating 6 data centers in South Africa, Nigeria, DRC
IXAfrica: largest player in East Africa, Nairobi
What type of companies are they?
I admit it is hard to decipher.
On the one hand, you see the presence of telco spin-outs (as in the case of other infrastructure components). For example, Airtel has launched Nxtra Africa, a company focused solely on building a data center network in Africa.
However, you definitely see a more specialized set of companies. The main reason is that data-center companies deal with two sets of clients we haven’t previously seen:
Cloud Service Providers: they aim at distributing computing power all around the globe to better serve customers;
large companies, governments and banks: they usually manage their own servers to have better control of their data, instead of relying on cloud storage services
A company building data centers doesn’t necessarily provide IT equipment, but rather manages the facility, making sure power, cooling, and connectivity are provided. This is why you might see real-estate investors tapping into the segment, something you wouldn’t see in other pieces of the infrastructure. This is also why sometimes they bundle other services that companies might be interested in, like cloud computing.
In any case, to whoever is building them, we need more of them.
Some might ask: why do we need more data centers in the continent? Why can’t we just store all of the applications (and user data) overseas?
The short answer is that localizing storage is important in order to:
reduce latency (better video calls or online gaming);
optimize intra-African traffic flows (data packets traveling shorter distances); and
slash operating costs and other risks related to currency fluctuations (as data storage is like an imported good that you pay in dollars)
Local data centers usually correspond to a better user experience. As data consumption and production increase in the continent, we need more data to be stored on the continent, if we don’t want to put a strain on the international bandwidth (i.e. too much traffic flowing in and out the continent).
One additional reason is law enforcement. In some cases, the government wants data on their citizens to be stored within the state borders, as you can hardly enforce any data protection law if the data is not stored within your borders.
For example, “there is a regulatory requirement by the central bank of Nigeria for all banks to localize their data storage and move them to tier-3 data centers” (source).
Finally, building data centers can act as an enabler for an ecosystem to develop around it, like IXPs and local cloud providers.
And this is why companies are doing it. In this case, as opposed to the cell tower business, we see a larger share of purely African players having their stakes in the game.
One of the names worth mentioning here is Cassava Technologies. The company is a full stack, vertically integrated holding of digital infrastructure in the continent. A true global champion that is behind businesses like Liquid Intelligent Technologies and Africa Data Centres.
Kudos to them!
The call to action: let’s make infrastructure cool again ✨✨
And here we are at the end of our journey.
Once having reached the Whatsapp servers, Kodjo’s message will travel backward through devices, cell towers, terrestrial fiber, subsea cables, and data centers.
For him to create partnerships and grow as an artist, all this digital plumbing had to be put in place. The ease, the quality, and the cost of it, is highly dependent on investments in the digital infrastructure we make in the next decade.
Looking back at our initial interview, we should give credit to Natalie Kolbe’s prophecy. Investments in digital infrastructure are real and they are moving fast: probably faster than what has been predicted. Even though their growth has been uneven across its different components.
We see the most rapid growth in the subsea cables segment, followed by cell towers. Device affordability is making important steps forward, with initiatives to push smart feature phones in the market, and asset-financing models like M-Kopa.
Pain points remain expanding the terrestrial fiber network (due to their complexity), and building more data centers (scale + power reliability).
With this in mind, we can all agree that there has been a surge in investments in infrastructure, driven by:
telcos, which remain big and relevant actors in the continent;
multinationals looking for opportunities to grow their market share (with tower companies in the first place);
some African champions (like Cassava Technologies)
The fact that large foreign multinational companies have outsized power over key infrastructures could be somewhat of a concern. At the same time, you don’t want to reproduce the situation of Cameroon, where state-owned company stifles competition and progress.
It is a balance that needs to be found.
I would like to close with a final note, that is a call to action: we need more people working on infrastructure.
To quote the words of Nomsa:
“There is a real big gap in terms of skills. There is no skills pipeline, most young people today they like to follow the buzzwords and what’s trending, like AI. Unfortunately, they don't know the significance of the subsea cable, they don't know that the day that they didn't have internet, it was because of the cuts that was the in on the shores of Ghana, and this is a real example I'm giving you. (…)
So most of the experts on the field that know how to run a wireless network are reaching their retirement age, and are actually not based in Africa (…)
So when you look at Africa, down the line in 10 years, we've got all of these cables and all this infrastructure, we need experts who are based on the continent, and who knows everything on how to run a cable.”
This is why she, together with other industry experts, has started ASEF Africa: to bring actors around the table and strengthen the ecosystem. Their first goal is to address the skills pipeline. I love their mission, and this is the type of initiative we need to support.
Infrastructure is important. Especially in the continent. We need to talk more about it. And we need to find a way to make it sound cool so that it becomes more attractive to people, especially youngsters.