Africa is building data centres faster than at any point in its digital history. Across South Africa, Nigeria, Kenya, Egypt, Ghana, Côte d’Ivoire, and Senegal, billions of dollars are flowing into new facilities designed to support cloud computing, artificial intelligence (AI), financial technology, and the continent’s rapidly growing digital economy.

On paper, it looks like exactly the infrastructure Africa has been waiting for. The Africa Data Centres Association (ADCA) projects the market could reach $9.2 billion by 2029, reflecting strong investor confidence and growing demand for digital infrastructure.

Yet behind the construction boom lies a quieter question that deserves more attention: who will actually fill these data centres?

According to Africa Data Center Association’s 2026 Economic Report on Data Centres in Africa , reaching 85% occupancy in African markets can take six to eight years, compared with only two to three years in Europe. At the same time, between 80% and 90% of Africa’s data is still processed and stored outside the continent in established cloud hubs such as Dublin, Frankfurt, and Amsterdam.

The challenge facing Africa is no longer whether it can build world-class data centres. It is whether enough digital workloads will stay on the continent to make those facilities commercially successful.

Africa Has Seen This Story Before

This is not the first time Africa has built infrastructure ahead of demand.

A few years ago, mobile broadband networks expanded rapidly across the continent. By 2023, coverage reached nearly 87% of the population in sub-Saharan Africa, yet millions of people remained offline because smartphones were too expensive, electricity was unreliable, mobile data was costly, and digital literacy remained low. Over time, those barriers gradually began to fall, and adoption followed.

Data centres appear to be entering a similar phase. Building a facility is only the beginning. A data centre only creates value when businesses consistently move their workloads into it. Banking platforms, cloud services, AI applications, fintech systems, government platforms, and enterprise software are what fill server racks and generate revenue. Empty floor space, no matter how modern the building, does not.

That is why occupancy has become one of the most important indicators of success. Enterprise migrations are often slow and complex, sometimes taking between six and twenty-four months. Governments continue to retain many critical systems on their own infrastructure because of security and sovereignty concerns. Meanwhile, many startups still choose European cloud regions where pricing, tools, and uptime are more predictable.

In many respects, the infrastructure has arrived before the surrounding digital economy is fully ready to absorb it.

The Economics of Empty Capacity

For operators, this delay creates a significant financial challenge.

Constructing a Tier III data centre requires enormous upfront investment. McKinsey estimates construction costs at approximately US$11.3 million per installed megawatt, with a substantial portion dedicated to power infrastructure, including generators, transformers, battery systems, and fuel reserves. These investments are essential because reliable electricity cannot always be guaranteed.

The costs, however, do not stop once construction is complete. If customer demand grows slowly, operators must continue maintaining expensive facilities while large portions of their capacity remain unused. Every empty rack represents infrastructure that has already been paid for but is not yet generating revenue.

Power further increases the pressure. Compared with mature markets, African operators rely more heavily on diesel generators and captive power generation to compensate for unstable electricity grids. As a result, the continent’s average Power Usage Effectiveness (PUE) stands at 1.67, above the global benchmark of 1.58. Although the difference appears small, higher energy consumption compounds over time, increasing operating costs and making local hosting less competitive.

Why Workloads Continue to Leave Africa

Even when African businesses adopt cloud computing, many of their workloads still end up outside the continent.

Recent research from PwC’s 2025 Africa Cloud Business Survey  shows that cloud adoption across Africa has matured significantly. Yet this maturity is not leading to complete migration into public cloud platforms. Instead, organisations are adopting hybrid and multi-cloud strategies.

Sensitive information often remains on-premises or within sovereign cloud environments to satisfy regulatory requirements, while less critical applications move to public cloud providers.

The challenge is that the largest share of cloud spending often follows those public cloud workloads. For many businesses, hosting applications in European cloud regions remains more economical. Once electricity, cooling, infrastructure costs, security, and currency fluctuations are factored in, hosting a standard cloud workload in cities such as Lagos can cost between 20% and 40% more than running the same workload in Dublin.

Global cloud providers also benefit from decades of investment in mature infrastructure, offering highly competitive pricing, extensive developer tools, and consistently reliable performance.

As a result, many African startups and enterprises continue to default to European cloud regions, even when local capacity exists. Until Africa can narrow the pricing and reliability gap, data sovereignty alone may not be enough to persuade organisations to keep their workloads at home.

AI Raises the Stakes

Artificial intelligence is adding a new dimension to the challenge.

Most existing African data centres were designed for traditional enterprise computing, where rack densities average around 5 kilowatts. AI workloads require considerably more power, often between 20 and 50 kilowatts per rack, with some advanced deployments approaching 100 kilowatts.

“Only five years ago, and even sometimes now, we see capacities of two to three kilowatts per rack. AI requires more than fifty kilowatts per rack. That changes the landscape dramatically. You need a completely different cooling system, a completely different power system. Everything is essentially new. That is the beauty of this industry, because you learn every day. Things are changing, and they are changing rapidly, so you have to be prepared, flexible, and scalable.”

 – Wojtek Piorko, Managing Director for Africa, Vertiv 

“You step back for a minute, and you see that every single day there is a new press release about a new AI data factory or AI data center being built here, there, and everywhere. All those factories are generating enormous amounts of data, and that data is not just moving within data centers anymore. It is moving across data centers.”

 – Joe Marsella, Vice President, International Business Development, Ciena

Only a relatively small number of facilities across the continent are currently equipped to support these high-density AI environments. The consequence is increasingly visible among African AI startups.

Many companies developing AI products train their models using infrastructure located in Europe or North America because GPU-ready facilities remain scarce within Africa. In effect, investment raised to build African AI solutions is often spent on overseas cloud infrastructure. This creates a difficult cycle. Africa is producing AI talent, attracting investment, and building innovative companies, yet much of the computing power required to support that innovation still sits outside the continent.

The ambition may be sovereign AI, but the infrastructure supporting it remains largely international.

The Next Phase Is About Utilisation

Africa’s data centre industry has already demonstrated that it can attract investment and deliver world-class infrastructure. That achievement should not be underestimated.

The next challenge, however, is no longer construction. It is utilisation.

Success over the coming years will depend less on how many new facilities are commissioned and more on how quickly banks, governments, cloud providers, AI companies, and enterprises move their workloads into those buildings.

Occupancy is becoming the metric that matters most. Africa is building capacity with confidence. Now the continent must create the conditions that encourage those servers to stay busy.

Because by 2030, the real measure of success will not be how many megawatts Africa has built. It will be how much of Africa’s digital economy is actually powered from within its own borders.