Who Owns Africa's Compute?

Africa is home to approximately 1.58 billion people in 2026, about 19.1% of the world's population (World Population Review). Its share of the world's active data centre capacity is approximately 0.6% (Africa Data Centres Association).

Those two numbers belong in the same sentence, because together they describe something more consequential than an infrastructure shortfall. They describe a structural condition: a continent that generates an enormous and fast-growing share of the world's data, identity records, transactions, health information, and language, while owning almost none of the physical capacity required to compute, store, and govern it.

The conventional framing of Africa's digital transformation has been a connectivity story — bandwidth, mobile penetration, last-mile access. That story is being solved, faster than most observers expected. The harder story sits deeper in the technology stack: electricity, compute, sovereign data infrastructure, AI governance, and cyber resilience. Africa is on the verge of gaining the ability to move data at global speed while still lacking the ability to process it at home.

The capacity picture, in megawatts

Facility counts are a poor measure of this gap. Databases disagree on what constitutes a data centre — a colocation site, a hyperscale campus, a carrier-neutral facility, an individual building within a campus — and totals vary accordingly. Installed capacity is the metric that reflects what actually matters for AI and cloud workloads.

The Africa Data Centres Association's Data Centres in Africa 2026 economic report puts the continent's position as follows:

The most important finding in that report is not the 360 MW. It is the association's own assessment that this growth is projected to maintain rather than increase Africa's global share (Africa Data Centres Association). Capacity on the continent is on track to roughly triple to about 1.2 GW by 2030 (reporting on the ADCA findings) — and Africa still ends the decade at under one percent of global capacity, because global hyperscale construction is expanding at least as quickly.

Africa is not closing a gap. On current trajectories, it is running fast enough to stay exactly where it is.

Bandwidth is being solved. Compute is not.

The 2Africa subsea cable makes the asymmetry concrete. The system spans roughly 45,000 kilometres, connects 33 countries, and carries a design capacity of up to 180 Tbps using spatial division multiplexing across 16 fibre pairs, with 168 Tbps supported on the West segment from the UK to South Africa (Vodafone; Submarine Networks). Meta completed the core build in late 2025 (Ecofin Agency).

That is a genuine step change in international bandwidth. It should reduce wholesale capacity constraints, improve redundancy, lower latency, and make new cloud regions commercially viable.

But a subsea cable is a transport asset, not a compute asset. Higher bandwidth in the absence of domestic capacity does something specific and often unexamined: it makes it cheaper and faster for African data to be processed somewhere else. The same infrastructure that enables local cloud regions also lowers the friction of permanent dependence on foreign ones. Which outcome materialises depends almost entirely on what gets built on the landing side of the cable.

Power is the binding constraint

Africa's compute ambitions collide with an electricity deficit that has not moved for years. As of 2024, around 600 million people in sub-Saharan Africa — 47% of the population — lacked access to electricity, with electrification barely keeping pace with population growth (International Energy Agency).

Meanwhile, the global demand curve for compute power is steepening sharply. Data centres consumed around 415 TWh of electricity in 2024, roughly 1.5% of world consumption, and the IEA projects that more than doubling to approximately 945 TWh by 2030 — comparable to Japan's total current electricity consumption (International Energy Agency).

AI clusters require continuous, high-quality, high-density power. Intermittent grids cannot host them, and diesel-backed workarounds destroy the economics. So Africa is being asked to compete for AI infrastructure in the same decade it is still trying to deliver first-time electricity access to hundreds of millions of people. Every serious continental compute strategy is therefore, first, an energy strategy.

Water belongs in this conversation too, carefully. Evaporative cooling consumes significant volumes of water, and 418 million people across Africa lack even a basic drinking water service (UNICEF/WHO Joint Monitoring Programme reporting). These two facts do not connect mechanically — household scarcity is not the same variable as industrial cooling supply. But they create a real political constraint. As African cities compete for scarce power and water, governments will increasingly have to demonstrate that hyperscale infrastructure can be built without displacing residential, agricultural, and industrial demand. Projects that cannot answer that question will not be permitted for long.

Sovereign AI: cloud adoption is not sovereignty

Here the argument sharpens. African enterprises are not lagging on cloud. PwC's 2025 Africa Cloud Business Survey found that 40% of African organisations are using or planning to use trusted national partner clouds for risk mitigation, resilience, disaster recovery, and security, and that 36% are using or planning to use sovereign public cloud specifically to leverage AI capabilities — a rate comparable to Europe and the Middle East at 37% (PwC South Africa; PwC).

Sovereign cloud in Africa has moved from compliance response to strategic requirement. But demand for sovereignty is not the same thing as possession of it. If African institutions consume AI through models trained elsewhere, hosted elsewhere, on hardware owned elsewhere, and tuned on datasets that under-represent African languages, contexts, and populations, then cloud maturity delivers capability without control.

The governance architecture is, notably, further along than the physical one. Agenda 2063 is the African Union's long-term development blueprint, not an AI framework. The operative instrument is the Continental Artificial Intelligence Strategy, endorsed by the AU Executive Council at its 45th Ordinary Session in July 2024, which positions AI as a strategic asset pivotal to achieving Agenda 2063 and sets out an Africa-centric, developmental approach spanning AI infrastructure, data, skills and research, governance and regulation, private-sector innovation, and international cooperation (African Union; Continental AI Strategy, July 2024).

Agenda 2063 supplies the political destination. The Continental AI Strategy is increasingly the governance architecture for reaching it. What remains missing is the substrate both documents assume.

Resilience: sovereign infrastructure raises the value of the target

There is a version of this story where Africa builds the compute, wins the cloud regions, and still loses — because the infrastructure becomes a higher-value target than its defences can support.

INTERPOL's African Cyberthreat Assessment Report 2026, drawing on law enforcement intelligence from 36 African member countries, reports that artificial intelligence enabled 55% of reported cybercrimes across the continent, and that cybercrime-related losses rose from USD 192 million in 2024 to USD 484 million in 2025 (INTERPOL; full report).

That is a 2.5x increase in a single year, in a region where institutional response capacity is thin. INTERPOL's 2025 assessment documented the capability side of the same problem:

• More than 30% of reported crime in parts of West and East Africa was cyber-related.
• 90% of surveyed countries said law enforcement or prosecution capabilities required significant improvement.
• Only 30% had an incident reporting system.
• 29% had a digital evidence repository.
• 19% had a cyber threat intelligence database.
• 86% said their capacity for international cooperation on cybercrime needed improvement.

Source: INTERPOL, Africa Cyberthreat Assessment reporting

Attackers are compounding their capability with AI faster than defenders are compounding theirs. Which yields the constraint that should govern every African data-sovereignty roadmap: sovereign AI infrastructure without sovereign cyber resilience simply concentrates more valuable assets behind weaker defences. Repatriating data is a security decision as much as an economic one, and it only improves the position if the receiving institution can defend, detect, investigate, and prosecute.

The real question

The infrastructure numbers are not the story. They are the evidence for a set of ownership questions that the next decade will answer, whether or not African institutions participate in answering them:

Who owns Africa's compute? Where does African data physically live, and under whose jurisdiction? Who trains the models that will make decisions about African citizens? Who defends that infrastructure when AI-enabled attacks are already implicated in more than half of reported cybercrime? And who ultimately captures the economic value generated by 1.6 billion people?

Africa is gaining bandwidth. It is adopting cloud, and demanding sovereign cloud, at rates comparable to Europe. The governance frameworks exist and are improving. What is not yet in place is the power, the capacity, the skilled workforce, and the institutional defensive capability required to convert all of that into ownership rather than throughput.

The continent does not primarily need more servers. It needs the electricity to run them, the institutions to govern them, the workforce to operate them, and the resilience to defend them. That is a capability agenda, not a procurement one — and capability is built, not bought.

References

1. Africa Data Centres Association. (2026). Data Centres in Africa 2026: The Economic Report. africadca.org
2. African Union. (2024). Continental Artificial Intelligence Strategy. au.int
3. International Energy Agency. (2025). Financing Electricity Access in Africa: State of Play. iea.org
4. International Energy Agency. (2025). Energy and AI: Executive Summary. iea.org
5. INTERPOL. (2026). African Cyberthreat Assessment Report 2026. interpol.int
6. PwC. (2025). Africa Cloud Business Survey. pwc.co.za
7. UNICEF/WHO Joint Monitoring Programme. Progress on Drinking Water, Sanitation and Hygiene in Africa. reliefweb.int
8. Vodafone. 2Africa: A Milestone for Africa and for Global Connectivity. vodafone.com
9. World Population Review. (2026). Africa Population 2026. worldpopulationreview.com