Every generation of infrastructure has a defining catalyst. For the last one it was the smartphone. For this one it is artificial intelligence — and the compute, connectivity and power it demands. In 2026, global data-centre capital expenditure will cross one trillion dollars for the first time, with the five largest hyperscalers alone committing roughly $725 billion to AI infrastructure this year. McKinsey estimates the world could spend up to $7 trillion on AI and data-centre capacity through 2030.
Set against that, the numbers for Sub-Saharan Africa are stark. The continent accounts for roughly 0.6% of global data-centre capacity and captures around 1.25% of global AI spending. The largest infrastructure mobilisation in modern history is under way, and Africa is starting it with a rounding error.
We do not read that as a reason for pessimism. We read it as the clearest signal in a decade that the next five years represent the most significant digital-infrastructure investment window the continent has seen — not despite AI, but because of it. This piece sets out the data behind that view, where we expect capital to flow, on what timeline, and how investors and operators can position now.
A demand shock meets a thin base
Sub-Saharan Africa enters the AI era structurally under-built. Internet penetration sits at roughly 38% of the population, fixed broadband at around one connection per 100 people, and about 10% of the population still has no mobile-broadband coverage at all (GSMA, Mobile Economy Sub-Saharan Africa 2024). Even where facilities exist, they are lightly used: Xalam Analytics estimates that in African markets outside South Africa, live IT capacity represents only about one-third of built potential.
Yet the demand signal is unambiguous. Africa’s AI market is projected to grow from $4.51 billion in 2025 to $16.53 billion by 2030 — a compound annual growth rate above 27% — with Sub-Saharan Africa absorbing 62% of that investment (Mastercard, AI in Africa 2025). By 2029, 85% of African businesses expect to be investing in AI, and AI-enabled activity could underpin as many as 230 million digital jobs across the continent by 2030.
The risk is a widening divide. AI diffusion — the share of the working-age population actually using AI — has reached 27.5% in the Global North but only 15.4% in the Global South, and the gap is widening each quarter (Microsoft AI Economy Institute, Q1 2026). South Africa leads the region at 23.1%, ranked 46th of 147 economies globally; most of the continent trails well behind. AI can compound Africa’s infrastructure deficit — or, financed correctly, it can become the demand catalyst that finally makes the build economics work.
Why AI rewrites the investment logic
For two decades, African digital-infrastructure returns were driven by a single equation: connect more people to mobile voice and data. AI changes that equation in three ways.
First, it shifts value from the access layer to the compute layer. Training and, increasingly, inference workloads require dense, power-hungry, low-latency facilities. The African Energy Chamber projects continental data-centre demand will reach 2 GW by 2030 — a tripling of today’s roughly 1.25 GW of total capacity, which breaks down into about 360 MW live, 238 MW under construction and 656 MW in the development pipeline (ADCA, Data Centres in Africa 2026). Importantly, 36% of African operators have already deployed AI-ready equipment and a further 35% are provisioning for it through modular designs. The retrofit and greenfield opportunity is live now, not theoretical.
Second, it turns power into the binding constraint, and therefore the opportunity. That 2 GW of demand is a catalyst for an estimated $10–20 billion of new energy investment, because in most African markets you cannot sell a megawatt of compute you cannot reliably power. The winning data-centre thesis is increasingly a power thesis: captive renewables, grid partnerships and cooling efficiency.
Third, it raises the value of connectivity as the middle mile. AI is useless without the fibre and subsea capacity to move data to and from it. Africa is now served by 77 active or under-construction submarine cable systems; the 2Africa cable alone delivers 180 Tbps across 46 landing points. The bottleneck has moved inland, to national backbones and metro fibre.
Where the capital is going — and when
We expect the coming wave to concentrate across four layers, each with a distinct geography and timeline.
The first is data centres and AI-ready colocation, the headline act of 2026 to 2030. Africa’s data-centre investment need is projected to exceed $5 billion by the late 2020s, and momentum is already visible. In July 2026, Google announced plans for Africa’s first Digital Exchange Port in South Africa’s Eastern Cape — a high-capacity gateway linking subsea cables, ISPs and cloud platforms. The near-term hubs are South Africa, the region’s anchor market and AI-diffusion leader, alongside Kenya and Nigeria, with secondary markets such as Ghana, Côte d’Ivoire and the DRC representing the higher-return frontier where live capacity is still a fraction of potential.
The second is fibre and subsea — the middle-mile rebuild of 2026 to 2028. Terrestrial fibre reach remains strikingly uneven: South Africa reaches roughly 58% of the population within 10 km of a fibre route, Kenya 46% and Nigeria 44%, but the DRC just 12% (Hamilton Research, 2025). Capital is now closing that gap. The DRC has partnered with China’s Genew Technologies on a $1.5 billion fibre-optic route along the Congo River; Vodacom’s $2.1 billion move to take control of Safaricom consolidates the region’s most valuable connectivity and mobile-money franchise; and Zanzibar’s TZS 300 billion fibre programme with Yas Tanzania shows the model reaching island and rural economies.
The third is power and adjacent energy infrastructure, the enabler of 2027 to 2031. This is the layer most under-covered by traditional telecoms investors and, in our view, the most mispriced. The $10–20 billion energy requirement tied to data-centre demand will draw in independent power producers, renewables developers and development-finance institutions. We expect the most sophisticated data-centre deals of the late decade to be structured as integrated compute-plus-power platforms.
The fourth runs continuously through 2030: mobile, 5G and the AI application layer. 5G remains early, with around 30 million subscriptions across the region today, forecast to reach 420 million by 2030 and about 28% of connections by 2031, and 29 countries having launched (Ericsson Mobility Report, June 2026). Total African ICT investment already runs at roughly $18.5 billion a year. Layered on top, the application economy — fintech, agritech, and sovereign and enterprise AI services — is where much of the value ultimately accrues, and where local players can leapfrog.
In short: 2026 to 2027 is the near-term window for AI-ready colocation retrofits, metro fibre and the marquee subsea and exchange gateways; 2028 to 2030 is the heavy build phase for hyperscale-adjacent capacity and integrated power; and beyond 2030 is the maturation of a genuine African AI-services layer.
The reasons for caution — and why they favour disciplined capital
None of this is a straight line. The World Bank has estimated that achieving universal, quality broadband across Africa by 2030 requires around $100 billion, including roughly 250,000 new 4G base stations and 250,000 km of new fibre. No single actor closes a gap of that size, and three risks recur in almost every mandate we advise on.
The first is cost of capital and country risk. Financing for African infrastructure carries equity risk premia several multiples above OECD benchmarks; deal structuring, not just deal sourcing, determines returns.
The second is power and land. Reliable power and permitting timelines routinely make or break data-centre economics, particularly outside South Africa.
The third is skills and regulation. Uneven spectrum policy, data-localisation rules and a thin AI talent pool can quietly erode a thesis that looked strong on a spreadsheet.
These risks do not argue against investing. They argue for investing with rigorous, evidence-based diligence — which is exactly where the returns diverge between informed and uninformed capital.
Digitalthings’ role in this market
Digitalthings’ role in this market is not to sell hype. It is to test whether the investment case matches the physical, commercial and regulatory reality.
Digitalthings is a boutique strategy and investment advisory firm specialising in digital infrastructure, cloud, platform and IoT solutions, broadband networks and services. Our work spans market assessment, commercial and technical due diligence, transaction advisory and strategy, and it is grounded in a continuously updated market-intelligence platform — the same platform that produced the figures in this article. Over more than a decade we have supported operators, tower and fibre platforms, private-equity and infrastructure funds, development-finance institutions and governments across South Africa, Kenya, the DRC, Angola, Malawi and beyond.
For investors and operators entering the AI-infrastructure wave, that combination of on-the-ground transaction experience and a proprietary, verifiable data platform lets us do three things. We size the opportunity with evidence rather than anecdote — country-by-country demand, capacity, competitive and regulatory reads drawn from a sourced database, not desktop guesswork. We de-risk the transaction through commercial and technical due diligence that pressure-tests power, connectivity, demand and cost-of-capital assumptions before capital is committed. And we shape the strategy — market entry, asset positioning and value-creation plans built for where the AI-era demand curve is actually heading.
The AI supercycle will build hundreds of billions of dollars of infrastructure this decade. The question for Africa is not whether it participates, but who captures the value — and that will be decided by the quality of the analysis behind each decision. If you are evaluating a data-centre, fibre, power or connectivity opportunity in Sub-Saharan Africa, we would welcome a conversation.
Figures are drawn from Digitalthings’ market-intelligence platform and its underlying sources, including ADCA, Mastercard, the Microsoft AI Economy Institute, GSMA, Ericsson, Hamilton Research, the African Energy Chamber and the World Bank, together with deal reporting from June and July 2026. Data is current as of July 2026.