The global data centre market is entering a new phase. For years, location decisions were mainly driven by fibre routes, latency, enterprise demand, cloud regions and proximity to major financial centres. Those factors still matter. But the next constraint is increasingly more basic: where can operators secure enough reliable, affordable and low-carbon power?
That question is becoming more urgent because of AI.
AI changes the data centre conversation in two ways: training and inference
Training is the process of building or improving a model. It requires very large clusters of GPUs, extremely high power density, advanced cooling and the ability to run intensive workloads for long periods. Training facilities can be less latency-sensitive, because the model does not always need to be trained close to the end user. They need scale, power and cooling above all else.
Inference is different. Inference is what happens when a trained model is used in real life: a bank runs fraud detection, a call centre uses a chatbot, a doctor queries a diagnostic tool, or a farmer uses an AI service in a local language. Inference is repeated constantly, often close to users, applications and data. It is more latency-sensitive, more distributed and increasingly tied to data sovereignty, enterprise cloud adoption and local regulation.
This distinction matters for Africa.
The continent is unlikely to become the world’s largest AI training hub in the near term. The largest frontier training clusters will continue to favour markets with massive grid capacity, deep hyperscaler ecosystems, specialist engineering depth and mature procurement channels. But Africa can be highly relevant for regional inference, sovereign AI workloads, enterprise AI, public-sector AI and selected training environments where power, connectivity and regulation align.
The global pressure is already visible
The International Energy Agency expects the United States, China and Europe to remain the largest regions for data centre electricity demand, with the United States alone adding around 240 TWh of data centre electricity demand by 2030 compared with 2024 levels. Goldman Sachs Research forecasts that global data centre power demand could increase by as much as 165% by 2030, driven heavily by AI. McKinsey estimates that meeting AI-related data centre demand could require trillions of dollars of investment by 2030, with AI workloads becoming a major driver of new capacity.
This is not only a technology issue. It is a power system issue.
In mature data centre hubs, operators are competing for land, grid connections, substations, water, permitting capacity and public acceptance. Traditional European hubs such as Frankfurt, London, Amsterdam, Paris and Dublin face land and power constraints, creating pressure for operators and hyperscalers to consider alternative locations. In the United States, data centre electricity demand forecasts vary widely, making grid planning more complex.
Africa’s strategic opening
Africa starts from a much smaller base. The Africa Data Centres Association has estimated that Africa has only a small share of global data centre capacity, with one industry estimate putting the continent at around 307 MW, less than 2% of the global total. That underdevelopment is often presented as a weakness. In many respects, it is. Grid reliability, utility creditworthiness, regulatory certainty, foreign exchange exposure, water availability and permitting remain real constraints.
But the low starting point also creates a strategic opening.
Africa is not locked into the same legacy data centre geography as Europe or North America. New capacity can be planned around renewable energy zones, grid upgrade programmes, submarine cable landings, metro fibre routes, cloud regions, enterprise demand and national digital policy from the outset.
In other words, Africa does not need to copy yesterday’s data centre model. It can build a more power-aware and AI-aware model.
South Africa is already showing parts of this direction. Teraco, now part of Digital Realty, has started construction of a 120 MW solar PV plant in the Free State to supply its data centres and support grid upgrades. NTT’s Johannesburg 1 data centre offers 12 MW of IT critical load once fully built out, reflecting continued demand in the Johannesburg / Gauteng market. Africa Data Centres, part of Cassava Technologies, operates a carrier-neutral network across key African markets and is positioned around cloud, carrier, enterprise and hyperscale demand. IXAfrica in Kenya, Africa Data Centres’ East African footprint, and new AI-ready infrastructure announcements all point to a more distributed African compute ecosystem.
The operator landscape needs to be understood carefully
Vantage, Equinix, Digital Realty / Teraco, NTT and Africa Data Centres all operate in the data centre sector, but they are not identical businesses.
Vantage is strongly hyperscale-campus aligned, serving hyperscalers, cloud providers and large enterprises, with large campuses designed for long-term scale.
Equinix combines enterprise colocation and interconnection with xScale hyperscale facilities, making ecosystem density and cloud connectivity central to its model.
Digital Realty, through its global platform and Teraco in Africa, spans colocation, interconnection, cloud on-ramps and large-scale deployments.
NTT Global Data Centers is one of the world’s largest data centre providers, with a global footprint and a South African facility designed for enterprise, colocation and cloud-related demand.
Africa Data Centres is one of the continent’s most important regional platforms, with carrier-neutral facilities supporting cloud providers, carriers, enterprises and regional interconnection.
This distinction matters for investors. A hyperscale campus has different requirements from an enterprise colocation facility. An interconnection hub has a different commercial model from a single-anchor AI compute campus. An inference-oriented facility serving local banks, governments, health systems and content platforms has different latency and network requirements from a frontier training cluster. Treating these as one category leads to weak market sizing and poor investment analysis.
Four layers of near-term opportunity
For Africa, the strongest near-term opportunity may sit across four layers.
First, core cloud and enterprise colocation in the major hubs: Johannesburg, Cape Town, Nairobi, Lagos, Cairo and selected North African markets.
Second, regional interconnection facilities close to submarine cables, internet exchanges and dense metro fibre.
Third, AI inference infrastructure that supports local enterprise AI, public-sector AI, fintech, health, education, agriculture and African-language applications.
Fourth, selective AI training or fine-tuning environments where renewable power, land, fibre and policy align well enough to make the economics credible.
This is where the energy angle becomes decisive. AI training requires very large, concentrated power loads. Inference may be more distributed, but it still increases data centre density and power consumption. Both require better grid planning, stronger wheeling and power purchase frameworks, more bankable renewable energy structures and closer coordination between utilities, regulators, municipalities and data centre developers.
Africa’s advantage is not that power is easy. It is not. The advantage is that parts of the continent still have the opportunity to design the next wave of digital infrastructure around power from the beginning. That could include renewable-backed campuses, co-investment in grid upgrades, edge and metro inference nodes, AI-ready colocation, sovereign cloud environments and regional compute corridors linked to submarine cables and cross-border fibre.
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 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. Its work includes data centre due diligence across multiple African markets, data centre asset carve-out feasibility, multi-country total cost of ownership modelling, infrastructure funding advisory, transaction support and technical-commercial assessment of African data centre assets. The nature of our work is helping investors, operators and institutions understand whether digital infrastructure assets are technically sound, commercially viable and bankable in African conditions.
That experience is increasingly relevant because Africa’s data centre opportunity is becoming more complex. It is no longer enough to ask whether there is demand for cloud services. The better questions are:
- Is power available, reliable and scalable?
- Can renewable energy be contracted on bankable terms?
- Is the grid connection credible and deliverable?
- Is there enough fibre diversity and route resilience?
- Is the site suitable for AI-era cooling and density?
- Is demand anchored by cloud, enterprise, public-sector or content customers?
- Is the regulatory environment clear on data sovereignty, licensing and cross-border data flows?
- Does the facility serve training, inference, enterprise colocation, interconnection — or a mix of all four?
Africa’s opportunity is real, but it will not be uniform. Johannesburg, Cape Town, Nairobi, Lagos, Cairo and selected cable-adjacent or renewable-rich locations will each follow different paths. Some will become cloud and enterprise hubs. Some will become interconnection points. Some may support AI inference at scale. A smaller number may support more power-intensive AI training or fine-tuning workloads.
The direction, however, is clear. As global data centre growth becomes constrained by power, land and permitting, Africa’s relative position changes. The continent’s renewable energy potential, young digital demand profile, improving fibre ecosystems and underbuilt data centre base create a credible investment case — if projects are structured carefully.
Africa’s next digital infrastructure advantage will not be defined only by towers, fibre or cloud regions. It will be defined by the intersection of power, data, AI and capital. That is where Africa may be better positioned than many assume.