The Megawatt Is Africa’s Data Centre Bottleneck

The headline out of Unstoppable Africa 2026 was a permanent African seat on the United Nations Security Council. The more consequential number for the continent’s digital economy was $300 million, and it was attached to solar mini-grids.

The Nigeria Distributed Renewable Energy Fund reached commercial launch on the sidelines of the 81st UN General Assembly. It is co-managed by the Nigeria Sovereign Investment Authority and Africa50, with Sustainable Energy for All as a partner and $25 million in IDA financing from the World Bank.

The fund will finance mini-grids, standalone solar systems and storage for communities and businesses that the national grid does not reach.

It sits under Mission 300, the World Bank and African Development Bank programme targeting electricity access for 300 million Africans by 2030. Organisers said the initiative has passed 50 million connections across 40 countries.

Read as an energy story, it is a rural access fund.

Read against what the digital infrastructure industry was debating in Nairobi last week, it becomes something else: a large pool of African capital targeting the input that increasingly determines where compute can be built.

What Nairobi was actually arguing about

At ITW Africa in Nairobi, one of the central questions was the cost of a megawatt.

Andile Ngcaba put the build cost of data centre capacity on the continent at around $12 million per megawatt, against a target closer to $5 million. JLL puts the global average build cost at roughly $11.3 million per megawatt for 2026.

The gap that matters is not simply the construction cost. It is everything wrapped around unreliable power: redundancy designed for a grid that cannot always be depended on, diesel and storage capacity sized for outages that are assumed rather than exceptional, and a cost of capital priced for both.

An operator does not pay a premium because concrete is necessarily more expensive in Nairobi or Lagos.

It pays a premium because reliable power is not always available, and every layer of engineering required to compensate for that uncertainty lands on the balance sheet before a single rack is energised.

That is why the scoreboard matters.

The IMF counts about 160 data centres in Africa, roughly 5.5 per cent of the global total by facility count. The African Actors of Data Center Association’s 2026 economic report puts the continent’s share of global capacity at about 0.6 per cent, or roughly 360 megawatts.

The distance between those two numbers is the heart of the argument.

Africa has a respectable number of buildings and a much smaller share of global capacity. Counting facilities flatters the position. Counting power does not.

The same sentence, from two rooms

The IMF’s AI Preparedness Index ranks sub-Saharan Africa lowest and identifies power shortages as a primary constraint. It projects that AI could add about 4 per cent to the sub-Saharan economy over a decade, conditional on electricity supply, internet access and skills.

Roughly 600 million people in the region still lack access to electricity.

In New York, African Union Commission Chairperson Mahmoud Ali Youssouf listed the continent’s binding constraints as affordable energy, infrastructure, access to finance, skills, technology and standards.

Energy came first.

UN Deputy Secretary-General Amina J. Mohammed, speaking about deploying AI at scale, said Africa has the talent and ideas and needs the opportunity, infrastructure and investment.

Neither was speaking specifically to a data centre audience.

Both, however, described a constraint that sits at the centre of the industry’s growth equation.

Access and industrial load are one market

The temptation is to treat a distributed renewable energy fund as a rural electrification instrument and therefore separate from hyperscale ambition.

The market mechanics suggest otherwise.

What unlocks private generation at data centre scale is closely related to what unlocks it at village scale: the ability of private producers to sell power to private buyers under viable commercial and regulatory frameworks.

Raxio chief executive Robert Skjødt has framed the requirement in those terms. South Africa’s 2022 Energy Action Plan, which opened the way for private power purchase agreements, provides a reference case. Kenya and Nigeria are working through comparable changes.

A fund that finances independent generation and helps establish viable contracting, credit structures and regulatory pathways at one scale can help build the infrastructure and market mechanisms required at another.

Alain Ebobissé of Africa50 described the vehicle as a model that can scale across Africa. Aminu Umar-Sadiq of NSIA said Nigeria’s distributed renewable market is investable, credible and ready to operate at scale.

Investability is precisely what the continent’s compute build needs.

The gap that does not close on its own

Kenya illustrates another side of the equation.

Telkom Kenya’s Kibati told the ITW Africa panel in Nairobi that roughly 90 per cent of the population sits under network coverage while usage is closer to 28 per cent.

Built capacity that goes unused is a familiar African outcome. It can emerge in power as it does in connectivity: capital deployed against demand that cannot yet support the investment.

Compute inverts that risk.

Data centre load is large, contracted and creditworthy. It is also less dependent on proximity to population centres than many traditional forms of electricity demand.

In generation terms, it can become the anchor customer that distributed renewable projects have historically lacked.

That creates an overlap between two markets that are often discussed separately.

A fund built to serve underserved communities and an industry searching for firm, affordable megawatts are looking for many of the same things: bankable contracts, viable private generation models and regulators capable of supporting them.

What to watch

Three markers will show whether this week’s announcement touches the digital infrastructure question.

First, whether any DRE Fund transaction names an industrial or digital offtaker rather than a household or SME portfolio.

Second, whether Nigeria and Kenya complete private-to-private power purchase frameworks that data centre operators can actually finance.

Third, whether the ADCA capacity figure moves.

At roughly 0.6 per cent of global capacity, that is the number that needs to change, not simply the facility count.

The $12 million megawatt is not merely a construction problem.

It is an electricity problem wearing a construction invoice.

Until the megawatt becomes cheaper and more certain, Africa can continue adding data centre buildings without closing the much larger gap in computing capacity.