Nigeria Could 10x Its AI Gains. The Grid Is the Only Thing Standing in the Way.

A new IMF paper just put a number on what good policy could unlock — and the gap is staggering.

---

The IMF has published a paper on artificial intelligence in Sub-Saharan Africa, and buried inside the technical language is a finding that should alarm Nigerian policymakers. Under current conditions, the region's AI adoption could lift productivity by just 0.2% over the next decade. Fix the infrastructure, and that number jumps to 2.1%. The technology is the same in both scenarios. What changes is electricity.

That tenfold gap is not a projection gap. It's a policy gap.

---

Nigeria Is in a Rare Position — For Better and Worse

Most Sub-Saharan African countries are classified by the IMF as "low exposure" to AI disruption, because most of their workers do manual, context-specific tasks that AI cannot easily touch. Nigeria is one of only six countries in the region whose job composition looks more like an emerging market economy than its neighbours. South Africa, Mauritius, Botswana, Namibia, and the Seychelles are the others.

That classification cuts both ways. It means Nigerian workers are more exposed to AI displacement. It also means Nigeria stands to gain more from AI adoption than almost any other economy in the region, ranking in the IMF's top five for potential productivity upside.

The problem is that potential and outcome are separated by one stubborn variable.

---

86% of Nigerian Firms Own a Generator. That's Not a Flex.

Roughly 78% of firms across Sub-Saharan Africa experience routine power outages. Those outages cost an average of 8.4% in lost sales, well above the global average of 5.2%. For AI infrastructure, which requires continuous, high-performance computing, a brief outage doesn't just inconvenience operations. It can interrupt a training run entirely and erase hours of compute time.

Nigerian firms have adapted. Eighty-six percent own or share a generator, the highest rate among the continent's three major data centre markets. Kenya sits at 65%, South Africa at 63%. The adaptation works, after a fashion. But generator power runs on imported fuel, costs more than grid electricity, and drags down total factor productivity. The IMF is careful to note that this affects public finances too, since petroleum subsidies added around 0.8% of GDP to Nigeria's fiscal burden in 2022 alone.

Running an AI economy on diesel is not a strategy.

---

The Data Centre Boom Has a Power Problem Attached to It

Sub-Saharan Africa holds under 1% of the world's AI-relevant data centre capacity. The IMF models a scenario where the region reaches just 0.5% of the global total by 2035, which sounds modest until you see what it implies: meeting that target would add electricity demand equal to roughly 10% of the region's entire installed generation capacity in 2023.

Nigeria is already in the middle of a data centre construction wave, partly driven by the CBN's June 2026 directive requiring financial transaction data to be stored locally by January 2027. Operators including OADC, Rack Centre, and Equinix have confirmed the pipeline is real. Billions are moving into Nigerian data centre capacity.

The IMF's question is pointed: will that investment deliver an AI productivity dividend, or create expensive server space running on diesel? Rack space and fibre alone won't close the gap between 0.2% and 2.1%. The grid has to move with the infrastructure.

---

Talent Is Leaving Because the Compute Isn't There

Power is the headline constraint, but the skills picture compounds it. Across Africa, only 5% of AI talent has reliable access to the high-performance computing needed for serious research. A practitioner in a G7 country might retrain and test a model every 30 minutes. Their counterpart in Lagos could wait up to six days for a single run.

Nigeria's Three Million Technical Talent programme, launched in late 2023, is a real attempt to address the skills deficit. Training more people faster is necessary. But as the IMF notes, it won't retain those people unless local computing capacity grows alongside the pipeline of talent. Skilled workers move toward compute. That's not sentiment, it's rational.

---

What This Requires

The CBN's localisation directive gave Nigeria a compelling reason to build data centres. The IMF paper is asking a harder question about what happens after the buildings go up.

The answers involve power purchase agreements, grid-extension mandates, renewable energy targets, and regulatory institutions that can keep pace with the infrastructure being built on top of them. Kenya's $1 billion Microsoft and G42 campus, powered by geothermal energy in a country where renewables already account for over 90% of generation, shows what the upper end of this can look like.

Nigeria has the demand signal, the market size, and the policy momentum. The grid remains the unresolved sentence at the end of every ambitious paragraph.

---