There is a joke that every Nigerian household understands without needing it explained: the light goes out mid-sentence, somebody shouts, and twenty minutes later somebody else shouts “up NEPA!” Forty years of that has made us, collectively, some of the most electricity-literate people on earth. We know what a stabiliser does. We know what a bad transformer sounds like before it goes. We know the price of diesel the way other countries know the price of bread.
So here is something worth sitting with. The richest companies in the history of the world — the ones building artificial intelligence, the ones with more cash than most national treasuries — have spent the last eighteen months running into exactly the same wall. Not a shortage of money. Not a shortage of engineers. Not even a shortage of computer chips, which is what everyone assumed the problem would be.
They cannot get enough electricity. And more specifically, they cannot get the equipment that moves electricity from where it is made to where it is used. Sound familiar?
America Just Discovered What Nigeria Has Known Since the 1980s
For roughly twenty years, the most boring fact in American economics was that electricity demand did not grow. Appliances got more efficient, factories moved offshore, and the grid coasted. Utilities built almost nothing new because nobody needed anything new.
That era is finished. US power demand is now rising for the fourth consecutive year — the first stretch like that since 2007 — and the reason is data centres. The US Energy Information Administration expects American generation to rise about 2.4% in 2026, to roughly 4,327 billion kilowatt-hours.
Now hold that number against ours. Nigeria has 13,625 MW of installed generation capacity on paper, according to the Nigerian Electricity Regulatory Commission. In April 2026, only 4,286 MW of it was actually available to dispatch — a plant availability factor of 31%. Nearly seven of every ten megawatts we supposedly own sat idle. Our all-time peak, set in March 2025, was 5,801.84 MW and has not been beaten since. The Transmission Company of Nigeria says it can wheel 8,700 MW, but what actually reaches consumers rarely clears 5,000.
Run the arithmetic and the United States generates something on the order of a hundred times more electricity than Nigeria does, for a population only about half again as large as ours. And they are the ones panicking about a shortage.
That is not a comment on Nigeria. It is a comment on how much electricity artificial intelligence eats.
The Real Bottleneck Is Not Chips. It Is Transformers.
This is the part that should make every Nigerian reader sit up, because it is our exact problem, scaled up and dressed in a suit.
You cannot energise a data centre — or a housing estate, or a market — without transformers, switchgear and high-voltage cable. That is the unglamorous iron that turns generated power into usable power. And the world has run out of the ability to make it fast enough.
By 2026, buyers were being quoted roughly 128 weeks of lead time on a power transformer. Generator step-up units run near 144 weeks. The largest high-voltage units have been pushed out four to five years. In 2020, the same equipment took about one year. High-voltage circuit breakers are now quoted around 125 weeks.
Two and a half years. To buy a transformer. If you have the money in hand.
The root cause is a material almost nobody outside the industry has heard of: grain-oriented electrical steel, the specialised alloy that transformer cores are wound from. There are very few plants in the world that make it well, and building a new one takes years. On top of that, China controls an estimated 60% of global transformer manufacturing capacity and is largely locked out of the American grid on national-security grounds — so the world’s biggest buyer is bidding for the same non-Chinese factory slots as everyone else. Industry analysts now warn that equipment shortages alone could delay or cancel close to half of all planned US data-centre developments.
I wrote a full breakdown of this for our research site, tracing the shortage through every layer of the supply chain: why electricity, not silicon, turned out to be the real AI bottleneck. The short version is that the constraint moved, and most people did not notice it move.
Why a Shortage in Texas Shows Up on a Lagos Electricity Bill
It would be comfortable to file this under “American problem.” It is not.
Nigeria does not manufacture most of its transformers. We import them, and China alone supplied about 38% of our transformer import value in 2024, with India and Turkey next. When the entire developed world starts bidding for grain-oriented electrical steel, copper windings and skilled transformer labour at the same time, the price of that input rises for everybody — including the buyer in Nigeria who has no leverage and no alternative supplier.
You can already see it in the numbers on the ground. A certified, IEC-compliant 100 KVA distribution transformer — the kind that serves one street — now runs roughly ₦5.8 to ₦6 million in 2026. Meanwhile our distribution companies posted around ₦2.4 trillion in financial losses, which means the entities responsible for replacing failed transformers are the entities least able to afford them at global prices.
That is the squeeze in one sentence: the equipment we most urgently need just became a globally contested commodity, at exactly the moment our distributors ran out of money.
It also explains, more honestly than most policy statements do, why Nigerians have voted with their wallets. Solar and battery storage now account for something close to 20% of national electricity usage, and the trend is accelerating. That is not fashion. That is millions of households and businesses independently concluding that waiting on the grid is a worse bet than buying their own generation. Every inverter installed in Surulere is, in a small way, the same trade the hyperscalers are making when they sign private power deals in Pennsylvania.
The Money Followed the Bottleneck — Not the Story
Here is where it turns into an investing lesson, and I think it is a genuinely useful one even for people who never intend to buy a share.
Through 2024 and 2025, the obvious trade was to buy the power producers — the companies that own the plants and sign the supply contracts. Everybody could see data centres needed electricity, so everybody bought the electricity sellers. Those stocks got expensive. Very expensive.
Then the bottleneck moved, and the market re-priced brutally. Through early August 2026, the companies that make the equipment and build the grid were up 40% to 72% for the year. The power producers everyone had piled into were down 12% to 25%. The speculative next-generation nuclear start-ups — the ones with the best story and no revenue — were cut roughly in half. Same theme. Opposite outcomes. The only variable was where in the chain you were standing. (Figures here come from our own equity research desk at AlphaEdge Hub, as of the 5 August 2026 close.)
The reason is simple enough to state in a sentence: a contract to sell something is a promise, but a factory that makes the thing nobody else can make is a moat. When there is a genuine shortage, the profit does not go to whoever signs the demand. It goes to whoever owns the constrained supply.
Four Things Worth Taking From This
- The obvious beneficiary is usually already priced in. By the time a theme is on every front page, the headline names carry the expectations. The money is usually one or two steps up the supply chain, in the boring company nobody can pronounce.
- Physical constraints beat financial ones. You can raise capital in a week. You cannot build a transformer plant in under three years. When a shortage is physical, it lasts — and it is far more predictable than a shortage caused by sentiment.
- A good story is not a good business. The pre-revenue nuclear developers had the most exciting narrative in the sector and lost roughly half their value. Revenue, backlog and contracts are what survive a re-pricing. Vision is not a financial statement.
- Energy costs feed everything else. Global competition for power equipment eventually reaches your diesel bill, your service charge, your food prices and your rent. It is not a niche technology story. It is an inflation story wearing a technology costume.
The Bottom Line
The world is discovering that intelligence — artificial or otherwise — is a physical product. It requires steel, copper, cooling and a very large amount of reliably delivered electricity. Every credible projection of the AI era now runs into a wall of transformers, substations and high-voltage cable that simply does not exist yet and cannot be conjured up quickly.
Nigeria has been living inside that constraint for four decades. We built an entire parallel economy around it: generators, inverters, stabilisers, and a national instinct for knowing when to charge your devices. What has changed is not our situation. What has changed is that the wealthiest corporations on the planet have arrived at the same wall, with vastly more money, and found that money alone does not move it any faster.
Which makes the last twenty years of Nigerian improvisation look less like a national embarrassment and more like a preview. We were simply early.
About the author: Georgi Kuzmanov is a senior equity analyst with over thirteen years in equity research and a Master’s in Financial Engineering from Columbia University. He publishes daily market analysis, sector research and quantitative strategy work at AlphaEdge Hub.
Data sources: US Energy Information Administration generation outlook; Nigerian Electricity Regulatory Commission operational factsheets; Transmission Company of Nigeria; company filings and industry lead-time surveys. Market figures as of the 5 August 2026 close. This article is for information and education only. It is not investment advice, and nothing here is a recommendation to buy or sell any security.
