The Cheapest Place on Earth to Run an AI Data Center Is Not Where You Think
The completion of the Grand Ethiopian Renaissance Dam is one of the most consequential infrastructure achievements in African history. With an installed capacity of over 5,000 megawatts, GERD positions Ethiopia as a serious energy power and immediately raises the obvious question: what do we do with the surplus?
The conventional answer has been to sell it. Build transmission lines to Sudan, Kenya, Tanzania, and eventually push cables all the way to South Africa and Europe. Monetize the electrons. Join the regional power grid and collect the proceeds.
I think that answer is wrong. Not because exporting electricity is a bad idea in principle but because we are thinking too small.
The Infrastructure Trap
Building transmission lines from western Ethiopia to South Africa spans over 5,000 kilometers of terrain, borders, and competing political interests. The capital cost runs into the billions. The wholesale price of electricity on regional African markets rarely exceeds a few cents per kilowatt-hour. By the time you account for transmission losses, maintenance, right-of-way negotiations, and financing costs, the return on that infrastructure investment is modest, slow, and fragile.
There is a better use of that electricity and a better use of that capital.
Bring the Compute to the Power
The global Artificial Intelligence industry is facing one crisis above all others: it is running out of electricity. AI data centers training the large language models and AI systems that are reshaping every industry consume hundreds of megawatts continuously. The largest hyperscale AI campuses being built today consume more power than some small nations. In the United States, data center operators are being turned away by utilities that simply cannot meet demand. In Europe, countries like Ireland and the Netherlands have placed moratoriums on new data center connections.
The world’s AI industry needs what Ethiopia has. The question is whether Ethiopia recognizes that.
Instead of building power lines to carry electricity to markets thousands of kilometers away, Ethiopia should be building the infrastructure to bring those markets here in the form of AI High-Performance Computing data centers, built directly alongside our hydroelectric projects.
Why Ethiopia Wins on Every Metric
The case is not simply about power. It is about the convergence of every cost input that determines where a data center gets built.
Electricity from GERD is projected to cost between one and two and a half cents per kilowatt-hour among the cheapest generation costs on the planet. For comparison, industrial power in the United States costs seven to twelve cents. In Germany, it reaches twenty cents. An AI campus in Ethiopia consuming 500 megawatts saves between two hundred and four hundred million dollars per year in power costs alone versus a comparable American operation.
Then there is cooling. Data centers generate enormous heat, and managing it is the second-largest operating cost in this industry. Ethiopia’s high-altitude highlands maintain temperatures between 10 and 22 degrees Celsius year-round conditions that allow facilities to cool themselves largely through outside air, without energy-intensive mechanical systems. This further multiplies our power advantage.
Land and construction costs are a fraction of what they are in Virginia, Singapore, or Frankfurt. Our workforce is young, growing, increasingly educated, and available at compensation levels that reflect our economic stage without compromising quality. Every major cost input points in the same direction.
The Renewable Imperative
There is one more dimension that I believe makes this argument not just compelling but urgent. The world’s largest technology companies Microsoft, Google, Amazon, Meta have made binding commitments to power their operations with 100% renewable energy. Meeting those commitments in developed markets is genuinely difficult. Solar and wind are intermittent. Nuclear faces political resistance. Grid constraints are severe.
Hydroelectric power from GERD is dispatchable renewable energy. It flows on demand, around the clock, every day of the year. A data center in Ethiopia powered by GERD can make a credible, auditable claim of 100% carbon-free power the gold standard that every major AI company in the world is chasing and cannot find at our price point.
As carbon pricing mechanisms expand globally, this is not a soft advantage. It is a structural, long-term competitive position.
A Lesson from History
When China opened its economy and combined cheap labor, available land, and state-directed infrastructure into an offer the world’s manufacturers could not refuse, it captured three decades of industrial growth and lifted hundreds of millions of people out of poverty. The analogy I keep returning to is this: the AI data center is to the 21st-century economy what the factory was to the 20th. Ethiopia has the opportunity to be the country that builds them and to do so on terms that no other nation can currently match.
This is not about abandoning electricity exports. Regional power cooperation remains valuable. It is about recognizing that the same electron that sells for a few cents on a regional grid can generate ten to twenty times more economic value if it powers a GPU cluster training the next generation of AI.
The Window Is Open But It Will Not Stay Open
The global AI industry is making location decisions right now. Power purchase agreements are being negotiated. Land is being secured. The countries and regions that move quickly will lock in anchor tenants, shape the regulatory environment, and capture the economic spillover the engineers, the startups, the capital, the supply chains that follows serious technology infrastructure.
Ethiopia has everything it needs: the power, the climate, the land, the people, and a moment in history where all of it is suddenly exactly what the world’s most important industry is looking for.
The choice is whether we see it.