The Environmental Cost of AI Data Centers in Africa

In Lagos, just outside the bustle of Victoria Island, a sleek new data center stands tall with glass walls and cooling towers humming quietly in the background. Its operators describe it as “AI-ready,” a hub designed to power everything from banking algorithms to generative AI tools. Step closer, and the less glamorous details surface. Diesel generators roar to life during power outages. Boreholes drilled nearby help offset water demand for cooling. This is the unseen reality of Africa’s bid to become a digital hub: the race to host data centers capable of handling artificial intelligence.

The Thirst of AI: How Servers Drain Our Water

Artificial intelligence, particularly large language models like OpenAI’s GPT-4 and Google’s Gemini, demand staggering amounts of computational power. Every question typed into a chatbot, every voice generated by an AI tool, sets thousands of servers into motion. Those servers heat up within seconds, and to prevent them from failing, they need relentless cooling. That cooling requires water, often drawn in thousands of liters daily, and steady electricity that can run uninterrupted for months.

Globally, researchers estimate that training a single large AI model can consume more electricity than 100 U.S. homes use in an entire year. In some cases, the water footprint of one AI training run equals the daily consumption of a small town. These numbers are abstract until they land in places already struggling to provide reliable utilities to their citizens.

The Grid That Can’t Keep Up

South Africa is a clear example. Johannesburg hosts several of the continent’s largest data centers, yet the country faces rolling blackouts that leave households and small businesses in the dark. To keep operations stable, centers turn to diesel generators, which produce both noise and emissions, further undermining the country’s push toward clean energy.

Nigeria faces a similar dilemma. Operators in Lagos build facilities with enormous backup systems, because the national grid cannot be trusted to provide uninterrupted power. The costs are passed down in diesel, in pollution, and in the higher price of digital services.

Kenya is an outlier. With geothermal energy forming a backbone of its grid, Nairobi is better positioned to power digital infrastructure sustainably. But energy is only part of the equation. Nairobi also struggles with recurring droughts, and data centers that depend on water-intensive cooling put additional strain on a city already rationing supply.

Progress vs. The Unseen Pollution

Governments celebrate the arrival of hyperscale operators like Microsoft, Google, and Amazon alongside regional players such as Raxio and Teraco. They see these centers as proof that Africa is ready to claim its place in the digital economy. Jobs are created, investors take notice, and countries like Nigeria and Kenya position themselves as continental leaders in technology infrastructure.

The symbolism is powerful. Africa is not just a market for apps and devices; it is hosting the machinery that powers the global digital economy. Yet behind the ribbon cuttings and promotional campaigns lies a more complicated picture. In water-scarce cities, locals question why servers drink while residents ration. In neighborhoods living with noise and fumes from endless generator cycles, the glow of server halls feels like a reminder of misplaced priorities.

Who Really Profits? The African Cost for Global AI

It is worth asking who truly benefits from this infrastructure. The servers may be physically located in Lagos, Nairobi, or Cape Town, but the workloads they run often serve users and clients far beyond Africa. When OpenAI’s GPT or Google’s Gemini models are trained or hosted partly in African facilities, the economic value largely flows outward, while the environmental costs remain behind.

For nearby communities, the benefits are less obvious. Jobs created by data centers tend to be specialized, often requiring skills beyond the reach of most residents. What locals experience more directly are the side effects: increased pressure on electricity, higher competition for water, and industrial noise in their neighborhoods.

Searching for Greener Paths

Some operators are aware of these tensions and are experimenting with alternatives. Liquid cooling systems, which replace or reduce water consumption, are being tested in facilities across the continent. Investments in solar and wind farms are slowly beginning to supplement grids and backup systems. A few data center operators even market their facilities as “green” to attract global clients conscious of sustainability.

But these efforts remain fragmented. The majority of facilities still rely on diesel for backup and consume large amounts of water for traditional cooling systems. The risk, critics argue, is that Africa becomes a convenient frontier for global tech giants to shift their environmental burdens while marketing the projects as a form of digital progress.

Balancing Digital Growth and Human Needs

Artificial intelligence is not slowing down. With every new release of models like Gemini or GPT, the computational demand multiplies, and with it the appetite for more data centers. Africa, eager to position itself as a player in the global tech economy, will continue to build these facilities.

The challenge is balance. Power and water are not abstract inputs; they are the lifeblood of households, farms, and industries. To prioritize servers without addressing these basic needs risks deepening inequality. If digital infrastructure is to be truly transformative for Africa, it must be designed with sustainability at its core, not as an afterthought.

The hum of AI in Africa is growing louder. The question is whether the continent can shape that growth in a way that strengthens communities instead of straining them. The decision will determine not just Africa’s place in the AI economy, but the resilience of its environment in the years to come.