The most transformative technology of the age is revealing a profound paradox. India’s data centres, the engine rooms of its artificial intelligence ambitions, consumed an estimated 150 billion litres of water in 2024-25, a figure projected to more than double by the end of the decade. Each AI prompt, each training run for a large language model, carries a hidden hydrological price tag that is only now entering public consciousness. Analysts in New Delhi note that this environmental burden sits awkwardly alongside the nation’s desire to become an AI superpower, a tension mirrored in the quiet admission, aired on World Environment Day, that the sector’s energy appetite and electronic waste trails are similarly spiralling. The pattern is global: the AI revolution is thirsty, power-hungry, and increasingly difficult to reconcile with sustainability pledges.

Yet the allure of machine intelligence remains irresistible, particularly for developing economies seeking competitive leaps. In Nairobi, Kenyan banks are wiring AI into their core operations, pairing transaction data with predictive analytics to anticipate customer needs—flagging early financial stress or recommending insurance the moment a property purchase completes. Indonesia’s private and public sector leaders, gathered at a recent AI Leadership Exchange in Jakarta, are urgently debating how to embed agentic AI not as a pilot but as an engine of enterprise advantage. And in the United Arab Emirates, banks are recasting financial literacy as a national priority, deploying AI-driven digital tools to build a more resilient, future-ready society. These advances, viewed from London, suggest that for many nations the immediate productivity and inclusion gains outweigh longer-term environmental qualms.

The calculus becomes more complex when set against other pressing financial and infrastructural demands. Kenya’s aviation sector is straining to reclaim its role as East Africa’s hub, with Kenya Airways targeting $1.5 billion in fresh capital and the Central Bank forecasting a wider current account deficit—now three percent of GDP—driven by costly fuel imports from a volatile Middle East. Simultaneously, the government is injecting billions of shillings into the state medical supplies agency to plug donor funding gaps, and amending banking laws to offer distressed lenders more generous emergency loans. These fiscal manoeuvres, analysts in Nairobi argue, highlight a delicate triangulation: countries must finance immediate stability, long-haul connectivity, and the digital future all at once, often with thin buffers.

Across the Atlantic, the backlash against the physical footprint of AI is hardening into policy. American communities, from Virginia to Oregon, are increasingly imposing moratoriums or outright bans on new data centre construction, rejecting the noise, water demand, and grid strain that accompany the server farms powering Silicon Valley’s leading models. A recent investigation mapped over 1,400 such facilities already dotting the country, with many more planned. Viewed from Washington, this local resistance signals a broader legitimacy crisis for an industry that has promised dematerialised, clean innovation. The irony is sharp: while Africa is often looked to as a source of learnings on operating under constraint, its businesses have long navigated the kind of infrastructural uncertainty that data centre critics now fear is being exported to the Global North.

The convergence of these trends points to an inflection point. The global conversation is no longer about whether artificial intelligence will reshape economies but under what terms. From Nairobi to Jakarta, policymakers are beginning to ask not just how to win the AI race, but how to ensure the prize is not a future too resource-starved to enjoy the victory. As the technology’s thirst and heat intensify, the measure of success may shift from raw computational might to the wisdom with which it is deployed.