🔗https://ig.ft.com/ai-data-centres/
The world is witnessing an extraordinary transformation—a global arms race for AI infrastructure that is reshaping the physical and economic landscape of technology. As companies chase breakthroughs in artificial intelligence, they are also racing to secure the one thing that makes AI possible: enormous computing power housed in massive data centres.
What was once a back-office function—running servers and storing data—has become the front line of AI competition. Industry giants like Microsoft, Amazon, Google, Meta, and Nvidia are investing hundreds of billions of dollars to build the next generation of data centres. These facilities are no longer just warehouses of servers; they are complex, energy-intensive ecosystems designed to run the world’s most powerful AI models.
In 2025 alone, global capital expenditure on data infrastructure is expected to top $350 billion, with some analysts projecting nearly $400 billion by 2026. The surge in investment is driven by the exponential growth of AI workloads—particularly the training and deployment of large language models (LLMs), which require vast amounts of energy and data throughput.
These new AI data centres are on a completely different scale from their predecessors. In some cases, they occupy hundreds of acres, draw gigawatts of power, and are built with custom cooling systems and dedicated power lines. Some are being designed to function like entire self-sustaining cities, optimized not just for computing but for long-term operational efficiency.
The land rush is global—but concentrated in hotspots like Northern Virginia, Texas, Ireland, and the Nordics. However, these areas are starting to reach their limits. In Virginia, for instance, power grid constraints are delaying new builds by up to seven years. That’s creating fierce competition for available space and energy, and pushing companies to seek alternative locations—often with looser regulations and more abundant land.
Behind the scenes, private equity firms and infrastructure investors are pouring money into AI data centre projects. One example is a $4 billion, 190-megawatt campus in Dallas backed by KKR, built in coordination with a natural gas producer to secure a reliable energy source. These financing structures show just how strategic—and capital-intensive—this new digital infrastructure has become.
But this race comes with a cost. The article raises sharp questions about energy consumption and environmental impact. AI models, especially during training, consume vast amounts of electricity. Some models are estimated to use as much energy as hundreds of homes per day. To power their operations, companies are resorting to electricity from nuclear plants, fossil fuels, and natural gas pipelines—despite public commitments to net-zero goals.
Water use is another major concern. Many of these hyperscale facilities rely on evaporative cooling, which can require millions of litres of water per day. In drought-prone areas like Arizona, Texas, and parts of Spain and India, this is raising alarms about sustainability and long-term viability. The article details how some communities are now competing with AI for access to basic natural resources.
To mitigate backlash, companies are exploring more sustainable approaches. Cooling systems using sea water, building in colder climates, and experimenting with next-gen chips that are more energy-efficient are part of the solution. But the article suggests that these improvements are not keeping pace with the scale of expansion.
What’s driving all this? The belief that AI will transform industries and economies—and that whoever controls the infrastructure will control the future. This conviction is not just technological; it’s geopolitical. The U.S., China, and Europe are all making strategic bets on AI capacity as part of their national digital sovereignty agendas.
Yet critics argue the industry may be building beyond what is necessary. Some analysts worry about “stranded assets”—facilities that are built but never fully utilized. Others point to the lack of regulation or accountability, especially around environmental disclosures. Many countries don’t even require reporting on water or energy usage from data centre operators.
The article also touches on the tension between private ambition and public oversight. Local governments often welcome the economic stimulus of a data centre—new jobs, tax revenue, construction contracts. But they may not fully understand the long-term environmental costs or infrastructure burdens until it’s too late.
At the heart of the article is a profound question: Are we building a better future—or just a bigger infrastructure footprint? The scale, speed, and secrecy of the AI race make it hard to tell. But what’s clear is that data centres are no longer just the pipes of the internet—they are the engines of a new AI-powered world.