AI is devouring water and energy: Is the time coming when we will have to choose between Data Centers and homes?

2026-08-18 19:03:10Lifestyle SHKRUAR NGA REDAKSIA VOX
Illustrative photo

The Guardian / In a summer marked by drought in the UK, wildfires in Europe and increasingly extreme temperatures around the world, John Harris arrives in Slough, about 30 kilometers west of London.

Slough was once known to many Britons as the drab suburb that became the backdrop for the hit sitcom The Office. Today, the landscape has changed. Among the traditional businesses, huge metal-clad buildings have risen, structures that are almost indistinguishable from the outside but inside which thousands of servers work.

They are Data Centers, the physical infrastructure upon which the digital economy and, increasingly, the artificial intelligence revolution are being built.

But behind the seemingly immaterial world of AI lies a very physical reality: servers require energy, produce heat, and need tremendous amounts of water to cool down.

This is putting Data Centers at the center of two of the biggest problems of the summer of 2026: energy and water.

Harris first visits a data center owned by the American company Equinix. Outside the building, there are about 100 staff and visitor vehicles, while almost nothing indicates what goes on inside. What can be heard, however, is a low industrial noise that continues uninterrupted.

Opposite, in an almost perfect irony, are the offices of Thames Water.

A young employee named Louise tells Harris that residents are aware of the high water consumption of AI infrastructure. She had seen material on Instagram warning that during heat waves, when servers need more cooling, their demand for water and energy also increases.

And the problem doesn't end with water.

A few minutes away is Britwell, a housing estate built after World War II for London families who had lost their homes to bombing raids.

Sheryl, a mother who lives there, tells Harris that there is a constant humming sound from the surrounding infrastructure at night. But she has also noticed something else: since the new centers were built, the area seems hotter.

This perception of residents is not without a scientific basis.

Preliminary research published this year by researchers at the University of Cambridge suggests that data centers can create a local “heat island” effect. According to them, temperatures in surrounding areas can increase by an average of about 2 degrees Celsius, while in some cases the effect can reach up to 9 degrees Celsius.

The owners of the Slough industrial estate have stated that they have not seen conclusive evidence that the activity of Data Centers there is causing local temperatures to rise.

But the size of the industry is extraordinary.

Slough today has the largest concentration of Data Centers in Western Europe: up to 40, and more are planned.

There are several reasons why this particular town has become a magnet for industry. The industrial area has benefited from a more streamlined planning regime; it is located close to London, which is particularly important for financial transactions where data transmission speed is important; and, by a coincidence of geography and infrastructure, Slough is located on the cables that connect Britain to the United States and Ireland.

The companies that own or operate these centers are not necessarily names familiar to the public: Equinix, Digital Realty, and Iron Mountain.

But their clients are among the most powerful companies on the planet: Amazon, Google, Oracle and Microsoft.

And what is happening in Slough is not just a British debate.

In the United States, data centers are already becoming a political issue ahead of the midterm elections.

Donald Trump has taken a clear stance in favor of their development, arguing that communities must be convinced of the advantages they bring and that technological development cannot be stopped.

But local resistance is growing.

In July, New York became the first US state to impose a one-year moratorium on the construction of new “hyperscale” data centers. Legal battles are underway in other parts of the US and pressure is mounting for similar restrictions.

Democratic politician Abdul El-Sayed, winner of the Michigan Senate primary, has brutally summed up the mood of a portion of the American electorate: people simply hate data centers.

Even in Texas, a state traditionally friendly to business and the energy industry, the Republican governor has called for a ban on their development in rural areas. Authorities have also ruled that new centers cannot connect to the state's power grid before regulators review their water and energy consumption.

The reason is simple: AI requires energy on a scale that was difficult to imagine just a few years ago.

More and more American data centers are installing their own gas turbines to produce energy.

Even in the UK, leaders of the energy regulator Ofgem have announced that around 100 planned data centers intend to use similar systems.

In the US, hubs are putting pressure on existing power grids, where coal and oil continue to be used. Diesel-powered backup generators are also a normal part of the industry's infrastructure.

There is an effort to develop “green” data centers, but the industry faces a fundamental problem: servers must operate 24 hours a day, 365 days a year.

Solar and wind power do not produce continuously. For this reason, part of the industry argues that renewables cannot yet be the main source of supply for centers that need to operate without any interruption.

But if energy is a problem, water could become a much more explosive political conflict.

Water UK, the organisation representing water companies in Britain, has warned that the government's current forecasts for future water needs are seriously flawed.

The reason is surprising: Data Centers are not properly included in these forecasts.

And the quantities involved are not small.

A Data Center can consume up to 5 million gallons of water per day.

This is approximately equivalent to the daily consumption of 50,000 people.

Here Harris raises one of the strongest questions of his analysis.

What will happen during future droughts?

If citizens are asked to limit their water use, will data centers continue to consume millions of gallons every day to keep AI servers cool?

This potential conflict is becoming even more serious due to the ambitions of the British government.

Whitehall's official target is to triple the number of data centers in the UK by 2030.

In fact, these structures have now been classified as critical national infrastructure, giving them a strategic importance comparable to other essential infrastructures.

But this brings up another conflict: Data Centers vs. Homes.

A report by the London Assembly has revealed that the extraordinary energy demands of data centers have already contributed to delays in new housing projects in the British capital.

A concrete case is the redevelopment of the former Truman Brewery site in East London.

The plans supported by local authorities focused on the construction of around 350 new homes.

But the approved project envisages a Data Center that will process City of London financial transactions, office space and only 44 apartments.

The difference is dramatic.

In a city that has suffered for years from a housing shortage and high prices, digital infrastructure is beginning to compete directly with the space and energy needed to house people.

And here Harris's analysis reaches the dilemma that could define the politics of the coming years.

The AI ??revolution is usually presented as something abstract: algorithms, language models, the cloud, automation.

But in reality AI needs giant buildings, land, cables, energy and millions of liters of water.

The more its use increases, the more these needs will compete with people's needs.

At the end of his visit to Britwell, Harris meets a woman named Jodie. She tells him that her brother had seen a claim on TikTok that, in the event of war, Slough could be targeted precisely because of the concentration of Data Centers.

The claim may sound dramatic, but it shows how much the perception of the city has changed.

On the horizon, giant buildings continue to hum.

And after all the promises of the artificial intelligence revolution, Harris leaves the reader with a much more mundane question:

When energy, water, and land are not enough for everything, what will governments choose to build — homes for people or servers for machines?


Video