The data centre boom is creating a new economic trade-off. Governments competed to attract this investment, but the infrastructure needed to support it is creating political, environmental and economic resistance. What changed?

From prestige projects to public opposition

Only a few years ago, governments were trying to attract AI data centres, viewing them as symbols of innovation and economic growth. Today, many of those same projects face fierce opposition from local communities concerned about electricity, water consumption and environmental costs. Has the race to build AI infrastructure begun to generate a global backlash?

Understanding data centres

Data centres are physical facilities or buildings that house a group of networked computer servers to run and deliver applications and services. They also store and manage enormous amounts of information related to them. The advancement of cloud computing in the early 2000s has transformed how we store data. Rather than dedicating machines to single customers, most modern data centres use cloud architecture, pooling and virtualising the hardware underneath. There are close to 9,000 data centres worldwide, a figure forecast to pass 10,000 by 20301. Among the most advanced is Microsoft's Fairwater site in Mount Pleasant, Wisconsin, which the company describes as the most powerful AI data centre it has built2. Introduced in the US Congress in March 2026 was a proposal called the AI Data Center Moratorium Act.3 Its key purpose is to pause all new or expanding AI data infrastructure until thorough laws are passed encompassing AI safety, sustainable practices, and worker rights. The bill indicates a turnaround in public opinion. A few years ago, data centres used to be prestige projects. However, now, public support has begun to erode. Across 42 US states, there have been 142 protests on a single day alone4, challenging the expansion of data centres that have grown over the years. Protests included signs saying, “Build community, not data centres” and “We thrive on water not data.” In fact, the support is so little that only 14% of those surveyed in a June Reuters/Ipsos poll would support a data centre in their area4. Recurrently, the issue of how much water AI uses is alarming to individuals and is frequently raised as a concern. With local temperatures rapidly rising, beating records year on year, individuals are worried it will magnify the consequences of global warming.

AI's physical limits

Backlash like this proves how AI has created new competition for scarce resources. In the UK, 2.5% of electricity is consumed by data centres and this usage is expected to rise four-fold by 20305. This is a drastic increase. The true figure may be higher still. The International Data Center Authority puts UK and US consumption at around 6% each, more than double the government's estimate6. That two credible measures of the same thing disagree by a factor of two says something in itself about how poorly the sector's footprint is understood. Machine learning models require constant processing power that far exceeds traditional web browsing or basic cloud storage, creating a strain on resources. AI is often described as a digital revolution. Ironically, its biggest constraints are physical. Electricity, water, land and skilled labour have become scarce, and sectors increasingly compete for them.

Many are concerned with whether building data centres are sustainable and compatible with the government's commitments of net zero carbon emissions by 2050. While they drive massive electricity demand, they also strain local power grids, create reliance on fossil fuels and raise significant environmental concerns. Construction and demand often outpace green energy, forcing grids to burn gas to prevent blackouts. At the moment, we have not eliminated the use of carbon-heavy energy, so this is a burden we have to carry. Ed Miliband, then Energy Secretary, said that the environmental impacts of data centres in the UK are “inherently uncertain”7, leaving the public in the dark.

Labour markets and the hidden costs

Resource pressures extend beyond electricity. Labour markets are also beginning to feel the effects of the AI infrastructure boom. Taking a look at the labour market, researchers at Georgia Tech found that data centres employ very few permanent workers in rural areas and generally import specialised services rather than domestically using them8. In fact, the employment benefits are often far smaller than expected. In Iowa, two massive projects carry exorbitant costs, one of $576 million and another of $750 million. Yet, legally, their requirement is to create at least 61 permanent jobs combined8. Are these projects really worth it if so few jobs are created for such an expensive cost? Not only are many data centre jobs temporary, the boom has also created a “two-speed construction market,” where very expensive AI projects jack up costs for other construction, while taking highly skilled labour with them. This creates severe skilled labour shortages.

Is the backlash against data centres happening everywhere, and is it causing economic tension in other countries?

Singapore and Ireland: warning signs

Data centre growth in Singapore is a double-edged sword to put it simply. It is the biggest data centre hub in South-east Asia with significant levels of investment into artificial intelligence and critical infrastructure. That status rests on its economic stability and its highly skilled workforce. However, the data centre sector takes up around 7% of the country’s total electricity usage9. An unchecked AI and electricity boom would create higher energy bills, delays in connecting new power to local grids and an increased reliance on fossil fuels. In addition, their clusters of high-rise buildings and lack of green space have created land constraints. Due to all these concerns, Singapore placed a temporary pause on new projects in 2019, allowing Johor, in Malaysia, to attract much more data centre investment. This has sparked tension over lost tech business. Tech moguls have voiced that constraining digital infrastructure risks Singapore’s status as Southeast Asia’s leading digital hub. Ireland has also become a cautionary tale for neighbouring countries. Data centres already account for a substantial share of the country’s electricity demand, 23% in 2025 according to its Central Statistics Office, up from 22% a year earlier and just 5% a decade ago10, forcing policymakers to confront difficult choices between attracting AI investment and maintaining grid reliability.

A new competition for AI investment

Unlike traditional manufacturing, AI investment is remarkably mobile. While Singapore’s policy was designed to ease pressure on electricity supplies and limited land, it made developers look beyond its borders. This highlights a broader shift in the global economy. Countries are no longer competing solely on tax rates or labour costs, but increasingly on their ability to provide the infrastructure required for artificial intelligence: modern power grids, high-speed digital networks and efficient planning systems. These have become strategic economic assets.

Not whether, but how

The backlash against data centres is unlikely to halt the AI revolution. If anything, demand for computing power will continue to grow. The more pressing question is whether governments can balance that demand with growing public concern over electricity, land, water and employment. Countries that restrict development may protect local resources but risk losing investment elsewhere. Those that embrace it may secure technological leadership at the cost of increasing pressure on infrastructure. The challenge is no longer whether to build AI data centres, but how to build them sustainably without undermining the very economic benefits they promise.

Every government wants the economic benefits of data centres, but few want to bear the costs of hosting the infrastructure.

Footnotes

  1. ABI Research, How Many Data Centers Are There and Where Are They Being Built? (opens in a new tab), July 16, 2024. ↩

  2. Microsoft, Inside the world's most powerful AI datacenter (opens in a new tab), September 18, 2025. ↩

  3. Brookings Institution, New evidence on data center employment effects (opens in a new tab), May 4, 2026. ↩

  4. Reuters, Data center opponents stage 142 protests across 42 US states (opens in a new tab), July 18, 2026. ↩ ↩2

  5. House of Commons Library, Data centres: planning policy, sustainability, and resilience (opens in a new tab), May 27, 2026. ↩

  6. The Guardian, Datacentres using 6% of electricity supply in UK and US, research says (opens in a new tab), May 13, 2026. ↩

  7. BBC News, Miliband says climate impact of data centres is uncertain (opens in a new tab), February 27, 2026. ↩

  8. Futurism, The Economics of Data Centers Creating Jobs Are So Bad That They Sound Like a Joke (opens in a new tab), July 13, 2026. ↩ ↩2

  9. fDi Intelligence, The dark side of south-east Asia data centres (opens in a new tab), February 27, 2025. ↩

  10. Central Statistics Office, Data Centres Metered Electricity Consumption 2025 (opens in a new tab), July 7, 2026. ↩