Data Centers Face Growing Pushback Amid AI Expansion

The conversation around data centers in the United States is hitting a boiling point, right as Paris Marx releases Hyperscale: The Ambition and Excess of Big Tech's Data Empires. With data center projects turning into major election talking points and sparking friction in local communities, Marx’s analysis draws clear parallels to what’s already unfolded elsewhere. According to Marx, the US isn’t breaking new ground with these clashes—it’s just taking longer than most to learn from the rest of the world.

People across the US are waking up to the sheer scale and side effects of data center expansion. These sites keep cloud services and generative AI humming, but they eat through staggering amounts of electricity and water. The late-2022 explosion of tools like ChatGPT only poured fuel on this fire, pushing demand for bigger and more numerous data centers.

As Marx documents, the pitfalls of unchecked growth have already shown up abroad, and now the US finds itself at a turning point—local resistance, regulatory fights, and environmental worries are all gaining steam.

What’s New: Global Lessons on Data Center Growth

Marx’s book and recent interview lay out how the dizzying rise of data centers, powered by AI and the cloud, is straining local resources and stirring up political drama. Take Ireland: in 2021, Dublin hit pause on new data centers after the national grid operator said there was simply no more capacity. Data centers were pulling so much electricity that the grid couldn’t safely handle any more.

This didn’t come out of nowhere—back in 2019, both the Netherlands and Singapore had already stopped new data center builds, recognizing just how much pressure these facilities place on energy and water supplies.

But Ireland’s moratorium didn’t end industry ambitions. Developers simply pivoted. When Dublin became off-limits, data center projects started popping up elsewhere in the country.

Some operators eyed direct gas network connections, and soon dedicated gas power plants were in the works, built to serve data centers alone. This let companies sidestep grid restrictions, but it sparked worries about climate fallout. The Irish Environmental Protection Agency warned that reaching the nation’s climate goals would be far tougher with so much fossil fuel being burned for data center operations.

Meanwhile, companies lobbied to change the law so private firms could build their own electricity infrastructure—a big shift in a country where only the public utility had ever constructed transmission lines. These moves triggered heated debates about fairness, public oversight, and the future of Ireland’s energy system.

Marx spotlights a revealing US example: Google in The Dalles, Oregon. There, Google used 29 percent of the town’s water supply—a fact the company kept under wraps, even convincing the city council to withhold the numbers for nearly a year. When the truth came out, residents pushed back and questions about corporate control over public resources intensified.

It’s a pattern Marx sees repeatedly: the opaque, convoluted nature of data center resource use often keeps communities in the dark until the impact is unavoidable.

This script plays out elsewhere, too. In Finland, Google cleared over 800 acres of forest for a €13 million data center, leading to public outrage and government apologies once it emerged that environmental permits were missing. In Texas, a data center surge now threatens to undo years of air quality gains, with private gas plants springing up to feed new facilities.

The message, as in Ireland, is that the US and other countries are wrestling with the same basic challenge: balancing economic ambitions against mounting environmental and social costs.

Marx also draws on history: during the dot-com bubble, overbuilding fiber networks and data centers led to bankruptcies and industry consolidation. Survivors like Equinix snapped up competitors’ assets, but today’s difference is the scale and speed of AI-fueled growth—and the risk that expectations for demand could prove wildly optimistic. If the AI boom fizzles, as Marx warns is possible, we could be stuck with data centers and gas plants nobody needs, with no easy way to repurpose them.

Unlike surplus fiber from the dot-com days, today’s excess could leave deeper environmental scars and financial headaches.

Why This Matters for Your Business

If you run a business or oversee tech investments, the data center surge isn’t just background noise—it’s a risk and a potential lever. The outsize energy and water demands of AI infrastructure can trigger local backlash, regulatory curveballs, and even operational bottlenecks. Imagine your cloud provider facing a sudden freeze on new capacity or stricter environmental rules—your digital initiatives might get delayed or become pricier overnight.

The regulatory landscape—moratoriums, emissions requirements, protests—can inject real uncertainty into long-term plans.

Still, Ireland’s story shows how big players adapt: they move, lobby, or build private energy sources. That flexibility can help the giants, but smaller businesses may get squeezed by higher costs or patchy access. If private gas plants and power lines are built mainly for the largest operators, smaller firms could find themselves edged out or left with less reliable infrastructure.

And with data centers now a hot-button political issue, public sentiment and activism can feed directly into business continuity and planning. Europe’s experience—where national AI ambitions now collide with local climate targets—proves this is a boardroom issue leaders can’t ignore.

How to Respond: Practical Steps for Business Leaders

  • Assess Your Cloud Dependencies: Know where your data sits and whether those regions face supply or regulatory threats. If your provider relies on areas with scarce water or strained grids, you could be vulnerable to sudden disruptions or price hikes.
  • Factor in Environmental and Political Risks: Weigh the solidity of underlying infrastructure and the odds of new limits or charges. Keep an eye on debates about emissions, water, and grid capacity. Where moratoriums or policy changes are on the table, have backup plans for possible interruptions.
  • Engage with Local Communities: If you’re building or depending on data centers, proactive outreach can help spot and address concerns early. Being upfront about resource use and environmental impacts can earn trust and reduce the odds of backlash or regulatory headaches.
  • Monitor Industry Trends: Track news about moratoriums, grid stress, and new rules in your operational hotspots. Watch for local planning moves, utility statements, and lobbying campaigns. Understanding these signals helps you get ahead of risks before they hit operations.
  • Plan for Flexibility: Structure your contracts and tech architecture to pivot if needed—think multi-cloud setups, exit clauses, or systems that can shift between regions without major pain.

My Take: Limits, Risks, and the Real Value of AI Infrastructure

From my perspective, the US is late to a global reckoning. Other countries have already struggled with—and sometimes controlled—the fallout from unchecked data center growth. Now, the AI rush is putting huge pressure on local resources, but there’s been little open discussion about the tradeoffs.

Ireland’s story is telling: when public infrastructure maxed out, private players found workarounds, sometimes undermining environmental goals and public control. The US now finds itself in a similar (and even bigger) scramble—companies are racing to put up private gas plants and push for regulatory tweaks, echoing Ireland’s experience on a grander scale.

There’s a real chance, as Marx suggests, that this AI infrastructure frenzy is a bubble. If demand forecasts fall flat, we could be left with stranded assets—idle gas plants and server farms. Unlike spare fiber from the dot-com bust, today’s environmental and financial costs would be harder to erase.

Fossil fuel investments and the risk of higher emissions create long-term baggage that doesn’t just disappear. Meanwhile, Marx points out a different path: in New Zealand, a community-led project built a better speech-recognition model for the Māori language, using local cloud providers and prioritizing local control. It’s proof that infrastructure can be designed for social and cultural benefit, not just scale or efficiency for global tech giants.

In short, don’t assume today’s path is set. Both costs and regulations could pivot unexpectedly. The promises of generative AI still haven’t all come true, and the social and environmental downsides are just surfacing.

Governments, as Marx notes, have been slow to set real guardrails on data centers or AI products, often buying industry hype that doesn’t match reality. So, for business leaders, the watchwords are diligence, flexibility, and active engagement—with providers and with communities. The central question isn’t just how much infrastructure you can build, but how much you actually need—and what the true cost is for society and the environment.

The Bottom Line

The US debate about data centers isn’t isolated, and experiences from Ireland, Singapore, and New Zealand all point to the same thing: unchecked growth brings real business, environmental, and social risks. As the AI boom heats up, savvy leaders will look past the hype and prepare for a future that’s more constrained—and likely, more contested. For deeper reading, check out Paris Marx’s Hyperscale and coverage from The Guardian and The Texas Tribune.