Google Used 29% of One Oregon Town's Water. It Fought to Hide That.
Paris Marx's 'Hyperscale' traces how Ireland's 2021 grid revolt foreshadowed the US data center fight — Google used 29% of one Oregon town's water while fighting disclosure.

Why it matters
- Google consumed 29 percent of The Dalles, Oregon's water and kept the figure secret for nearly a year.
- Dublin passed a data center moratorium in 2021 after its grid operator said capacity was exhausted.
- Marx's book 'Hyperscale: The Ambition and Excess of Big Tech's Data Empires' publishes as data centers shape US midterm races.
- Google cleared over 800 acres of Finnish forest without permits for a €13 million data center project and apologized.
- Dotcom-era precedent: Equinix bought distressed data centers after the bust — but AI-era chip cycles may erode that playbook.
By James Calloway
Google consumed 29 percent of The Dalles, Oregon's water supply — and spent the better part of a year persuading the city council to keep that figure secret, according to Paris Marx, author of the new book Hyperscale: The Ambition and Excess of Big Tech's Data Empires. The community eventually got the numbers. By then, the precedent was set.
Marx, a longtime Big Tech critic and host of the Tech Won't Save Us podcast, has been tracking data center conflicts for years. His book arrives as data centers have become a live issue in US elections — one that could decide key midterm races in less than two weeks, he notes. The book tours a longer history: antiwar protests against American computers during the Vietnam War, resistance in Ireland and Chile, and a community-led alternative in New Zealand.
"I certainly didn't predict that it was all going to kind of come to a head as my book was coming out," Marx said in an interview this week.
The timing may be accidental. The pattern he documents is not.
What did Ireland show before the US fight began?
Marx traces his interest to 2022, when Ireland's energy grid strain became impossible to ignore. Data centers were consuming huge quantities of electricity. Dublin had already passed a moratorium in 2021 — notably, imposed not by politicians but by the grid operator. "Hey, there are too many data centers. We do not have the grid capacity to connect any more of them," Marx recounts the operator saying.
The industry's response is the part Marx calls predictive. Blocked from Dublin grid connections, operators looked elsewhere in Ireland. Then they tapped the gas network to run generators. Eventually they began building dedicated gas power plants to serve their data centers directly, bypassing the grid altogether.
There is a caveat worth flagging: this is Marx's account, drawn from his research and advocacy perspective, not an industry response. But the mechanism he describes is concrete. The Irish grid is public — only the public utility can build electricity lines. Data center companies, Marx says, lobbied for legislation that would allow private companies to build their own electricity cables. That raises equity questions about who controls essential infrastructure.
Meanwhile, operators were obtaining emission licenses from the Irish Environmental Protection Agency to burn fossil fuels. The agency itself warned that climate targets would become much harder to achieve.
The US is now replaying this script at larger scale, with companies racing to build massive private gas plants for data centers. "When I started to see that happen in the US, I wasn't surprised. I'd already seen the story play out before," Marx says. "But the scale of it is staggering, and how quickly it's escalated."
Are we in an AI bubble — and what happens if it pops?
Marx's position is blunt: "I think we're in an AI bubble." The harder question is what a collapse looks like when so much physical infrastructure is already committed.
The build-out faces real bottlenecks. There are issues sourcing turbines, chips, and other components. Marx asks whether the data centers and gas plants will even be completed before investors reassess AI demand. That is a genuine uncertainty, not a forecast — and he presents it as one.
There is a historical precedent, and it cuts both ways. The dotcom boom and bust produced an excess build-out of fiber capacity, and also of data centers — though at a far earlier technological stage. Many data center companies went under. Survivors like Equinix bought up facilities cheaply and waited for demand to return. The strategy worked.
Marx argues this time differs in one crucial respect. AI data centers are planned around chips that must be cycled out within a few years. "Even if the data center is there, what makes it most valuable is potentially not going to be useful by the time that demand returns," he says. The shell survives; the value inside may not.
The gas plants are the stickier problem. "Inevitably, even if the data centers don't come, then those gas plants will exist. They will be looking for new customers to justify their existence," Marx says. "I'm sure that people are not building them with the intention to just shut them down." In plain words: the power infrastructure will outlive the AI rationale, and will find something to burn for.
Is there a non-extractive alternative?
Marx closes his book with New Zealand, where a community-led effort built a more accurate speech-recognition model for the Māori language using local cloud providers to host the data. The example is small. That is partly the point — it demonstrates a different relationship between computation and community, without claiming it scales.
His policy prescription for governments is two-track. First, firmer regulations on data center development. Second — and this is the part usually missing from US debate — regulations on the AI products themselves.
"We had a lot of promises about what generative AI was going to deliver that haven't been realized," Marx says. "There has also been a long list of harms that have arisen from these technologies that we were not told would come along with them."
He faults governments for buying "the narrative that generative AI is still what Sam Altman sold it to be in early 2023 rather than what it has become today." The claim is arguable — AI products have millions of users and real revenue — but the gap between promise and delivery is documented enough to take seriously.
The core question Marx poses is resource accounting: how much computation do we actually need, and what justifies it?
"Does what we're getting from what Silicon Valley is creating really justify the resources that we are collectively sacrificing in order to realize it?" he asks. "If we actually did the assessment, I think that we would very clearly find that there's no way to justify what we're doing right now — from a financial perspective, but also from a social and environmental one."
That is a strong claim, and Marx offers it as judgment rather than measurement. No such comprehensive assessment yet exists in either direction. In plain words: the book's argument is that the burden of proof has shifted, and the industry has not met it.
Is Europe the next flash point?
The Irish climate warnings are not isolated. Joel Khalil, covering the EU for the same outlet, reports that Europe's thirst for AI is colliding with member states' climate goals. He expects data centers to become a political flash point in Europe as countries go to the polls next year, mirroring the US.
"The fascinating thing about the European data center build-out is the collision of competing interests: national-level policy butts up against local priorities, ambitions for economic growth stand in conflict with climate objectives, and planning bureaucracy foils attempts by AI companies to rapidly expand their fleets," Khalil says. "It's hard to see how Europe could possibly have it all."
Other developments this week reinforce the trend:
- The AI and data center boom has boosted demand for a surprising commodity: gray paint, according to The New York Times.
- Texas's massive data center build-out threatens years of state progress cutting air pollution, the Texas Tribune reports.
- Google apologized for clearing more than 800 acres of forest in Finland without the required environmental permits, part of a €13 million data center project. The Guardian reports the backlash could bleed into Finland's elections next spring.
The Finland case is the sharpest signal. When a company clears 800 acres without permits and frames the remedy as an apology, enforcement — not just regulation — is the variable to watch.
What happens next
Marx's book makes one wager explicit: the US is not the first mover in this fight, only the largest. Ireland shows the industry's playbook when blocked; New Zealand shows an alternative that few governments are pursuing. Whether US midterm contests driven by data center opposition produce regulation, moratoriums, or merely delay is the open question — and Marx's history suggests the gas plants get built either way.
Original: nytimes.com
More from James Calloway
Show full bio
News editor covering people, launches and day-to-day practice in artificial intelligence and machine learning at AI In Context.
369 articles
Related articles
- Google's Emissions Jumped 50% in 2024 to Feed Generative AI
- Data Center Backlash Goes Global as AI Buildout Hits Local Resistance
- Amazon warns communities: block AI data centers, risk the economy
- Australian Officials Call Datacentre Backlash a US Import: 'We Are Not the United States'
- AI Policy Becomes Early Flashpoint for 2028 Democratic Field