OpenAI pledged to build responsibly in Texas, Bloomberg reported that top AI labs still have not disclosed their emissions, and a recent Ezra Klein Show episode chronicled the growing movement against data centers.
Paul Roetzer's read: the fight is less about water and electricity than about an industry that has lost the public's trust.
OpenAI sent a letter to Texas Governor Greg Abbott committing to build AI infrastructure in the state responsibly by:
paying its own way
protecting residential and small business customers from added costs
conserving water
reporting its electricity and water use accurately.
Abbott says OpenAI will comply with the data center standards he set for Texas this summer.
Meannwhile, Bloomberg reported that top AI companies, including OpenAI and Anthropic, have not disclosed their greenhouse gas emissions or published sustainability reports, even as both prepare for IPOs. That may change in November, when a California law known as SB 253 starts requiring companies with more than $1 billion in revenue that do business in the state to report emissions. Anthropic is already working with Watershed, a carbon accounting platform, to measure its footprint.
The backlash itself was the subject of a recent episode of The Ezra Klein Show featuring writer Jasmine Sun, who covers the movement against data centers, the massive facilities packed with the computer servers that run AI. An overwhelming majority of Americans oppose having one built near their homes, New York Governor Kathy Hochul has paused new hyperscale data centers, the largest class of these facilities, for up to a year, and more than 100 similar proposals are in play nationwide. A Cato Institute commentary pushed back, arguing bad policy is the real problem and noting data centers used just 0.3% of the US public water supply in 2023.
On Episode 232 of The Artificial Intelligence Show, SmarterX founder and CEO Paul Roetzer looks at why the backlash keeps growing.
0.3% - Share of US public water supply used by data centers in 2023, according to the Cato Institute
100+ - Proposals nationwide to pause data center construction
1 - Year, which is the maximum length of New York's pause on new hyperscale data centers
$1 billion - Revenue threshold for California's SB 253 emissions reporting
$20 million - Approximate tax revenue for one small community from a Microsoft data center, per the Klein episode
The questions changed first. Roetzer has spoken publicly about AI dozens of times a year since roughly 2015. "Up until last year, I could count on one hand how many questions I probably got about data centers and the environment," he says. "Every talk I do, regardless of who it's for, I get at least one or two now."
The secrecy backfired. On the Klein episode, Sun described a dynamic that destroys local trust: When a data center comes to town, everyone from trade leaders to city council members signs nondisclosure agreements. In small communities, word gets out anyway. Residents complain at council meetings, members legally cannot respond, and people conclude their leaders are hiding something. And they sort of are., Roetzer says.
Water usage seems misunderstood by the public. Roetzer points out that nobody wants any industrial facility in their backyard, an Amazon warehouse or a solar farm, for example. And according to Sun, much of the water panic is outdated: New data centers are almost all closed-loop systems that recycle water the way air conditioning does, using a fraction of what a golf course uses. Sun also noted watching YouTube or Netflix consumes more energy and water than querying ChatGPT.
But the electricity concern is real. The grid lacks capacity for rapid data center growth, and when companies promise bills will not rise, nobody believes them. The episode also flagged real positives, including major tax revenue and high-paying jobs that Roetzer says can last seven to 10 years or longer.
The benefit is invisible to the average person. A car factory makes cars. But when a data center arrives, the average person asks what they get out of it. "You get to use ChatGPT" is not a welcome benefit at the moment. "The AI industry seems to have far more of a reputation and trust problem," says Roetzer. "For years, the Silicon Valley leaders communicated to the public about AGI and solving math problems and this future of abundance when they should have been making the benefits real and tangible to the average consumer and worker." (AGI, or artificial general intelligence, is AI that can match or beat a human's intellect.) About six weeks ago, those leaders simultaneously changed their tone on jobs, which Roetzer sees as a commuications strategy that will not cut it. "They have a communications and PR problem," he says. "But to Ezra's point, they have a product problem."
Even the data centers' own employees are worried. Roetzer recently did a private event for roughly 400 executives at one of the companies building data centers. "I got questions about the environmental impact of their own technology from the people within the companies building it," he says. Questions expressing concern about their local communities and jobs.
"Just because you work at a company that wants to be AI forward doesn't mean all your employees are on board with it. And there could be a number of different reasons, including some of them just really have concerns about this stuff."
— Paul Roetzer, founder and CEO of SmarterX, Episode 232 of The Artificial Intelligence Show
Environmental concerns have moved inside companies. Employees who believe AI harms their communities might quietly resist using it at all, which makes any AI transformation effort harder. Roetzer's prescription starts inside the company with communication and transparency. Many people building and deploying this technology don't understand its actual footprint any better than the public.
Roetzer's advice for leaders? Educate, don't advocate. Acknowledge the electricity problem honestly, correct any outdated water claims with facts, and be straight about the tradeoffs.
California's SB 253 will produce the first real numbers in November. Once large AI companies must report emissions, the debate gets a factual baseline, and disclosures from OpenAI and Anthropic ahead of their IPOs will show whether the industry gets ahead of the story or gets dragged into it.
State moratoriums on data center builds. If more governors follow Hochul on where AI infrastructure gets built and how fast, this becomes a complicated state-by-state political fight.
Fear or mistrust of AI is a top-four barrier to adoption, cited by 29% of professionals in the 2026 State of AI for Business Report, alongside training, awareness, and time. A smaller share of respondents, 3%, named environmental impact when asked what concerns them most about AI. The trust gap Roetzer describes in local communities shows up inside organizations, too.
The full report, built on more than 2,100 responses from professionals across roles, functions, and industries, maps how organizations are navigating adoption, governance, training, and employee sentiment. Read the full report →