July 28, 2026
Data centers need better cooling, but fast-tracking new PFAS compounds would put communities on the wrong side of another industry experiment.
The Guardian reported on July 27, 2026, that a coalition of environmental groups is urging the EPA to reject a new PFAS compound proposed for data-center cooling. That story sits at the intersection of two major American trends: the explosive growth of AI infrastructure and the long, ugly record of industry asking communities to trust chemicals before the public understands the risks. We should not repeat that mistake because the server farms are newer and the branding is cleaner.
Data centers are real infrastructure. They power cloud services, AI tools, communications, finance, healthcare systems and the ordinary digital conveniences people now expect. They also use enormous amounts of electricity and, in many places, water. The industry is searching for cooling methods that can handle rising demand without burning through local resources. That problem is legitimate. But a legitimate engineering challenge does not justify rushing a forever-chemical answer past serious public scrutiny.
PFAS are not ordinary substances. They persist in the environment, move through water and have been linked to serious health concerns. Communities across the country have already learned what happens when chemicals are treated as manageable until they are everywhere. The cleanup costs are massive. The health fears linger for generations. The companies responsible often fight liability while taxpayers, water systems and families absorb the consequences. That history should make regulators more cautious, not less.
According to The Guardian's reporting, the chemical company Chemours wants approval for a PFAS compound intended for two-phase cooling systems. Environmental advocates say the company has understated health and climate risks and that the review is being pushed through under a Trump-era order designed to speed data-center-related approvals. If that description is accurate, it is exactly the wrong model for public decision-making. Fast growth is not an excuse for weak review. Speed is valuable only when the safety work is real.
The industry will argue that closed-loop systems limit emissions and that the technology can reduce water use. Those claims deserve to be examined. They do not deserve to be accepted on faith. Closed systems leak. Equipment ages. Workers make mistakes. Disasters happen. Disposal becomes somebody's problem. A chemical's lifecycle does not end when a company brochure stops describing it. Regulators need to examine manufacturing, use, leakage, breakdown products, end-of-life handling, climate impact and cumulative exposure before giving a green light.
This is where the politics become familiar. Corporate America often frames environmental protection as a barrier to innovation. But innovation without accountability is just risk transfer. Companies get the upside, communities get the uncertainty and public agencies inherit the cleanup. That is not pro-growth. It is lazy industrial policy. A serious country can build advanced technology while refusing to let chemical companies externalize danger.
Democrats should be careful not to sound anti-technology. The better argument is pro-competence. America needs digital infrastructure, but it also needs rules that force that infrastructure to be clean, efficient and accountable. Data centers should be pushed toward energy efficiency, water reuse, clean power procurement, transparent local reporting and safer cooling technologies. If a proposed chemical cannot clear an independent safety review, the answer is not to lower the bar. The answer is to find a better design.
Environmental justice has to be part of the decision too. Industrial risks rarely land evenly. Communities near chemical plants, data centers, waste facilities and overburdened water systems are often asked to accept exposures that wealthier communities would resist immediately. If PFAS-based cooling becomes a national solution, the harm will not be distributed by press release. It will follow facility locations, supply chains, waste routes and regulatory gaps. People living near those points deserve more than assurances.
The climate angle is also bigger than water savings. If a compound carries climate-warming risks, regulators must count those risks honestly. It would be absurd to solve one environmental pressure by deepening another. The AI boom is already challenging clean-energy goals because of rising electricity demand. Adding questionable chemical impacts to that buildout would make a difficult transition even messier.
The EPA's job is not to help the technology sector move as fast as possible. Its job is to protect public health and the environment. Sometimes that means saying yes with strict safeguards. Sometimes it means saying no. On PFAS and data-center cooling, the burden should be on industry to prove safety across the whole lifecycle, not on the public to prove harm after deployment. We have seen that movie. The ending is expensive, unfair and preventable.