PFAS and the Regulatory Turn: The Coming Reformulation

Now to the force reshaping the coolant market from the outside. The engineered fluorinated fluids that make two-phase immersion so effective belong, chemically, to a vast family of substances known as per- and polyfluoroalkyl substances, or PFAS, widely called forever chemicals because the carbon-fluorine bond that makes them so stable and useful also makes them extraordinarily persistent in the environment. That persistence, and mounting concern about their accumulation in water and living tissue, has put the whole class under regulatory pressure that is now arriving at the data center's door.

The signals have been unmistakable. In Europe, five national authorities jointly proposed a broad restriction on the manufacture and use of PFAS as a class, one of the most sweeping chemical restrictions ever contemplated, working through the European Chemicals Agency. In the United States, environmental regulators have moved to designate certain PFAS compounds as hazardous substances and to set drinking-water limits measured in parts per trillion, an astonishingly low threshold that signals how seriously the risk is being taken. The most consequential single event for our industry came from a manufacturer, not a regulator: a major producer of these fluids announced it would exit PFAS manufacturing entirely, removing from the market some of the exact engineered coolants that two-phase immersion had been built around. When the supplier of a critical input announces its own withdrawal, that is not a distant policy debate; it is a supply cliff with a date on it.

I want to be precise about who this hits and who it spares, because sloppy generalization causes bad decisions. The regulation strikes hardest at two-phase immersion and at any single-phase immersion using fluorinated dielectric fluids, because those approaches depend on exactly the chemistry under restriction. It largely spares water-based direct-to-chip cooling, whose water-glycol coolants are not PFAS. This asymmetry is a major, and I think underappreciated, reason direct-to-chip consolidated its lead at the very moment liquid cooling scaled: the alternative carried a regulatory cloud that made conservative operators and their lawyers nervous. Timing matters in markets, and the fluid controversy arrived at precisely the wrong moment for immersion.

The industry's response is a scramble to reformulate. Chemical companies are developing next-generation dielectric fluids that deliver the non-flammable, low-boiling behavior two-phase cooling needs without the persistent fluorinated backbone, and researchers are revisiting hydrocarbon and other non-fluorinated chemistries for single-phase immersion. This is genuinely hard. The properties that made PFAS fluids excellent, inertness, non-flammability, precise boiling points, tunable across a range, are difficult to reproduce all at once in a benign molecule. And every candidate replacement must run the long qualification gauntlet described in the previous lesson: years of material-compatibility and stability testing before an operator will trust it under a rack of irreplaceable accelerators. Reformulation is not a quarter's work. It is a multi-year transition, and the deadline pressure and the validation timeline are pushing against each other.

For the three audiences I am speaking to, the implications differ but all point the same way. For the investor, coolant chemistry is now a diligence item: a cooling company heavily committed to restricted fluids carries a risk that belongs in the model, and firms developing compliant replacements sit on real opportunity. For the operator, the message is to favor coolant chemistries with a clear regulatory runway, to write flexibility into long-term supply contracts, and to avoid architecting a facility around a fluid that may be legally constrained before the building is fully depreciated. For the policy audience, this is a live illustration of environmental rules colliding with strategic infrastructure: the same persistence that makes these chemicals an environmental problem made them an engineering asset, and society is now negotiating that trade-off in real time. The lesson is not that regulation is an obstacle to route around. It is that in liquid cooling, the molecule is now a strategic variable, and the operators who treated it as an afterthought are the ones being forced into expensive, hurried changes.