LG and SK Enmove’s PFAS-Free Cooling Bet: Why LG Must Sell the Whole System
Their targets of GWP below 150 and no PFAS are table stakes for single-phase immersion. The commercial test is whether LG can sell chiller, coolant distribution and fluid together against oil majors that sell only fluid.
LG Electronics and SK Enmove have agreed to develop refrigerants, refrigeration oils and immersion-cooling materials for high-density computing. SK Enmove will develop the fluid candidates; LG will test them in cooling systems. The partners set a global-warming-potential target below 150 and want the materials to avoid PFAS. The agreement builds on work begun last year with U.S. immersion specialist GRC. LG’s release describes the programme and the October 2025 partnership notice names GRC.
The two headline specifications apply to different products, and neither sets the partners apart in immersion fluid.
The GWP target is a refrigerant specification. Global-warming potential measures the climate effect of a gas that escapes, which is why it governs the refrigerants in chillers. Single-phase immersion fluids are synthetic hydrocarbons or oils that stay liquid and do not evaporate in normal use; suppliers describe them as having no direct GWP. The sub-150 target therefore matters for LG’s chiller business, not for the fluid in the tank. It is also stricter than the EU’s coming rule for large equipment: from January 2027, the F-gas Regulation bars new chillers above 12 kW that use fluorinated gases with a GWP of 750 or more, and applies the 150 limit only to small units. A sub-150 refrigerant would put LG’s data-centre chillers well inside the law. The EU regulation sets the thresholds.
This is also where SK Enmove’s contribution is most specific. Moving a chiller to a low-GWP refrigerant changes the chemistry inside the compressor, and the lubricating oil has to be matched to the new refrigerant for miscibility, stability and material compatibility. The agreement covers those refrigeration oils explicitly. For LG, a qualified refrigerant-and-oil pairing from a domestic partner is a practical input to its chiller line. It is a less visible part of the deal than immersion fluid, but it applies to equipment LG already sells.
The PFAS target matters more in two-phase systems, where the fluid boils and condenses and some widely used products have been fluorinated, such as 3M’s Novec line. 3M’s planned exit from PFAS manufacturing by the end of 2025 hit that supply most directly, and the European Chemicals Agency is still assessing a broad PFAS restriction. GRC specialises in single-phase immersion, where fluids are usually not fluorinated in the first place. For single-phase buyers, PFAS-free is a procurement requirement every supplier already meets. GRC’s overview, the ECHA process and the 3M notice set the background.
That leaves a crowded fluid market. Shell and Castrol sell single-phase immersion fluids, and ExxonMobil markets fluids for data-centre immersion. SK Enmove brings lubricant chemistry of its own, but it enters against three oil majors with global distribution and years of qualification data. On fluid alone, the partners would have to win on heat transfer, compatibility with seals and server materials, stability over operating cycles and service support. Shell, Castrol and ExxonMobil already compete.
LG’s advantage is that it sells the rest of the cooling chain. It markets an air-cooled centrifugal chiller for AI data centres with an oil-free magnetic-bearing compressor, a 1.4-megawatt coolant distribution unit, computer-room air handlers and control software. At Data Center World in April it presented these as an end-to-end portfolio alongside GRC immersion tanks and fluid co-developed with SK Enmove. None of the oil majors sells the chiller. A buyer who takes LG’s chiller, distribution unit and a qualified fluid from one supplier gets a single party responsible for the thermal performance of the loop. LG’s chiller launch and Data Center World showcase set out the line-up.
The gap in that offer is the tank, which is GRC’s rather than LG’s. The integration therefore depends on a three-way partnership, and a buyer still deals with more than one supplier for the immersion hardware. The strongest version of the pitch would be a jointly warranted system in which LG, GRC and SK Enmove qualify one fluid against one tank and one cooling plant. Warranty is the operative word. Operators hesitate to submerge servers when it is unclear who pays if a component fails in the fluid, and a supplier group that answers that question in a contract has something to sell that a fluid datasheet cannot provide.
AI racks give operators a reason to test immersion where air cooling or direct-to-chip liquid systems limit density or energy use. Most operators are adopting direct-to-chip cooling first, because it fits existing racks and service practices; immersion asks them to change how technicians handle servers, how warranties apply and how facilities are laid out. That makes a single accountable supplier more valuable in immersion than in any other cooling choice. A fluid that meets environmental thresholds but increases downtime, complicates maintenance or damages components will not be deployed. The partners have given no pilot timetable; the release refers only to proof-of-concept and commercial review with GRC since last year. The desk expects a named-customer pilot of the combined system within 12 months if the system pitch is serious. If LG instead sells the fluid as a stand-alone product against Shell and Castrol, the partnership becomes one more entrant in a market that has plenty.