In January 2015, on the show floor at CES in Las Vegas, three companies unveiled a wine cooler. It chilled a rack of bottles with no compressor, no hiss of refrigerant, and no moving gas at all. Inside was a wheel of gadolinium metal spinning past a magnet. Every time the alloy entered the magnetic field it warmed; every time it left, it cooled. The demo was the joint work of BASF, the German chemical giant; Haier, the world’s largest appliance maker; and a company almost no consumer had heard of, Astronautics Corporation of America of Milwaukee, which for fifty years had mostly built cockpit displays for military aircraft.

The message was that the compressor’s days were numbered. Refrigeration had run on the same basic trick since Willis Carrier bolted together the first modern air conditioner in 1902: squeeze a gas, let it expand, move heat. Now solid metal would do it, silently, with a magnet. The technology even had an origin myth from the Energy Department’s Ames Laboratory in Iowa, where physicists had found a “giant” magnetocaloric effect in a gadolinium-silicon-germanium alloy in the late 1990s.

Eleven years later, you cannot buy a magnetic refrigerator. But something did break loose in the patent record, and it points somewhere the 2015 demo did not.

The number

Across the full US grant record there are 247 patents on caloric cooling, the family of solid-state effects where a material dumps or absorbs heat when you hit it with a field. The single largest holder is not BASF, not Haier, not a startup. It is Carrier, the company Willis Carrier founded, with 16 grants.

Here is the part that matters. Of those 16, fifteen are electrocaloric and one is elastocaloric. Zero are magnetic. The direct corporate descendant of the man who invented air conditioning has quietly assembled the largest US caloric-cooling portfolio in the country, and it skipped entirely the approach that got two decades of press and a wine cooler at CES.

To see why, you have to stop counting and read what these patents actually describe. When you do, the 247 grants sort into four different bets on the same idea, and the smart money has moved off the one everyone reported on.

Four ways to make a solid cold

The shared mechanism is worth stating plainly, because it is the thing that makes this a story and not a word cloud. Certain materials sit right at a phase transition. Nudge them across it with an external field and their internal disorder changes so sharply that they must give up heat or pull it in. Pump that cycle and you have a heat pump with no compressor and no gas. The only question is which field you use to do the nudging.

Magnets. This is the 2015 approach, and in raw patent volume it still dominates: since 2021 alone, more than 200 US grants touch magnetic refrigeration. BASF holds nine, all on the materials themselves, iron-phosphorus and manganese-iron-phosphorus-silicon alloys arranged in graded “cascades” so different layers hit their transition at different temperatures. Astronautics holds a dozen, and reading them tells you what went wrong: they are elaborate machines, drums and rotary beds and fluid valves engineered to shuttle a heat-transfer fluid past the alloy fast enough to matter. The physics needs gadolinium or its cousins, which are expensive, and it needs large permanent magnets, which are heavy and expensive. You end up with a refrigerator that costs like a car.

The company that proved this was Cooltech Applications of Strasbourg. It built the largest magnetocaloric prototype ever made, a 15-kilowatt unit shown in 2018, and then went bankrupt. Its patents were sold off to a French intellectual-property fund in 2019. Twenty years of the giant magnetocaloric effect, and the flagship company’s assets ended up in a licensing portfolio.

Electric fields. This is Carrier’s bet, and it is a different animal. Instead of a gadolinium wheel and a magnet, the working material is a thin plastic film, a copolymer of vinylidene fluoride and trifluoroethylene. Apply a voltage across the film and its molecular dipoles snap into alignment, releasing heat; drop the voltage and it cools. It is the same relaxor-fluoropolymer family that already sits inside piezoelectric sensors and ultrasound transducers, which means the manufacturing base exists. Carrier’s fifteen electrocaloric grants describe how to stack these films into modules, pattern the electrodes, and route fluid between them. Three more electrocaloric patents belong to United Technologies, Carrier’s former parent, which spun the HVAC business off in 2020. Read together, it is one continuous R&D program that has been running quietly for a decade. No magnets. No rare earths. A film you could imagine printing.

Mechanical strain. Bend a nickel-titanium wire and it warms; let it snap back and it cools. This is elastocaloric, and its most cited champion is Ichiro Takeuchi at the University of Maryland, whose nitinol-tube device published in Science hit temperature swings of 22.5 degrees Celsius, ranking, by his group’s accounting, in the top 15 percent of every caloric cooler built in four decades. Nitinol is the same shape-memory alloy used in arterial stents and eyeglass frames, so again the material is already manufactured at scale. The Irish company Exergyn has four recent US grants on shape-memory-alloy heat pumps, and Carrier’s lone non-electric caloric patent is a “thermally driven elastocaloric system” built on the same alloys.

Pressure. The newest and strangest. Squeeze a waxy organic solid called a plastic crystal and its floppy molecules stop tumbling, and the entropy change is enormous. Xavier Moya at Cambridge and Josep-Lluís Tamarit in Barcelona measured a “colossal” barocaloric effect in neopentylglycol, a cheap compound sold by the drum, reporting in Nature Communications an entropy change of 510 joules per kilogram per kelvin, on par with the hydrofluorocarbon fluids in your fridge. They patented it in 2017; the US grant, shared between Cambridge and two Catalan universities, issued in 2022, and MIT landed its own barocaloric heat-transfer patent in 2025.

Delete the word “caloric” from every one of those and the descriptions still belong together: a solid material that changes temperature when you apply a field, pumping heat with no compressor and no gas. That is the engineering DNA. The four camps are arguing about which field is cheapest to apply at scale, and the patent record shows the argument tilting away from magnets.

Why an R&D director should care

Cooling is not a niche. Refrigeration and air conditioning account for roughly three-quarters of all hydrofluorocarbon use in the United States, according to the EPA, and those refrigerants are potent greenhouse gases. R-410A, the standard in home AC, traps heat about 2,088 times as effectively as carbon dioxide, pound for pound. Under the AIM Act, the US is phasing HFC supply down by 85 percent by 2036, and the Kigali Amendment behind it is projected to avoid up to half a degree of global warming this century. The regulatory clock is the forcing function. A cooling technology with no refrigerant at all does not have a phasedown problem.

That reframes Carrier’s portfolio. The company sells compressors and the HFC-charged systems built around them. It is also the largest US patent holder in the one class of technology that would make compressors and refrigerants unnecessary. It is hedging across three of the four caloric mechanisms while sitting out the one that just bankrupted its most committed competitor. You do not need to guess at strategy to find that interesting. The grants are public, dated, and readable.

The 2015 wine cooler was aimed at the wrong field. The heat was never the hard part. The magnet was.

Method note. Counts come from roughly 9.3 million US utility patent grants sourced from USPTO bulk data, searched full-text for the four caloric effects (magnetocaloric and magnetic refrigeration, electrocaloric, elastocaloric, barocaloric), with grants issued through mid-June 2026. The 247 figure is distinct granted patents matching any of those terms; assignee totals combine variant spellings and subsidiary filings, so a company that files under several legal names is counted once. Full-text search catches patents that discuss a caloric effect without centering on it, so treat the totals as a floor, not a census. Material and device details are quoted from the patents’ own abstracts and claims. Scientific and regulatory figures are attributed inline to the University of Cambridge, Nature Communications, the University of Maryland and Science, BASF, Astronautics Corporation of America, the trade outlet R744, and the US Environmental Protection Agency.