Opening
A nickel-titanium alloy, stretched and released inside a small prototype, recently pulled a heat sink from room temperature down below zero Celsius. No refrigerant. No compressor. Just a metal that gets cold when you stop pulling on it, described in a recent Nature paper on elastocaloric cooling.
That device is not going into a window any time soon. But it points at something the cooling industry has been quietly betting against for a decade: that the vapour-compression machine bolted to the side of half the world’s buildings is, in fact, a finished technology.
The story sold to consumers is that the air conditioner just needs a greener refrigerant and a smarter thermostat to ride out the climate century. That framing is wrong, and the engineers trying to replace the compressor entirely are the clearest evidence of it. A small but growing group of physicists and startups argue the vapour-compression cycle has hit a wall. Their pitch: rip out the refrigerant, the compressor, and the copper coils, and cool rooms with solid blocks of metal, plastic, or ceramic that change temperature when you squeeze, stretch, or magnetise them.
The promise is real. So are the doubts.
The problem the industry would rather not name
Cooling is one of the few technologies where the public health case and the climate case point in opposite directions. Air conditioning has prevented significant numbers of heat-related deaths in recent decades, particularly as extreme heat events have become more frequent. That is not a marginal benefit. It is one of the largest public health interventions of the modern era, distributed one window unit at a time.
The same machines are also a problem. Air conditioning accounts for a significant share of global electricity use and greenhouse-gas emissions. The global stock of AC units is projected to roughly triple by 2050 as middle classes expand across South Asia, Southeast Asia, and sub-Saharan Africa.
That trajectory is locked in. The only open question is what those tripled units run on.
Why the compressor has a ceiling
Modern HVAC systems already perform better than most people realise. The coefficient of performance for a typical HVAC unit sits around 3, meaning it shifts three units of heat for every unit of electricity it consumes.
That is good. It is also close to the practical ceiling of what compressing and decompressing a refrigerant can deliver in a unit small enough to bolt to a wall.
The deeper issue is the refrigerant itself. R410A, the working fluid in many residential systems, has a global-warming potential more than 2,000 times that of carbon dioxide. Every leak, every botched recycling job, every landfilled unit is a small atmospheric event. The industry is migrating to lower-GWP fluids, but the chemistry is constrained: anything that boils and condenses at useful temperatures tends to be either flammable, toxic, or a potent greenhouse gas.
The solid-state argument starts here. If the refrigerant is the problem, get rid of the refrigerant.
Four bets on a compressor-free future
There is no single solid-state cooling technology. There are at least four, each chasing the same physical phenomenon from a different angle: a material that gets cold when you stop doing something to it.
Thermoelectrics use semiconductors that pump heat when a current passes through them. They are reliable, silent, and already power the mini-fridges in hotel rooms. They are also inefficient at room scale.
Magnetocaloric systems exploit the fact that certain metals, with gadolinium as the textbook example, heat up when magnetised and cool down when the field is removed. German startup Magnotherm has moved this from the lab into commercial refrigeration, launching a pilot with the REWE Group in German supermarkets earlier this year.
Elastocaloric materials, mostly shape-memory alloys, release heat when stretched or compressed and absorb it when relaxed. Think of a rubber band that warms when you pull it, scaled up and made out of nickel-titanium.
Barocaloric systems use pressure rather than magnetic fields or mechanical strain. A UK company called Barocal, built around Cambridge materials research, is developing solid refrigerants intended to replace gas-based cooling fluids.
Each approach avoids fluorinated refrigerants entirely. Each one is also, today, trying to prove that it can compete with the boring compressor in the window.
The efficiency gap nobody can fully explain
This is where the scientific community starts to disagree with the press releases.
The open question is why solid-state coolers cannot yet consistently match the efficiency, durability, manufacturability, and cost profile of conventional thermodynamic cycles. The theoretical limits suggest they should be able to. The lab benches say the road is harder.
Some of the gap is engineering. Heat exchangers in prototype devices are crude, and parasitic losses eat into the gains. Some of it is materials: the best magnetocaloric and barocaloric compounds can be expensive, rare, or difficult to cycle reliably. And some of it is genuinely unresolved physics, because entropy changes that look promising on paper can degrade once a material is cycled millions of times in a real device.
Two to three years is the kind of timeline researchers and founders often talk about for more convincing room-scale demonstrations of elastocaloric and barocaloric systems. Not mass-market products. Demonstrations.
The industry calendar runs in decades. That window is tight.
Why the edges matter
Climate technology coverage tends to fixate on full replacement. Either the new technology kills the old one, or it failed. Cooling does not work that way.
Even a small share of the global cooling market could matter if solid-state systems solve problems that compressors handle badly. Efficiency is not the only metric. Durability, refrigerant-free operation, quiet performance, and the ability to serve niches that compressors handle poorly all count.
That math gets more interesting when you consider where solid-state wins first. Vibration-sensitive environments. Medical cold chains. Vehicle cabins, where compressor weight and noise are penalties. Spaces where occupants cannot tolerate the temperature swings or sudden failures of conventional units. A pilot programme is already testing solid-state heating and cooling for people with disabilities, on the theory that quiet, refrigerant-free systems may serve users for whom a conventional AC’s quirks are not merely annoying but disabling.
None of those markets need to dethrone the window unit. They just need to exist.
The demand curve is not waiting
The urgency is not abstract. Hotter summers are already changing how people think about cooling, resilience, and grid demand. Multiply that by projected AC growth and the load on grids built for a cooler century becomes the story.
India installed millions of room air conditioners last year. Indonesia, Vietnam, and Nigeria are tracking similar curves at lower absolute numbers but steeper growth rates. Each of those units, today, almost certainly contains a fluorinated refrigerant and a compressor designed in its essentials before the moon landing.
The locked-in emissions from cooling demand over the next twenty-five years dwarf any plausible solid-state rollout in the same period. That is not a reason to dismiss the new technologies. It is a reason to be honest about what they are for.
Who actually wants a colder room
There is a behavioural layer to this that engineers tend to skip over. Cooling demand is not purely a function of outdoor temperature. It is shaped by income, by building stock, by cultural norms about indoor climate, and by personal history. Thermostat preferences often track childhood economics as much as physiology.
That matters for solid-state cooling because the technology’s strongest pitch is not raw output. It is granularity. A magnetocaloric or thermoelectric module can be small, silent, and local. A cooled chair, a cooled bed, a cooled vest, in ways a compressor cannot. If the future of cooling is partly about cooling people rather than rooms, the efficiency comparison with a central HVAC system changes.
You do not need to outperform a 3-tonne unit if you are cooling a square metre of skin.
The investor problem
Hardware is hard. Climate hardware is harder. Cooling hardware sold into a commodity market dominated by Daikin, Carrier, Mitsubishi, and a handful of Chinese giants is harder still.
The startups working on solid-state cooling face a financing reality that does not reward the patience the science requires. Barocal raised on Cambridge research and a pitch that resonates with European climate funds. Magnotherm took the supermarket refrigeration route: a real customer, a defined use case, a workable cost structure today rather than a theoretical one in 2032. That sequencing is probably the playbook. Solve refrigeration first, where margins tolerate premium technology, then move toward residential cooling once volumes drop unit costs.
It is also the playbook that takes a decade. The demand curve does not slow down to wait for it.
What the scientists are actually saying
The scepticism in the academic literature is not that solid-state cooling will not work. It is that the public conversation treats it as imminent when it is not, and treats it as a silver bullet when it is one piece of a portfolio.
The honest version of the pitch goes like this. Solid-state systems will likely reach commercial viability first in specific niches: commercial refrigeration, specialised cooling, medical or mobility-related climate control, and high-end residential markets willing to pay for silent or refrigerant-free operation. They are not going to displace the window AC in a Mumbai apartment by 2050, and pretending otherwise does the technology no favours.
The scientific case is still advancing. A recent Nature paper on elastocaloric cooling described a device using low-transition-temperature nickel-titanium alloys that achieved sub-zero Celsius cooling from a room-temperature heat sink. That is not a mass-market air conditioner. It is evidence that the underlying physics keeps moving toward practical machines.
That is a long way from the marketing language. It is also a long way from doing nothing.
The uncomfortable middle
Climate technology coverage tends to swing between two failure modes. One celebrates every lab announcement as the end of the fossil era. The other dismisses anything pre-commercial as vapourware. Solid-state cooling sits squarely in the uncomfortable middle, which is also where most of the important climate technologies actually live.
The compressor will not disappear in this decade. The refrigerant problem will not solve itself. The demand curve will not flatten because the physics is interesting.
What can happen is a slow substitution at the edges, starting in supermarkets and specialised applications, then moving toward the rooms where people actually live as the cost curves cross. Whether that substitution becomes material by 2040 or stalls as a specialist niche depends on engineering breakthroughs that have not happened yet, financing patience that does not yet exist, and policy frameworks (refrigerant pricing, building codes, procurement standards) that mostly remain to be written.
It is worth asking, though, whether the delay is really about engineering risk. The vapour-compression cycle is a century-old supply chain. Compressor plants, refrigerant chemistries, distributor networks, certification regimes, the trained technicians who service the existing fleet. Every one of those represents a fixed cost that an incumbent has already amortised and a competitor has to absorb from scratch. The conservatism the industry calls prudence looks, from a different angle, like protection of a depreciated asset base.
The startups are not the ones who need to justify themselves. The window AC in the room already has.
