The Chinese bus company couldn’t work it out. Some days when its buses merely crawled along Shanghai’s streets, their power consumption would go through the roof. The reason why was a mystery. So, a team from the US firm Montana Technologies flew out to investigate. They started clamping electricity meters onto parts of the bus and soon realized what Yutong Bus, the Chinese company in question, had missed.

“They didn’t instrument air conditioning,” says Matt Jore, CEO of Montana Technologies, explaining how the buses’ AC, used to counter the often hot and humid weather in Shanghai, had a huge effect on power consumption. “The driver would turn the air conditioning on, it would just spike up.”

Whenever anyone, anywhere, reaches for the button that activates air conditioning, or lowers the desired temperature in their room a degree or two, energy use rises. A lot. In humid conditions, air conditioners have to work especially hard—more than half of the energy they consume can go toward dehumidifying the air rather than cooling it. The buses struggling in China’s muggy weather gave Jore and his colleagues an idea. If they could make dehumidification more efficient somehow, then they could make air conditioning as a whole much more efficient, too. They headed back to the US wondering how to make this happen.

Cooling devours one-tenth of the world’s energy, or 20 percent of all energy used in buildings. The International Energy Agency expects demand for cooling to skyrocket in the next 25 years—with two-thirds of the world’s households expected to have an air conditioner of some kind by 2050. As the climate crisis deepens, judicious use of cooling will only be more important. For one thing, it has the potential to save many lives. But air conditioners, for all their usefulness, are far from sustainable. Cooling tech could, in principle, be far more efficient.

“I have here 50-gallon barrels of this stuff. It comes in a special powder,” says Jore, referring to the moisture-loving material that coats components inside his firm’s novel dehumidifier system, AirJoule. This is the result of years of research and development that followed his team’s trip to China. The coating is a type of highly porous material called a metal-organic framework, and the pores are sized so that they fit around water molecules extremely well. It makes for a powerful desiccant, or drying device.

“Just one kilogram can take up half or more than half—in our case 55 percent—of its own weight in water vapor,” says Jore.

The AirJoule system consists of two chambers, each one containing surfaces coated with this special material. They take turns at dehumidifying a flow of air. One chamber is always drying air that is pushed through the system while the other gradually releases the moisture it previously collected. A little heat from the drying chamber gets applied to the moisture-saturated coating in the other, since that helps to encourage the water to drip away for removal. These two cavities swap roles every 10 minutes or so, says Jore.

This process doesn’t cool the air, but it does make it possible to feed dry air to a more traditional air conditioning device, drastically cutting how much energy that secondary device will use. And Jore claims that AirJoule consumes less than 100 watt-hours per liter of water vapor removed—potentially cutting the energy required for dehumidification by as much as 90 percent compared to a traditional dehumidifier.

Montana Technologies wants to sell the components for its AirJoule system to established HVAC firms rather than attempt to build its own consumer products and compete with those firms directly—it calls the approach AirJoule Inside. The firm is also working on a system for the US military, based on the same technology, that can harvest drinkable water from the air. Handy for troops stationed in the desert, one imagines. However, AirJoule is still at the prototype and testing stages.

“We’re building several of these pilot preproduction units for potential customers and partners,” says Jore. “Think rooftops on big-box retailers.”

Rival firm Blue Frontier has also come up with a desiccant-based dehumidifying system, though it uses a liquid desiccant, a salt solution that is capable of collecting moisture from the air. CEO Daniel Betts says that his firm is installing the tech at multiple undisclosed locations around the US—including office spaces, warehouses, and restaurants. Three are live, with six more to be installed by the end of the year.

As with AirJoule, Blue Frontier’s approach would link to a separate, secondary air-conditioning process to cool the dried air. And Blue Frontier must similarly factor in the need to regenerate its desiccant, though this process can be separated out from dehumidification and run at times when there is less demand on the electricity grid. “We are shifting the load of air conditioning from the peak,” says Betts.

Really big air-conditioning systems work differently to the unit that you might have in your house or apartment. Take centralized chiller plants in hotels, for instance. They circulate chilled liquid to guests’ rooms where it is used to cool the air. Chiller plants that drop the temperature of this liquid are reasonably efficient already. But they still have to draw power from the grid at peak times, say in the late afternoon, when everyone wants to cool down from the heat of the day, notes Yaron Ben Nun, founder and chief technology officer at Nostromo Energy, which focuses on energy storage.

To get around this issue, Nostromo has created a system called IceBrick, which it installed last year at two adjacent hotels in California: the Beverly Hilton and the Waldorf Astoria Beverly Hills. The IceBrick, a rectangular module, sits on the roof of a building. It contains nearly 200 insulated capsules of water that can be frozen when off-peak energy is available. Then, in the middle of a hot day when hotel guests begin to swelter, the chiller plant can use that stored coolth, as it were, to avoid paying top electricity prices. This doesn’t mean a reduction in energy consumption—actually, it goes up slightly—but Ben Nun says the system can reduce annual cooling costs by 30 percent and associated emissions by up to 80 percent, because the IceBrick can wait to draw power at times when lots of renewable electricity is available on the grid (for instance, when wind turbines are busily spinning in the middle of the night).

The system wouldn’t suit a single house, as it only works with these large centralized cooling installations. However, it can be bolted to a variety of chiller plants—it doesn’t really matter what circulating fluid they use, for example, Ben Nun says.

“It was really exciting to see that,” says Nicole Miranda, senior research associate in sustainable cooling at the University of Oxford. “It’s nice to see an off-the-shelf, quite stackable, flexible solution.”

Not every scenario will require a brand-new approach. Traditional air-conditioning technology can benefit from better design, argues Vince Romanin, CEO of Gradient, which makes window-based heat pumps that can provide heating or cooling to a specific room. The units hang either side of window sills, internally and externally, and are connected by pipes. The external part of the unit is larger than is common for traditional air conditioners, which means it can incorporate a bigger, more efficient heat exchanger, says Romanin. But he stresses that trying to make the most efficient air conditioner possible wasn’t a key goal for the company. Rather, it was “getting high-efficiency systems into more types of buildings.”

On top of such gains, smart controls and other features can also help ensure users don’t waste energy, he notes. A typical Gradient window heat pump currently costs $3,800, but Romanin says the firm hopes to reduce this to $1,000 eventually, in order to make the device as accessible as possible.

“I cannot imagine living without air conditioning. I would not be able to work,” says Jaka Tušek, assistant professor in mechanical engineering at the University of Ljubljana in Slovenia. Temperatures in Slovenia have recently hovered around 35 degrees Celsius (95 degrees Fahrenheit). And yet, many non-centralized air conditioners today only function at around 20 percent of their theoretical maximum efficiency, he notes.

An overheating world is going to need more active cooling—of that, there is no doubt. But exactly how and to what extent we deploy it is going to matter. There is no point over-cooling ourselves either. Eva Horn, professor of German literature and cultural theory at the University of Vienna, called out the irony of doing so in her 2016 essay on air conditioning: “We are getting less and less able to tolerate the very same warming temperatures that we are creating, largely through the massive CO₂ output of air conditioning technology itself.”

That makes it imperative to find ways of reducing energy consumption on air-conditioning overall. All of the above technologies are versions of, or adjuncts to, established air-conditioning technology. But there are ideas floating around for how to make an air conditioner of a completely different design. Among the options that excite Tušek is “electrocaloric” cooling, which uses an electric field to induce a subtle change in the movement of atoms within a material, a kind of phase transition, changing the material from one state to another. This causes a temperature change within that material.

Last November, Emmanuel Defay at the University of Luxembourg and colleagues published a paper in which they described an electrocaloric cooling device. It’s made using thin strips of ceramic material stacked on top of one another, with only a thin gap of air between them. This small, ceramic stack sits inside a tube filled with fluid. When an electric field is applied, the phase transition happens within the solid ceramic material, causing it to heat up. The fluid dutifully absorbs heat from the stack and displaces it to one side of the tube. Then, the electric field is turned off and the stack cools down. As that happens, the fluid moves in the opposite direction, ultimately cooling the other end of the tube.

“They are somehow dancing together,” says Defay. Over roughly 100 cycles, one end of the tube gets hotter (reaching around 30 degrees Celsius) while the other end gets significantly cooler (around 15 degrees Celsius). “Little by little, you create this gradient of temperature,” he adds. It is radically different from the cooling technology people rely on today and could be around 20 percent more efficient overall for space cooling one day, Defay estimates. However, the laboratory prototype has only about 4 watts of cooling power, and the ceramics are made using less-than-ideal materials such as lead, which is toxic. Tušek notes that the long-term reliability of the materials used in such a system is yet to be proven.

In the race to develop better active cooling technologies, we might miss the opportunity to adopt more passive cooling measures, says Miranda. Passive cooling—as simple as shutters that keep the sun off windows on summer days—costs nothing to run, she notes: “The risk here is again that we have an easy, off-the-shelf solution, but that makes us a bit lazy about using the passive options that could save us electricity.”