The History of Air Conditioning

Let's start our story with the hot blast of air in front of an outdoor unit. An air conditioner is a machine that loads heat from inside a room onto a substance called a refrigerant and ferries it outdoors. Inside the room, the refrigerant turns into a gas and absorbs heat; outside in the unit, it turns back into a liquid and spits that heat out. It doesn't eliminate heat—it simply moves it somewhere else. That is why the cooler a room gets, the hotter it feels in front of the outdoor unit.

On top of that, the electricity used to move that heat eventually turns into heat as well and gets dumped outside with it. The heat dumped outside is always greater than the heat removed from the room. This simple fact will follow us all the way to the final chapter of our story, because what happens when an entire city turns on its air conditioning hinges right on this point.

Let's take a quick look at the name, too. The original term for this technology wasn't "cooling," but "air conditioning." It means treating not just the temperature, but humidity, cleanliness, and airflow all together. Cooling is just one part of that.

So saying someone "invented air conditioning" fits this story much better than saying they "invented the air cooler." It wasn't that cooling machines came first and humidity control was tacked on later; rather, the technology to condition air came first, and cooling later became the product that was sold.

The order of events is also the exact opposite of what common sense suggests. The problem this technology first tried to solve wasn't temperature—it was humidity. The amount of moisture air can hold depends on its temperature. So if you cool air down enough by bringing it into contact with a cold surface, the moisture it can no longer hold condenses into droplets and falls away. Temperature was lowered simply to extract moisture. The goal and the means were the reverse of what we assume today.

Here is the road ahead. A machine born for printing presses moves into theaters, offices, and homes. Along the way, the fluid circulating inside becomes the subject of international treaties twice. And in the end, this machine redraws the limits of a city's summer peak electricity demand and the heat a human body can endure.

Our first scene opens with ice. Before machines came along, coolness was something you mined or hauled in.

Even before machines existed, people managed the heat. They just did it with architecture instead of machinery. In the desert cities of Iran, homes featured windcatcher towers known as badgirs. These tall towers caught the breeze and funneled it down over underground pools or water channels; as the water evaporated, cooled air flowed into the living quarters.

In the Dowlat Abad Garden of Yazd stands a 33.8m-tall badgir. Believed to have been built around 1750, it is the tallest surviving historic badgir today. Combined with courtyards, thick earthen walls, and wooden lattice windows fitted with porous water jars, techniques for cooling indoor spaces without using a single drop of electricity had been built up over at least several centuries.

Another path was ice. In 1806, Frederic Tudor of Boston loaded ice cut from New England ponds onto a ship and sent it to the Caribbean. That first voyage ended in a loss of about $4,500. Yet he refused to back down, eventually turning ice into a viable, marketable commodity.

In May 1833, the brig Tuscany sailed from Boston carrying 180 tons of ice, arriving in Calcutta, India, that September. By the time it arrived, 100 tons of ice remained. More than half had survived a four-month voyage. Aided by innovations like sawdust insulation and ice cutters, the ice trade along this route continued for nearly 40 years.

That same year, 1833, Dr. John Gorrie took up his post in the Florida cotton port of Apalachicola. When yellow fever struck in 1841, he hung basins of ice from the ceiling so the cool air would sink over his patients. What is important here is his motive: he was not trying to beat the heat, but to cure disease. He believed bad air transmitted illness, and that cooling and drying the air could stop the disease in its tracks.

Gorrie succeeded in making ice mechanically in 1844, and around 1850 he was routinely producing brick-sized blocks of ice. In 1851, he received the first U.S. patent for an artificial ice-making machine. However, facing ridicule from New England ice merchants and a lack of funding, his business failed, and he died in 1855 at age 52.

Just how expensive coolness was in this era is illustrated by the White House in 1881. As the shot President Garfield lay dying, Navy engineers packed layers of cotton cloth into a coffin-sized cast-iron box, filled the top with ice, salt, and water, and used a blower to force outside air through it. According to widely cited accounts, the indoor temperature dropped to 80°F, about 27°C, consuming about 227 tons of ice over 58 days. The president died that September.

In July 1901, 25-year-old Willis Carrier, having completed his master's degree in mechanical engineering at Cornell, joined the heating and ventilation equipment maker Buffalo Forge. The following year, the Sackett-Wilhelms lithographic printing plant in Brooklyn brought a problem to the company.

Here was the problem. Color printing requires pressing one color on top of another, one layer at a time. But when humidity fluctuates, paper absorbs and releases moisture, expanding and contracting. When that happens, previously printed colors and later printed colors fall out of alignment. Paper curling and wrinkling at the edges even jammed the feeding mechanisms. The problem wasn't the heat; it was the moisture.

On July 17, 1902, Carrier penned his initials onto the mechanical drawings designed to solve this problem. Today, this date is celebrated as the birthday of modern air conditioning. To be clear, the air conditioner wasn't switched on that day. It was simply the day he signed the blueprints.

The method was just as we saw in the first chapter. Run cold water through coils and pass air over them. Once the air cools down to the surface temperature of the coils, it drops the moisture it had been holding right onto them. According to Carrier's company records, the design target was not temperature, but maintaining a relative humidity of 55% year-round. Its cooling capacity was equivalent to using about 100,000 pounds—a bit over 45 tons—of ice per day. If you convert the ice melted over 58 days at the White House 20 years earlier into a daily rate, this machine pulled off roughly ten times that amount every single day.

According to Carrier's company records, late that autumn on a foggy train platform in Pittsburgh, he had a crucial flash of insight. His thought was that if you thoroughly saturate air with water and then control its temperature, you can produce air with whatever exact amount of moisture you want. Spraying water to dry the air—an idea that sounds contradictory at first glance. However, because this anecdote leans on the company's official history and Carrier's own reminiscences, it is more accurate to view it as part of a corporate founding myth.

In January 1906, Carrier received U.S. Patent 808,897 for an "Apparatus for Treating Air." It was a cooling patent granted in the dead of winter.

So this machine's first customer was not a human being. Cooling for people would have to wait roughly 20 more years.

In 1906, the year the patent was issued, Stuart Cramer, a textile engineer from North Carolina, gave a speech at the American Cotton Manufacturers Association convention in Asheville. It was there that he used the term "air conditioning." Carrier was not the person who coined this phrase.

Why a textile engineer? Cotton yarn snaps if the air gets too dry. That was why Cramer had been developing equipment to add moisture to mill air since 1895. Pre-hydrating yarn back then was known as "water conditioning," and by analogy, he called treating the air "air conditioning." This term was born not to mean "cooling," but "conditioning the air to make spinning yarn easier." He went on to design around 150 Southern cotton mills and held more than 60 patents related to humidity control.

While the name was taking shape, an entire scientific field was emerging alongside it. In December 1911, Carrier presented his "Rational Psychrometric Formulae" at the annual meeting of the American Society of Mechanical Engineers. It was the landmark paper that made it possible to calculate the exact relationships among humidity, temperature, and dew point—the temperature at which water droplets begin to condense. The industry calls this the Magna Carta of Psychrometrics. It was the precise moment guesswork turned into engineering.

It even ventured into private homes. In 1914, mechanical air cooling was installed in the mansion of Minneapolis millionaire Charles Gates. It is recognized as the first case of residential air conditioning in the United States. Yet Gates died without ever getting to live in that house. Even into the 1930s, a residential setup cost at least $1,500, keeping it far out of reach for ordinary people.

In June 1915, Carrier set up his own company alongside six fellow engineers with $32,600 in capital. Then, in May 1922, the company unveiled the centrifugal chiller. It was a massive machine capable of cooling an entire large building, and it made it possible to use safer refrigerants instead of hazardous ammonia.

Without this machine, the next scene in history could never have unfolded: the moment people began paying money just to buy cool air.

In 1924, Carrier equipment was installed in the J. L. Hudson department store in Detroit. It was the moment air conditioning shifted from factory quality control to a business tool for attracting customers.

Then, over the Memorial Day weekend in 1925, the Rivoli Theatre in New York's Times Square opened its doors with a 133-ton Carrier cooling system—equivalent to melting 133 tons of ice per day. As the story goes, last-minute adjustments delayed the startup, so it was still warm inside when the audience entered. Adolph Zukor of Paramount was sitting in the audience that day. Over the next 5 years, air conditioning was installed in some 300 theaters across the United States.

Here is a common misconception: the claim that the term "summer blockbuster" originated in movie theaters around this time is unconfirmed. "Blockbuster" originally referred to a bomb in the early 1940s capable of destroying an entire city block, and it was first used in a movie advertisement in May 1943. Summer became the peak moviegoing season only after two box office hits in 1975 and 1977. While theater air conditioning did create the habit of going to the movies in the summer, the phrase itself has a different origin.

In January 1928, the Milam Building opened in San Antonio, Texas. It was the world's first high-rise office building designed and constructed with air conditioning from the start. Carrier installed the equipment under the name "Manufactured Weather," and blocks of ice were placed in the basement to assist the chillers. That refrigeration unit ran for 60 years until October 1989. There is another reason this building matters: air-conditioned offices reportedly commanded 10–15% higher rents. This was the turning point where air conditioning evolved from a utility into a source of revenue.

In April of that same year, lawmakers sitting in the windowless chamber of the U.S. House of Representatives also got to enjoy air conditioning before the year was out. It was ordered in May and brought into operation within the year. What is interesting is why the representatives embraced it: during the first three months, about 227 kg of dust was trapped in the filters, and seeing this with their own eyes led them to view air conditioning as health equipment. In June 1938, a central air conditioning plant was installed for the Capitol and House office buildings. Even with air conditioning, however, the tradition of summer recess never disappeared.

What about the tropics? In May 1930, mechanical air conditioning was introduced at the Capitol Theatre in Singapore—just 5 years after theater air conditioning debuted in New York. Carrier established its Asian regional headquarters in Singapore in the early 1930s, and in July 1938, the Alhambra Theatre reopened as Malaya's first modern air-conditioned cinema.

The reason theaters came first becomes obvious when you look at the price tag. In 1937, installing home air conditioning in Singapore cost about $600, while a skilled worker's daily wage back then was about $1.40. Even in the tropics, cool air was something you rented for a few hours by buying a ticket.

Before making its way into homes, it made a stop in the car. In 1939, Packard offered automotive cooling as an option for the following year's models. It was the world's first factory-installed car air conditioner. Yet after about 2,000 units were sold, it vanished after 1941. It took up the entire trunk, and without a dashboard switch, turning it off reportedly required stopping the engine and removing the drive belt.

It entered homes only after the war ended. In 1947, Henry Galson introduced a lower-cost, compact window air conditioner, selling over 43,000 units that year. It was a time when factories and parts were in surplus.

Here lies a common misconception: that mass-produced suburban developments like Levittown, built starting in 1947, came standard with air conditioning. That is not true. By 1951, 17,447 homes had gone up on Long Island alone, and the standard features of those early models were radiant heating and a television, not cooling. Air conditioning in American suburbs spread by being added to windows one by one after the homes had already been built.

Commercial buildings were a different story. Completed in 1952 in New York, Lever House was wrapped in sealed glass walls that could not be opened. It was a design that simply could not exist without air conditioning, and cities around the world soon copied it. Air conditioning had ceased to be mere building equipment and had become the building's core prerequisite. When the power goes out, older buildings open their windows, but in buildings like this, there are no windows to open.

The inflection point came in 1953. Annual sales of room air conditioners in the United States surged from 74,000 units in 1948 to 1,045,000 units in 1953. That was a 14-fold increase in just five years. It marked the watershed moment when air conditioning shifted from a luxury to an everyday household essential.

And at this point, the most famous claim emerges: that air conditioning drove the population explosion of the American South and the Sun Belt. There is no doubt that the South grew. Its share of the nation's population rose from 24% in 1950 to 30% by the mid-2000s. Between 1940 and 2020, Arizona's population grew by about 14.3 times and Florida's by about 11.4 times. Over the same period, the entire United States grew by about 2.5 times. Historian Raymond Arsenault's 1984 paper is the starting point of this narrative.

In 2007, however, Edward Glaeser and Kristina Tobio quantified the drivers of Southern growth across productivity, amenities, and housing supply, arriving at a different conclusion. They found that productivity growth was the main driver before 1980, while policies that made housing easy to build took over afterward. One piece of evidence lies in prices. Over the same period, the South's average home prices rose from 83% to 91% relative to the national average, while incomes rose from 76% to 94%. Since housing prices grew more slowly than incomes, it suggests people were not willing to pay a premium simply for the amenity of sunshine. There are also several competing explanations: the aerospace and defense industries, Cold War federal investments, the end of the Jim Crow system, right-to-work laws, and even mosquito control.

Does this mean air conditioning did nothing? Not at all. But what we can state with much firmer evidence concerns lives, not population. A 2016 study estimated that the impact of days with average temperatures exceeding 80 degrees Fahrenheit on American mortality fell by about 75% in the second half of the 20th century compared to the first half. The authors concluded that this decline was virtually entirely explained by the adoption of residential air conditioning.

Still, we cannot simply take this figure and apply it elsewhere. It measures how much the link between heat and death weakened, not where people chose to move. Not turning correlation into causation is the key takeaway here.

Here, we need to turn back the clock for a moment. To understand what comes next, we have to follow the story of the refrigerant circulating inside the machine from the very beginning.

In the 1920s, refrigerators and cooling units used ammonia, sulfur dioxide, and methyl chloride. At least the first two had a noticeable smell if they leaked. Methyl chloride, however, had almost no odor and became explosive when mixed with air. After a series of accidents were reported in Chicago in August 1929, people even put their refrigerators out in the yard before going to sleep.

Another event from the same year is still frequently misquoted today. That is the disaster at the Cleveland Clinic on May 15, 1929, in which 123 people died. This was not a refrigerant accident. It was a fire where cellulose nitrate X-ray film stored in the basement ignited from the heat of an incandescent light bulb, releasing toxic gas. Because genuine methyl chloride accidents occurred around the same time, the two stories became conflated.

In April 1930, at the American Chemical Society meeting in Atlanta, Thomas Midgley Jr. presented the answer. According to a widely told story, he inhaled a lungful of the new gas and blew it out over a candle to extinguish the flame. It was a demonstration that it was neither toxic nor flammable. This substance would later be known as Freon-12. Midgley was also the very same person who, a few years earlier, had proposed adding lead to gasoline. R-12 went on the market in 1931, followed by R-11 in 1932 and R-22 in 1936. By the standards of the time, Freon was an obvious improvement in safety.

The bill arrived 44 years later. In 1974, Mario Molina and F. Sherwood Rowland published a paper in Nature. It showed that CFCs leaking from spray cans, refrigerators, and air conditioners drifted up into the stratosphere and destroyed ozone. Here lies the paradox: The very property chosen for safety—the fact that it reacted with nothing—allowed CFCs to reach the stratosphere without breaking down.

In 1985, Joseph Farman and his colleagues reported that ozone over Antarctica had depleted by up to 70%, and that same year 20 countries signed the Vienna Convention. The Montreal Protocol was signed in September 1987 and entered into force in January 1989. It became the first treaty in UN history to be ratified by every member state.

It is important here not to confuse two distinct issues. Ozone depletion and global warming are separate problems with different causes and consequences. The Montreal Protocol targeted the former. It phased out CFCs, which had high ozone-depleting potential, in developed nations by 1996. R-22, an HCFC with lower depleting potential, was kept a bit longer as a transitional refrigerant before being pushed out on a schedule of 2020 for developed nations and 2030 for developing nations.

When new products in the United States and Europe were barred from using R-22 in 2010, R-410A stepped in to fill the void. R-410A is an HFC, so it does not destroy ozone. The ozone problem was solved right there. However, its 100-year global warming potential is 2,088 according to the IPCC Fourth Assessment Report and 2,285 according to the Sixth Assessment Report—with values varying even for the same substance depending on the edition of the report.

So in October 2016 in Kigali, the very same treaty revisited the same ground. This time, the regulations targeted them not as ozone depleters, but as greenhouse gases. The amendment took effect in January 2019, aiming to cut production and consumption by more than 85% relative to the 2011–2013 baseline. The completion dates vary by country group. There is an estimate that this could avoid 0.4–0.5°C of warming by the end of the 21st century, though the exact range differs across the literature. In December 2020, the United States codified this same direction into law through the AIM Act.

Today, the market is shifting toward R-32, R-454B, propane, and carbon dioxide. Yet these new refrigerants are not necessarily safer. R-32 and R-454B are classified as mildly flammable, and propane is classified as flammable. The fact that the propane charge limit for residential split-system air conditioners was raised from 150g to up to 988g does not mean it has become safer; it means the allowable limit was expanded under strict safety-design conditions. When you recall that the reason Freon was chosen 100 years ago was precisely because it was 'non-flammable and non-toxic,' today's transition is the exact same kind of trade-off.

Meanwhile, the machine itself evolved. In 1981, Toshiba commercialized a residential air conditioner that used an inverter to adjust the compressor's rotation speed. A commercial version had arrived the previous year. Instead of constantly turning on and off, it ran only as much as needed. In 1982, Daikin introduced a system where a single outdoor unit could independently regulate the refrigerant flow to multiple indoor units. Afterward, this Japanese inverter and heat pump design became the standard across Asia and Europe. By the way, reversing the direction of refrigerant flow turns the exact same machine into a heater. That is why a large share of wall-mounted air conditioners in Korea already double as heaters.

Korea's story began earlier, in 1968. Goldstar, known today as LG Electronics, partnered technologically with General Electric to release the GA-111, a window-unit room air conditioner roughly the size of a television set. While records agree on the release year, the launch month and model name notation vary slightly depending on the source. Samsung Electronics began manufacturing air conditioners in the mid-1970s, and Goldstar followed with a wall-mounted model in 1979 and Korea's first floor-standing model in 1983.

Yet at the time, air conditioners were found in government offices, dabang coffee houses, and the homes of the wealthy, not ordinary households. According to Gallup Korea surveys, the household cooling appliance ownership rate was 6% in 1993, 14% in 1996, 24% in 1998, and 36% in 2001. It then surged to 74% in 2012, 80% in 2016, and 87% in 2018. The year 2018 is the latest confirmed data point. These figures reflect whether a household owned at least one unit, not the number of units per household. Unverified tables circulating online claiming "Korea's penetration rate is 98%, No. 1 in the world" have no traceable source.

The look of the appliance changed as well. During the 1990s, the market shifted from window units to floor-standing units. In 2000, LG Electronics launched Whisen, and in July 2002, Samsung Electronics introduced Hauzen. Throughout the 2000s, the two companies solidified a structure where their combined market share exceeded 90%. However, this figure fluctuates depending on whether it is measured by revenue or unit volume at a given time. Either way, the air conditioner transformed from a machine temporarily mounted for the summer into a piece of furniture standing in the corner of the living room year-round.

The sheer weight of that shift surfaced on September 15, 2011. Right after the Chuseok holiday in mid-September, temperatures nationwide topped 30 degrees Celsius. Demand forecasts, built on the assumption of autumn weather, were off by 3,281,000 kW, and power plant maintenance had already begun under the assumption that summer had ended. At 2:20 PM, reserve power dropped to 838,000 kW. The threshold for grid stability was 4,000,000 kW. For 4 hours and 45 minutes, from 3:11 PM to 7:56 PM, rolling blackouts hit parts of the country without warning. The Korea Power Exchange rotated power cuts across regions in 1,000,000 kW blocks. The Minister of Knowledge Economy stepped down, and compensation for damages totaled about 7 billion KRW.

What matters most is the cause. The incident was ultimately deemed a management failure—not a shortage of generation capacity, but a misalignment of demand forecasting, maintenance scheduling, and supply tallying. The fact that air conditioners were still running in September was the trigger, but what collapsed was the system for forecasting demand, scheduling maintenance, and tallying supply capacity.

Today, South Korea's peak electricity demand occurs on summer afternoons, not winter evenings. On August 20, 2024, peak power demand hit an all-time record of 97.1 GW, and as of September 2026, that record still stands. Records are not rewritten every single year. Meanwhile, on August 25, 2025, a larger figure of 104.3 GW was reported, but that was a real-time estimate based on total demand including behind-the-meter solar power outside the market, making it a completely different metric from standard peak power demand. Comparing the two figures in the same breath would be inaccurate.

In May 2018, the International Energy Agency published 'The Future of Cooling.' It reported that cooling was the fastest-growing source of energy demand in the building sector. Under its baseline scenario projections, the global stock of air conditioners in buildings would grow from 1.6 billion units to 5.6 billion by 2050, and electricity for cooling would exceed 6,000 TWh. In India, cooling's share of peak electricity load was projected to rise from 10% to 45%. All of these are projections.

Here are the confirmed figures. In 2022, the electricity the world used for cooling was about 2,100 TWh, accounting for about 7% of global electricity. That is more than double the level in 2000. When it comes to the share of greenhouse gas emissions, institutions differ. The United Nations Environment Programme counts refrigerants and electricity together and puts it at about 7%, while Our World in Data, drawing on IEA data, estimates it at about 3.2%. Depending on what gets counted under cooling, how indirect emissions from electricity are allocated, and whether refrigerant leaks are included, the difference widens by more than double. It is a figure where you cannot simply pick a single number and say, 'Cooling accounts for X percent of global greenhouse gases.'

Yet this electricity is not used evenly across the globe. Based on a 2023 IEA commentary, household ownership rates exceed 90% in the United States and Japan, while standing at 15% in Southeast Asia and 5% in India and Africa. Among the 2.8 billion people living in the world's hottest regions, only one in ten has access to cooling. The World Health Organization estimates an annual average of about 489,000 heat-related deaths from 2000 to 2019, though this is an estimate of excess mortality rather than a count written on death certificates.

There are also many misunderstandings about human limits. A 2010 paper suggested 35℃ as the theoretical upper limit of wet-bulb temperature that humans can endure, a figure frequently cited, but this assumed a person who is completely healthy and resting. In a 2022 experiment, even young, healthy adults reached their threshold at temperatures 5 to 10℃ lower than that. During the 2003 European and 2010 Russian heatwaves, the wet-bulb temperature never exceeded 28℃.

And that heat from the first chapter comes right back. An air conditioner dumps the heat it extracts indoors plus the heat generated by the electricity it consumes out into the street. In a simulation of central Paris, street temperatures rose by 0.5℃, and in another simulation of Phoenix, nighttime ground-level temperatures rose by more than 1℃. Because these are separate models with different cities, climates, and equipment configurations, you cannot simply average the two numbers, and both are model results rather than real-world measurements. Yet the direction is the same.

Finally, here is what to keep an eye on moving forward. First is the refrigerant transition. As the shift toward inexpensive, flammable refrigerants accelerates, we have to watch whether charge limits, installation regulations, and maintenance technician training keep pace. Second is efficiency. The Global Cooling Pledge, signed by more than 71 countries at COP28 in December 2023, called for cutting cooling-related emissions by 68% by 2050 and raising the global average efficiency of newly sold air conditioners by 50%. It is a voluntary pledge with no legal binding force.

Third is solutions beyond the machine. The Global Cooling Prize, announced in April 2021, sought residential cooling units with one-fifth or less of the climate impact of standard air conditioners; two teams were named joint winners, but the technology is still at the prototype stage. Radiative cooling materials, which send heat out toward the sky even under direct sunlight, report performance levels around 100 to 150 W/m², but their efficiency plummets under humidity and cloud cover. And what the badgir (windcatcher) once did—architecture that delays heat through eaves, thick walls, and cross-ventilation—remains just as valid, yet it demands generous land area and low-rise buildings.

Fourth is peak load. Rather than how much total electricity was consumed over the course of a year, what dictates the power grid is how much power is drawn all at once during a single hour on the hottest afternoon. Power plants and transmission grids have to be built to handle that single hour.

In 1902, the client was paper. Today, what this machine pushes up is a city's electricity peak, and what it divides is who can endure the heat. The distance between what it was made for and what it has become—that is this story.