When the satellites zoomed in, you could see the panels gleaming from space. Pairing images taken miles above the Earth with Chinese customs records, BloombergNEF solar analyst Jenny Chase and her team discovered this year that the rooftops of homes and factories across Pakistan are blanketed with solar panels. Catching their own government by surprise, Pakistanis have been installing a massive amount of solar power.

In the process, Pakistan has gone from an inconsequential solar market to the sixth-largest in the world. The country of 242 million has a power grid with a peak capacity of 46 gigawatts — that’s less than 4 percent of the US power supply for a country with more than two-thirds as many people. In the last three years, Pakistanis have imported more than 25 gigawatts of solar panels from China. This disorganized, bottom-up boom has increased Pakistan’s power supply by 50 percent.

The solar surge is driven by high local electricity costs. At 16.6 cents per kilowatt-hour, Pakistan’s electricity rate for businesses is 37 percent higher than its neighbor India, and more than double the average rate in Asia. Agreements made in the 1990s have kept the state stuck in expensive contracts with independent power producers, and power plants burn lots of liquefied natural gas, which became costlier after Russia invaded Ukraine in 2022. That same year, Pakistan fell into a foreign exchange crisis as the country’s dollar reserves plunged, which made everything more expensive.

All of this opened an opportunity for businesses and better-off Pakistanis to begin importing solar panels from China, which can pay for themselves in as little as two years and free their users from the expensive, unreliable grid. The middle class has started to do the same. The state has come under pressure to raise rates for the conventional grid to satisfy its contracts with power producers — which the increasingly shrinking, poorer customer base struggles even more to afford. Consumers who have made the switch to solar panels, like the owner of a factory that makes soccer balls in Sialkot, told the Financial Times, “Allah has given us this gift to get out of this mess.”

But there’s a bigger story here, beyond one nation’s problems with its power grid. What’s happening in Pakistan is the latest sign that energy authorities are underestimating how much clean power the world demands — and that energy models can suffer from the same biases as their makers. Those failures in number-crunching are not merely abstract. Failing to grasp how much energy is wanted, and the things people in places like Pakistan might be willing to do to get it, leaves the world unprepared to build, fund, and plan for a cleaner future.

Why our energy demand projections are always wrong

History has shown that cheap energy creates its own demand. When steam engines got more efficient in 19th century Britain, coal consumption grew. When oil got cheap and plentiful after World War II, humans didn’t enjoy the savings. They built more cars. More recently, in 2000, when estimating an industrializing China’s electricity demand for 2005, the Energy Information Administration’s projections were off by 25 percent.

Modelers try to project how much energy will be demanded years into the future. But those projects often fail to distinguish between how much energy is needed and how much is wanted.

In part, this is about the particular outperformance of solar, whose growth the International Energy Agency (IEA), an intergovernmental organization that oversees the global energy sector, has drastically underestimated every year since 2006 — as have countries’ own renewable energy targets. The IEA’s Net Zero by 2050 report, a plan for how to eliminate net greenhouse gas emissions by the mid-21st century, was seen as ambitious when it came out in 2021. It called for the world to add 630 gigawatts of solar power annually by 2030. This is actually proving a very easy target: The world is already on track to add nearly 600 gigawatts in 2024 — 334 gigawatts in China, 53 gigawatts in the US, and, stunningly, at least 16 gigawatts in Pakistan.

The pattern is that Western energy forecasters are continually surprised by how much energy people in developing countries will consume. As countries like Pakistan grow wealthier, their people are going to demand the same energy-dependent conveniences that people in wealthy countries already do — and our energy forecasts must reflect that reality or we’re going to keep getting it wrong.

One recent projection illustrates the problem. Working with Pakistan’s energy authorities, the Danish Energy Agency examined in 2023 how Pakistan’s electricity sector might develop to align with the IEA’s net-zero-by-2050 pathway. It expected Pakistani electricity generation to reach about 350,000 gigawatt-hours (a unit of energy use over time) in 2045, doubling its 2022 capacity of 173,000 gigawatt-hours. But that would still leave the hundreds of millions of people living in future Pakistan to get by with much less power than Texas generates today, with a fraction of Pakistan’s population.

But that estimate fails to account for Pakistan’s middle class, which is almost 100 million strong with millions more on the cusp of exiting poverty and joining its ranks. As people get richer, they ask for — and use — more electricity. Today, only 11 percent of Pakistani households have air-conditioning. In six days this June, when temperatures in southern parts of the country hit 120 degrees Fahrenheit, at least 568 people died. In a 3 degrees Celsius scenario, which is what we are on track for by 2100, South Asia’s major cities will face multiple heat waves every year that last an average of 23 days — nearly twice as long as they do today. Even if air conditioners grow increasingly efficient, in a scenario in which power production merely doubled between now and 2045, a large share of the population would lack protection from the deadly temperatures of the 2040s.

Either the rising middle class of Pakistan will continue to endure heat waves like this year’s without using technology to ease their suffering — or they are going to end up using a lot more power than prominent forecasts for energy consumption assume. That would have precedent: global demand for energy to power air-conditioning and other cooling measures grew from 300 terawatt-hours in the 1990s to 800 terawatt-hours in 2023, which is one reason why worldwide carbon emissions still haven’t peaked. In Pakistan, a recent study found that higher temperatures will help drive electricity demand to grow twice as fast (6 to 8 percent annually) as current estimates. The IEA itself has once again revised its forecasts for electricity demand growth upward by 6 percent, adding that demand for air-condition is a key uncertainty.

Models of energy and climate systems are full of assumptions that often don’t filter through to those reading the headline findings: some scholars call it “status quo bias.” When an energy agency in Paris or Copenhagen posits that the world can reach net-zero with millions of Pakistanis having just enough power to spin ceiling fans, they are postulating certain limits on how much power could be installed in Pakistan, and how much Europeans could cut emissions to leave Pakistan space for growth. When seemingly apolitical integrated assessment models, which evaluate the economic results of climate change, tell policymakers which temperature targets are possible or desirable, they are also revealing political assumptions that are often mistaken.

Accepting his Nobel Prize in 2018, William Nordhaus, the godfather of integrated assessment models, asserted that the “cost-benefit optimum [for warming] rises to 3C in 2100.” In other words, the most cost-effective outcome of climate change — in terms of how much we spend to stop it and how much damage it inflicts — is a global average temperature increase of 3 degrees Celsius, twice the warming target set in the 2015 Paris climate agreement (a target the world is well on its way to missing). But whose cost, and whose optimum? Models like Nordhaus’ assume economic development can protect people from the climate change that the same economic activity causes. This line of thinking assumes there is a limit to how much the world’s wealthy countries can slow down (or green) their energy consumption to arrest climate change — and that therefore there is a limit on how much new energy can be allocated for poorer countries with their rapidly developing economies.

The world’s growing middle class isn’t waiting for permission to buy air conditioners. The task now is to make sure that the energy that powers them is clean — and that means having more than enough solar panels for Lahore as well as Copenhagen.