by Bill Gates
The situation with Suntown isn’t merely a hypothetical example. Something similar has been happening in Germany, which through its ambitious Energiewende program set a goal of 60 percent renewables by 2050. The country has spent billions of dollars over the past decade expanding its use of renewables, increasing its solar capacity nearly 650 percent between 2008 and 2010. But Germany produced about 10 times more solar in June 2018 than it did in December 2018. In fact, at times during the summer, Germany’s solar and wind plants generate so much electricity that the country can’t use it all. When that happens, it ends up transmitting some of the excess to neighboring Poland and the Czech Republic, whose leaders have complained that it’s straining their own power grids and causing unpredictable swings in the cost of electricity.
There’s another problem caused by intermittency, and it’s even harder to solve than the daily or seasonal variety. What happens when an extreme event forces a city to survive several days without any renewable energy at all?
Imagine a future where Tokyo gets all its electricity from wind power alone. (Japan does, in fact, have quite a bit of onshore and offshore wind available.) One August, at the peak of cyclone season, a massive storm hits. The winds are so strong that they will rip the city’s wind turbines apart if they aren’t shut down. Tokyo’s leaders decide to switch off the turbines and get by solely on electricity stored up in the best large-scale batteries they can find.
Here’s the question: How many batteries would they need in order to power Tokyo for three days, until the storm passes and they can turn the turbines back on?
The answer is more than 14 million batteries. That’s more storage capacity than the world produces in a decade. Purchase price: $400 billion. Averaged over the lifetime of the batteries, that’s an annual expense of more than $27 billion.*5 And that’s just the capital cost of the batteries; it doesn’t include other expenses like installation and maintenance.
This example is entirely hypothetical. No one seriously thinks Tokyo should get all its electricity from wind or store all of it in today’s batteries. I’m using this illustration to make a crucial point: It’s extremely difficult and expensive to store electricity on a large scale, but that’s one of the things we’ll need to do if we’re going to rely on intermittent sources to provide a significant percentage of clean electricity in the coming years.
And we’re going to need much more clean electricity in the coming years. Most experts agree that as we electrify other carbon-intensive processes like making steel and running cars, the world’s electricity supply will need to double or even triple by 2050. And that doesn’t even account for population growth, or the fact that people will get richer and use more electricity. So the world will need much more than three times the electricity we generate now.
Because solar and wind are intermittent, our capacity to generate electricity will need to grow even more. (Capacity measures how much electricity we’re theoretically capable of producing when the sun is shining its brightest or the wind is blowing its hardest; generation is how much we actually get, after accounting for intermittency, shutting down power plants for maintenance, and other factors. Generation is always smaller than capacity, and in the case of variable sources like solar and wind it can be a lot smaller.)
With all the additional electricity we’ll be using, and assuming that wind and solar play a significant role, completely decarbonizing America’s power grid by 2050 will require adding around 75 gigawatts of capacity every year for the next 30 years.
Is that a lot? Over the past decade, we’ve added an average of 22 gigawatts a year. Now we need to install more than three times that much each year, and keep up the pace for the next three decades.
That will be a bit easier as we make solar panels and wind turbines cheaper and even more efficient—that is, as we invent ways to get even more energy from a given amount of sunlight or wind. (The best solar panels today convert less than a quarter of the sunlight that hits them into electricity, and the theoretical limit for the most common type of commercially available panels is about 33 percent.) As these conversion rates go up, we can get more power from the same amount of land, which will help as we deploy these technologies widely.
But more efficient panels and turbines aren’t enough, because there’s a major difference between the build-out America did in the 20th century and what we need to do in the 21st. Location is going to matter more than ever.
Since the beginning of the electric grid, utilities have placed most power plants close to America’s rapidly growing cities, because it was relatively easy to use railroads and pipelines to ship fossil fuels from wherever they were extracted to the power plants where they’d be burned to make electricity. As a result, America’s power grid relies on railroads and pipelines to move fuels over long distances to power plants, and then on transmission lines to move electricity over short distances to the cities that need it.
That model doesn’t work with solar and wind. You can’t ship sunlight in a railcar to some power plant; it has to be converted to electricity on the spot. But most of America’s sunlight supply is in the Southwest, and most of our wind is in the Great Plains, far from many major urban areas.
In short, intermittency is the main force that pushes the cost up as we get closer to all zero-carbon electricity. It’s why cities that are trying to go green still supplement solar and wind with other ways to generate electricity, such as gas-fired power plants that can be powered up and down as needed to meet demand, and these so-called peakers are not zero-carbon by any stretch of the imagination.
Just to be clear: Variable energy sources like solar and wind can play a substantial role in getting us to zero. In fact, we need them to. We should be deploying renewables quickly wherever it’s economical to do so. It’s amazing how much the costs of solar and wind power have dropped in the past decade: Solar cells, for example, got almost 10 times cheaper between 2010 and 2020, and the price of a full solar system went down by 11 percent in 2019 alone. A lot of the credit for these decreases goes to learning by doing—the simple fact that the more times we make some product, the better we get at it.
We do need to remove the barriers that keep us from making the most of renewable sources. For example, it’s natural to think of America’s electric grid as one single connected network, but in reality it’s nothing of the sort. There isn’t one power grid; there are many, and they’re a patchwork mess that makes it essentially impossible to send electricity beyond the region where it’s made. Arizona can sell spare solar power to its neighbors, but not to a state on the other side of the country.
We could solve this problem by crisscrossing the country with thousands of miles of special long-distance power lines carrying what’s called high-voltage current. This technology already exists; in fact, the United States already has some of these lines installed. (The biggest one runs from Washington State to California.) But the political hurdles to a massive upgrade of our electric grid are considerable.
Just think about how many landowners, utility companies, and local and state governments you’d need to bring together to build power lines that could move solar energy from the Southwest all the way to customers in New England. Merely picking the routes and establishing rights-of-way would be a massive undertaking; people tend to object when you want to run a big power line through the local park.
Construction on the TransWest Express, a transmission project designed to move wind-generated power from Wyoming to California and the Southwest, is scheduled to begin in 2021. The project is supposed to become operational in 2024—some 17 years after planning began.
But if we could pull this off, it would be transformative. I’m funding a project that involves building a computer model of all the power grids covering the United States. Using the model, experts have studied what it would take for all western states to reach California’s goal of 60 percent renewables by 2030, and for all eastern states to reach New York’s goal of 70 percent clean energ
y by that same year. What they found is that there’s simply no way for the states to do it without enhancing the power grid. The model also showed that regional and national approaches to transmission—rather than leaving each state to its own devices—would allow every state to meet the emission-reduction goals with 30 percent fewer renewables than they would need otherwise. In other words, we’ll save money by building renewables in the best locations, building a unified national grid, and shipping zero-emissions electrons wherever they’re needed.*6
In the coming years, as electricity becomes an even bigger part of our overall energy diet, we’ll need models like these for grids around the world. They’ll help us answer questions like: Which mix of clean energy sources will be the most efficient in a given place? Where should transmission lines go? Which regulations stand in the way, and what incentives do we need to create? I hope to see a lot more projects like this one.
Here’s another complication: As our houses rely less on fossil fuels and more on electricity (for example, to power electric cars and stay warm in the winter), we’ll need to upgrade the electrical service to each household—by at least a factor of two, and in many cases even more than that. A lot of streets will need to be dug up and electrical poles climbed to install heavier wires, transformers, and other equipment. So it will be felt in a real way by nearly every community, and the political impact will get down to the local level.
Technology might be able to help overcome some of the political barriers involved with these upgrades. For example, power lines are less of an eyesore if they’re run underground. But today, burying power lines increases the cost by a factor of 5 to 10. (The problem is heat: Power lines get hot when there’s electricity running through them. That’s no problem when they’re aboveground—the heat just dissipates into the air—but underground there’s no place for the heat to go. If the temperature gets too high, the power lines melt.) Some companies are working on next-generation transmission that would eliminate the heat problem and reduce the cost of underground lines significantly.
Deploying today’s renewables and improving transmission couldn’t be more important. If we don’t upgrade our grid significantly and instead make each region do this on its own, the Green Premium might not be 15 to 30 percent; it could be 100 percent or more. Unless we use large amounts of nuclear energy (which I’ll get to in the next section), every path to zero in the United States will require us to install as much wind and solar power as we can build and find room for. It’s hard to say exactly how much of America’s electricity will come from renewables in the end, but what we do know is that between now and 2050 we have to build them much faster—on the order of 5 to 10 times faster—than we’re doing right now.
And remember that most countries aren’t as lucky as the United States when it comes to solar and wind resources. The fact that we can hope to generate a large percentage of our power from renewables is the exception rather than the rule. That’s why, even as we deploy, deploy, deploy solar and wind, the world is going to need some new clean electricity inventions too.
* * *
—
There’s already a lot of great research going on. If there’s one thing I love about my work, it’s the opportunity to meet with, and learn from, top scientists and entrepreneurs. Over the years, through my investments in Breakthrough Energy and in other ways, I’ve heard about some potential breakthroughs that could be the revolution we need to get to zero emissions in electricity. These ideas are in various stages of development; some are relatively mature and well tested, while others are, frankly, nuts. But we can’t be afraid to bet on some crazy ideas. It’s the only way to guarantee at least a few breakthroughs.
Making Carbon-Free Electricity
Nuclear fission. Here’s the one-sentence case for nuclear power: It’s the only carbon-free energy source that can reliably deliver power day and night, through every season, almost anywhere on earth, that has been proven to work on a large scale.
No other clean energy source even comes close to what nuclear already provides today. (Here I mean nuclear fission—the process of getting energy by splitting atoms apart. I’ll get to its counterpart, nuclear fusion, in the next section.) The United States gets around 20 percent of its electricity from nuclear plants; France has the highest share in the world, getting 70 percent of its electricity from nuclear. Remember that by comparison solar and wind together provide about 7 percent worldwide.
And it’s hard to foresee a future where we decarbonize our power grid affordably without using more nuclear power. In 2018, researchers at the Massachusetts Institute of Technology analyzed nearly 1,000 scenarios for getting to zero in the United States; all the cheapest paths involved using a power source that’s clean and always available—that is, one like nuclear power. Without a source like that, getting to zero-carbon electricity would cost a lot more.
Nuclear plants are also number one when it comes to efficiently using materials like cement, steel, and glass. This chart shows you how much material it takes to generate a unit of electricity from various sources:
How much stuff does it take to build and run a power plant? That depends on the type of plant. Nuclear is the most efficient, using much less material per unit of electricity generated than other sources do. (U.S. Department of Energy)
See how small the nuclear stack is? That means you’re getting far more energy for each pound of material that goes into building and running the power plant. It’s a major consideration, given all the greenhouse gases that are emitted when we produce those materials. (See the next chapter for more detail about that.) And these figures don’t take into account the fact that solar and wind farms generally need more land than nuclear plants, and they generate power only 25 to 40 percent of the time, versus 90 percent for nuclear. So the difference is even more dramatic than this chart shows.
It’s no secret that nuclear power has its problems. It’s very expensive to build today. Human error can cause accidents. Uranium, the fuel it uses, can be converted for use in weapons. The waste is dangerous and hard to store.
High-profile accidents at Three Mile Island in the United States, Chernobyl in the former U.S.S.R., and Fukushima in Japan put a spotlight on all these risks. There are real problems that led to those disasters, but instead of getting to work on solving those problems, we just stopped trying to advance the field.
Imagine if everyone had gotten together one day and said, “Hey, cars are killing people. They’re dangerous. Let’s stop driving and give up these automobiles.” That would’ve been ridiculous, of course. We did just the opposite: We used innovation to make cars safer. To keep people from flying through the windshield, we invented seat belts and air bags. To protect passengers during an accident, we created safer materials and better designs. To protect pedestrians in parking lots, we started installing rear-view cameras.
Nuclear power kills far, far fewer people than cars do. For that matter, it kills far fewer people than any fossil fuel.
Nevertheless, we should improve it, just as we did with cars, by analyzing the problems one by one and setting out to solve them with innovation.
Scientists and engineers have proposed various solutions. I’m very optimistic about the approach created by TerraPower, a company I founded in 2008, bringing together some of the best minds in nuclear physics and computer modeling to design a next-generation nuclear reactor.
Because no one was going to let us build experimental reactors in the real world, we set up a lab of supercomputers in Bellevue, Washington, where the team runs digital simulations of different reactor designs. We think we’ve created a model that solves all the key problems using a design called a traveling wave reactor.
Is nuclear power dangerous? Not if you’re counting the number of deaths caused per unit of electricity, as this chart shows. The numbers here cover the entire process of generating energy, from extracting fuels to turning them into electricity, as well as the environmental problems they cause, such as a
ir pollution. (Our World in Data)
TerraPower’s reactor could run on many different types of fuel, including the waste from other nuclear facilities. The reactor would produce far less waste than today’s plants, would be fully automated—eliminating the possibility of human error—and could be built underground, protecting it from attack. Finally, the design would be inherently safe, using some ingenious features to control the nuclear reaction; for example, the radioactive fuel is contained in pins that expand if they get too hot, which slows the nuclear reaction down and prevents overheating. Accidents would literally be prevented by the laws of physics.
We’re still years away from breaking ground on a new plant. So far, TerraPower’s design exists only in our supercomputers; we’re working with the U.S. government on building our first prototype.
Nuclear fusion. There’s another, entirely different approach to nuclear power that’s quite promising but still at least a decade away from supplying electricity to consumers. Instead of getting energy by splitting atoms apart, as fission does, it involves pushing them together, or fusing them.
Fusion relies on the same basic process that powers the sun. You start with a gas—most research focuses on certain types of hydrogen—and get it extraordinarily hot, well over 50 million degrees Celsius, while it’s in an electrically charged state known as plasma. At these temperatures, the particles are moving so fast that they hit each other and fuse together, just as the hydrogen atoms in the sun do. When the hydrogen particles fuse, they turn into helium, and in the process they release a great deal of energy, which can be used to generate electricity. (Scientists have various ways of containing the plasma; the most common methods use either powerful magnets or lasers.)
