For seven years, the U.S. government has been searching for an energy miracle via the Advanced Research Projects Agency for Energy, or ARPA–E. And yet, over the same time span, the biggest advance has been the quick deployment of various renewable energy technologies everyone knows about, accompanied by a rapid drop in cost.

But Bill Gates, co-founder of the Bill & Melinda Gates Foundation and former founder of Microsoft, says that’s not enough to really revolutionize energy supply and significantly reduce carbon emissions.

So the world's richest man is putting his money where his mouth is, corralling cash from his fellow billionaires as part of the Breakthrough Energy Coalition, formed in 2015. Their intent is to put money into far-out energy ideas like turning air and sunshine into energy-dense fuels, using kites to harvest energy from high-altitude winds or devising a nuclear reactor that will not melt down.

Their funding is not entirely altruistic. After all, if this money can be patient enough, replacing and expanding the world's energy infrastructure could mean trillion-dollar payouts. The Breakthrough Energy Coalition, paired with an international effort known as Mission Innovation, is meant to be that patient money, at least $20-billion worth.

Scientific American sat down with Gates in New York City to discuss this vision for the future of energy.

[An edited transcript of the interview follows.]

What do you mean by an energy miracle?
People may have a stronger definition of that term than I do. I mean, I consider the computer-on-a-chip a miracle. The Internet's a miracle. The cell phone is a miracle. Big scientific advances—unanticipated, broad-impact—they're coming and coming at a faster rate than ever. If we can get [energy] that's measurably less expensive than hydrocarbons, completely clean and providing the same reliability, that I will consider a miracle.

I predict that in the next 15 years we have a high probability of achieving it. When you have about a dozen [technology] paths—all of which at least to me appear to have better than 20 percent chance independently—if you get the R&D up, if you do things on the demand side and include great things [governments have] done like production tax credit, investment tax credit, Renewable Portfolio Standard, many tens of billions of dollars just in the U.S. alone, then you're very much tilting the odds to have a very positive surprise.

We've heard about your efforts in nuclear fission, but you also talk about solar fuels and what we call at Scientific American reverse combustion—turning CO2 back into fuel, the way plants do. Why does that stand out? The energetics don't seem so great.
Nature shows that it can be done. The reason I like that one in particular—well partly I like Professor Nate Lewis [at the California Institute of Technology] who works on that stuff—and he showed me that he's within a factor of 100, which is a long, long way, but they can make tiny, tiny amounts [of energy]. It doesn't guarantee that they can get there but if you want an example of something that should get double or triple the amount of R&D it gets today, I'd say that's it. Not just the one lab but several labs taking hopefully slightly different and competitive approaches.

The beauty of it is that a liquid hydrocarbon, say gasoline, is really a magic energy carrier. You put it in a steel tank. It sits there for arbitrary periods of time. It's got an [energy] density that's still an order of magnitude better than the best battery that we can make. It's incredible.

Natural gas isn't as easy as the liquid stuff to store but we can store it pretty darn well, either in liquid or gaseous form. Natural gas [power plants for times of peak electricity demand], if you don't have a carbon [emissions] constraint, they're really kind of miraculous.

Those things are really, really cheap. And only if compressed air or flow batteries were super, super successful, can they even be in that order of magnitude, of that regime, of super, super-high amounts of energy. Most battery technologies have a hard time playing a grid role, which is different than a role in consumer transport, where another factor of two or three probably would get you into the mainstream product profile.

Didn't the energy miracle already happen with the fracking revolution? And if we just get the CO2 captured and stored—CCS—underground, or anywhere besides the atmosphere, then we're done?
Well CCS is hard to do. The one thing you can never say about CCS is that it will make energy cheaper than it is today. Remember, my initial entry into the energy problem isn't so much climate, although I think that's super important, but I start out with let's make energy cheaper. Because when I say I want more fertilizer for Africa, I want more light at night, I want vaccines to stay cold for Africa—that's all about the price of energy.

People say, "Oh, there's a water shortage." Another way to say that is that energy's too expensive, because converting this thing we have called ocean into clean water anywhere, that's all a matter of paying for the desalination and paying for the pumping. So in a certain sense there isn't a water shortage. It's just that the price of water, given a certain price of energy, means that doing agriculture in the Middle East, in northern China is uneconomic.

So, yes, fracking is an incredible thing. It's some mix of government R&D, entrepreneurial attitude—and the digital revolution, of course, is playing a nice role in all these things. That ability to take seismic data or magnetometer data, all this different data, and create an understanding of the various geological layers and therefore say, "Okay, I think this deposit probably extends out to there", and control all the things to be very precise. The IT part of it is never to be underestimated. People kind of take that for granted. But the oil industry didn't have the magic of big data, and now they do and the sensors that let you impute all that stuff get dramatically better as well.

And CCS, I absolutely think we should pursue it. To guarantee it stays in the ground—probably only a governmental entity can take on that multicentury liability, that they're going to be around and ensure the quality of that—so you have to have the right regulation. You have to find the right geological structures and then you have to understand what the capital and energetic costs are for CCS. Somebody should have done a large-scale CCS project, both with a high-percentage recovery against coal and with a high percentage recovery against natural gas.

Because CCS is expensive, can we expect all nations to try to use it?
It's not clear what you would do with India. India is a big enough energy consumer over the next 40 years that you can't give them a free pass. Sub-Saharan Africa, which in the worst case over the next 40 years will be 4 percent of global greenhouse gas emissions, you can give them a pass and say, "Hey any way you guys can get energy, if it's natural gas, coal, gasoline, whatever you want.... Hey, you are such a small percentage and your per-person emissions are essentially zero,"—putting aside land use and livestock, which is a complicated area. Even then they wouldn't get to our greenhouse levels until the end of the century if they went full speed ahead with their current coal-based plans.

Then there's some sort of CO2 capture from the air coming out of rich countries, maybe, which we should not assume that such a thing becomes economic. There are people, including some I fund, who are working on that. But the cost per ton recovered is in the $100-plus regime, and you need to get down in the probably $20 to $30 regime before you can start to say, "Okay, let's multiply that by 36 billion [metric tons of CO2 emitted annually] and start thinking about whether or not we should do this or not."

India to me is kind of paradigmatic, because they have this imperative in terms of women not breathing smoke and people being able to have fertilizer. The imperatives of giving them what we take for granted is that they will live probably a five-times-more energy-intensive life, which would still leave them less than a third of the U.S., even somewhat below European. They should use more energy.

Now you can have a footnote and say, “Okay, poor countries are still allowed to emit.” Mostly, however, when we talk about energy we mean: household, factory, office, transport.

Cars, coal and cattle.
Yeah, factories, too. I don't know if you saw this book, Sustainable Materials with Both Eyes Open?

It just goes through steel, wood, plastic, paper, aluminum, and how much are we likely to need in the future, and what sorts of process improvements would reduce the energy intensity. They end up having to say, "Okay, you've got to start reusing stuff." It was a great book because it teaches you when a refrigerator is “obsolete”; it's actually the plastic parts that have gotten messy and ugly and need to be replaced. The metal parts are not fatigued in any way.

It could be reused, is basically what you're saying.
The key point the book makes is that reuse is a stronger benefit energy-wise than recycle, because in recycle, in the case of the metals, you melt it again, and so you're actually using almost as much energy—aluminum, say half as much—but almost as much energy. In the digital world being able to label everything—who made it, what alloys does it have, when was it made, and even someday putting sensors in that will understand the degree of fatigue. It's big. Both eyes open means that you can't just take the metal manufacturing process and optimize that. You have to look at the whole design and reuse it as a societal materials thing in order to get to these drastic reductions.

What gives you hope about more efficient and less carbon-intensive energy production, besides reverse combustion? What are some of the other pathways?
Well I have a ton of money, not as a percentage, in a company called TerraPower, which is making a fourth-generation nuclear reactor technology. It may specifically be a solution; it's a great, great company because of computer simulation. We built a piece of software which, with today's supercomputers shouldn't be too surprising, where we sit and simulate the [reactor] material's fatigue. We can throw a tidal wave at it. We can throw a Richter scale 10 earthquake at it, a volcano at it. We've tried out more nuclear power designs in this one company than in all of the history of mankind put together.

The software is amazing, and we do all these tests to make sure it's right. We have materials out of the last U.S. fast reactor, and they have some neutron bombardment, but then we put those into another reactor and we do additional radiation, and then we look at these things and see what's happened to the crystal structure. We think we really understand how to make steel that doesn't [degrade].

The knock on nuclear has always been Adm. Hyman Rickover's comment that the best reactors are always paper reactors, and we might update that to say the even better reactors are the computer reactors. Everything changes when you get them in actual working experience. Do you anticipate that you'll be able to overcome those challenges—and quickly? It seems too slow.
We wouldn't want to count on TerraPower. We need 12 paths, five companies per path. We need at least 60 TerraPowers. I'll stick up for TerraPower because the assignment was inherent safety based on physicsno humans, no buttons, no training—superior economics; more than a factor-of-10 waste reduction; strong antiproliferation—and that you don't run out of uranium. On paper they have met their assignment.

The idea, through a partnership with China or some other country, is: Can you get the pilot plant built, if everything went well, by 2024? And then have six years of operating experience where by 2030 you would say to the world, "Hey, build as many of these as you want. All new nuclear starts should be this and nuclear starts as a percentage of new or replacement energy should be very high." That's the possibility, but we've got to get the pilot plant built. We've got to get it approved. It has to work super well. The time frames can't slip too dramatically. It's a serious entrant and, from my potentially biased point of view, in the nuclear fission category I don't know many other entrants that you look at and say, "Okay, if you go from paper to real, then this is a meaningful contribution to cheap energy, and to global warming as an incredible problem."

I can see that working for India, but it doesn't seem like that would work for many African countries, because they don't have the grid to go with that TerraPower reactor?
There's always been this thing called ship-based nuclear. [Basically, a nuclear power plant on a ship.] Now my credibility will drop, but the ocean is pretty magical [for dealing with the challenges of nuclear power, like safety or cooling].

But what about a country like Chad, nowhere near the ocean?
You're right. We're going to have to fly that baby into Lake Chad; landlocked countries are always tough.

It's not just landlocked countries though, it's poor countries, even rich countries.
One of the interesting simulations we're doing nowadays is, Can you build a high-voltage direct current grid without making the cost of your energy superhigh? If you assume you have that magic grid, and if you diversify your energy sources—like California wind, Washington wind, various things would win out in that model—you can get up to 80 percent [of electricity from renewable resources], which I was always skeptical about and doesn't seem intuitively right to me, but the models are pretty clear. So then you have this 20 percent that is unsolved—assuming the price of wind and solar come way down, which if you can extend the current learning curve on solarthat gets to be pretty good.

One thing that's cool is that about 25 percent of the time you have more energy than you need. You need a new type of customer that's called the "I can use intermittent energy" customer. You need a profile where their capital cost is very low, and the cost of energy in their equation is very, very high, so that they outcompete plants that get 24-hour energy.

Maybe somebody will create some standby heat solution or something, but it would be interesting because you would have two prices. You'd have the price of 24-hour energy and then you would have this significantly lower specialized price, which with the magic grid you could deliver to essentially any parts of the geography. Fertilizer, steel, aluminum, making hydrogen, whatever thing you like, making hydrocarbons, electrofuels, you get the price low enough—and who knows what kind of breakthroughs will be there. Then you can do electrofuels, [using microbes or other means to turn electricity into liquid fuels].

Then we export those to Africa to solve their problem.
Exactly. The more fungibility of energy you can get, it makes these problems a lot easier. It's hard to beat liquid hydrocarbons for flying airplanes. But it's a small percentage of all energy usage. It's small enough that some biofuels or electrofuel-type approach or some expensive per-ton free-air-capture thing can offset that piece.

The land use and livestock one is tricky, and there's less going into that then you might like.

It sounds like you’re saying the energy miracle we need is on the deployment side. We have all these options. TerraPower might be one. There are the solar panels that are going up across the U.S. We need more batteries out there. Is deployment the miracle we're looking for?
No, no. Just do the equation of how much we put into the demand side for clean energy—and look at Germany, Japan, the U.S. and do the accounting for the Renewable Portfolio Standards and things like that. It's tricky. Then look at how much incrementally we put into the supply side. Did Germany increase their R&D budget when they bought some of the solar that's at one of the stranger places in the world to put solar? What did they end up with? In my view there's been an imbalance towards funding the demand side. I love all of it, and of course you need to do both.

There are deployment challenges but high-altitude wind is not a deployment challenge. If there's some fusion path, if there's some solar fuels path…. Nate Lewis's problem is not a deployment problem. His problem is a basic research, materials understanding.

Put aside energy—we are on the verge of understanding alloys on a rational basis, understanding catalysts on a rational basis. When you talk about an energy R&D budget, it's not just, "Okay, this is only useful for an energy-type thing." This is really basic materials science. Why do we have to keep repairing roads and bridges? Come on, let's create something that takes away that ongoing cost or at least reduces it.