The Lifecycle Energy of Ed's New Dishwasher

Podcast Transcript

Ed Coyne: Hello and welcome to Sprott Radio. I'm your host, Ed Coyne, Senior Managing Partner at Sprott. I'm pleased today to welcome a new guest to our show, Mark Mills. Mark is the founder and executive director of the National Center for Energy Analytics. Mark, thank you for joining us today on Sprott Radio.

Mark Mills: I'm delighted we found each other after my email failure when you sent me a note nine months ago. Thanks for inviting me.

Ed Coyne: Mark, we're going to talk about energy, but we're also going to talk about tech. Maybe we can solve that problem for you today.

Mark Mills: It's on me, right?

Ed Coyne: [laughs]

Mark Mills: I need a much better AI chatbot to manage my email and inform me of what to do. Thank God that's coming.

Ed Coyne: It is coming, for sure. Mark, in preparing for this conversation today, I see you have an incredible background. Between physics and engineering, your academic accomplishments are impressive. You've also served, as I was reading, at the White House in the science office under Ronald Reagan, which I find incredibly interesting and impressive. You've had multiple books and covered a wide range of topics in tech and energy. You've spoken at a who's who of events over the years. Let's go back in time and maybe talk about how this all started and then work us up to what you're working on today.

Mark Mills: I began my career in semiconductor manufacturing and moved into fiber optics and missile guidance work. I worked in the early years of the smart weapons systems and the latter years of the Cold War. Then, I got involved in the nuclear energy industry in Canada. I'm Canadian. I went to Queen's University and got my physics degree there. I was always fascinated by the energetics of things. What physicist isn't? I had an opportunity to work for Eldorado Nuclear, which is now Chemical. I learned a lot about nukes, and I’ve always been a nuclear bull, so to speak.

I moved to the U.S. just in time, as a 20-something, for the Three Mile Island to happen, and I was at the site for the week of the accident. Then, I spent the next seven or eight years of my life defending the virtues of nuclear energy. I failed because the U.S. and Canada, to a great degree, not as much, let its nuclear industry lag and shrink. Roughly speaking, the intersection of my interest in computers from an early age is semiconductors and manufacturing, energy and nuclear energy, and then all things energetics, how engines work, and microprocessors work. How rockets work means they're all energetic functions that matter.

It's always informed of how I look, analyze, and forecast. I have always been something of a forecaster. My most recent book, The Cloud Revolution, is a forecasting book. The subtitles are Roaring 2020s, and the book has a lot about AI. The book came out about a year before ChatGPT was a word, got announced. The book has a lot about AI because AI is not new. When I wrote the book, it was obvious that we were at the inflection, and I wrote why that matters.

I did, I believe, the first study ever on the energy appetite of the internet back in 1999. Then I updated that in 2013, looking at the energy appetite of the cloud, which I titled to annoy people, but also, as a matter of semantic fact, The Cloud Begins With Coal. I titled it that way, other than to annoy people because it's a fact. Suppose you look at the evolution of cloud infrastructure construction, which started in 1999 and ended in 2013. In that case, the cloud was beginning to become a real part of the infrastructure of modern economies.

The primary source of net new electricity from 2000 to 2013 was coal. It's about 60% of all net new supply. Ipso facto if you build a net new infrastructure that consumes electricity, and you build new things that consume electricity, it is consuming the marginal supply, so to speak. I'm going to be updating that study again this year. Maybe I'll title it The Cloud Begins with Natural Gas for reasons we can discuss. That's a long explanation in the sense that I did forecasting. Also, I did forecasting relevant investments. I wrote an investment newsletter with a colleague about four or five years back. I filed two venture funds. Of course, in finance and venture, what you're trying to do is, like everybody else, guess the future. You're making a forecast about a company and the market the company's in. I’ve invested in everything from advanced semiconductors and new classes of semiconductor materials to LEDs and fiber optics, lithium batteries, and software for the oil and gas industry. That's our most recent activities.

Before AI became a hot word, we put money to work in a small fund I helped co-found. We put capital to work in companies that developed AI-centric software to improve the oil and gas industry's productivity. We're picking a niche that everybody was ignoring. The real applications for AI, which are far less fun, are materially relevant ones in improving productivity, which, of course, is the engine of growth. Productivity comes from technology, and productivity growth gives us wealth; wealth gives us more demand for energy and everything else.

It's not just that; it makes people healthier and happier. By saying my book is about the coming productivity boom, we raise people out of poverty. I could have titled it the AI Productivity Boom instead of The Cloud Revolution. That has a lot of implications. How are we going to use AI? How are we going to fuel it? What's it going to do? Who's going to be disrupted? How much energy will it use? These are all questions that are now being actively debated and explored.

Ed Coyne: To that point, tech and energy, my initial question was, do you see them becoming codependents? Just hearing you talk right now, I think they've been that way since day one.

Mark Mills: Of course.

Ed Coyne: They're evolving now. Talk about that evolution a bit. As we get more tech demand, we need more energy, and where that energy will come from. Can we expand on that a bit?

Mark Mills: Analogies are helpful. Analogies are never perfect, but they're illustrative. Think in dollar and energy terms because people don't understand, and I don't mean this as an insult to your listeners. As a physicist, I'll tell you that no one can visualize a BTU, a bale of oil equivalent, or a joule. These units don't have salience. You could visualize a gallon of gasoline; you know what a dollar's worth.

If you build a billion dollars worth of cars, it's a lot of cars. A billion dollars worth of cars will consume about $200 million of energy over their 10-year use. They're codependent. This was obvious in the early days; the auto age is still obvious. Most of the energy is oil, and very little is electricity globally still. Even in Norway, where we have lots of EVs, it is still mostly oil. A billion dollars of cars consumes over a 10-year operating life of $200 million of energy. A billion dollars of data centers, pre-AI, over a 10-year life consumes $600 million of energy.

Ed Coyne: Wow.

Mark Mills: If you add AI to the mix, the billion dollars of data centers consume about $2,000 million of electricity over the 10-year life. They're codependent, but the codependency is locked into the physics of the universe we live in. Everything uses energy. It takes energy to build things, to move things, and to run things. Those are the three domains of everything: every service, every product. Not some, but all of them. The codependency comes from who is responsible for the energy. If you're a car maker, fundamentally, you don't care how the car is fueled. The value of the car is just making a profit. Consumers are looking for fuel for the vehicles they want to use.

If you're a data center, you're selling a cloud service to people, whether mapping, inference, or fake video. It doesn't matter. You're directly responsible, not the user of your product; the product is the software in the cloud. You're responsible for the energy. Ironically, the codependency is far higher in data than in cars. Data centers are more like factories.

For example, if you build a chip fab to give you the same metrics, a billion dollars of chip factories consumes about $300 million of electricity and gas over a 10-year life. They consume more energy than cars or EVs per billion dollars, but the codependency is more like data centers. If you're running the factory, you care about your energy's cost and reliability.

The data center world is measured in three ways. People always wear their dollars if you're an investor, but the data center builders measure it in square feet and megawatts. The primary metric of merit is megawatts, which is ironic. The megawatt equivalent is the horsepower of the engine. But you don't measure the number of cars sold and how many megawatts or horsepower of cars are sold. You could do that because we know what the engines are like, but you don't.

You don't measure factories in terms of their ability in megawatts. They do care because that's when they hook up to the grid. Data center guys have always thought about megawatts. We now have data centers being built by the gigawatts a year. The numbers are crazy. More data center capacity will be added in gigawatts in the next three years than in the last decade.

Ed Coyne: Wow.

Mark Mills: Where's the power going to come from? We know. Wherever you can get that much new energy fast. The speed is going to matter because you're building this stuff quickly. You can't build enough windmills. You can't build enough solar panels. They don't supply the energy when you need it, so you have to develop even more of them and then build batteries. You can't build enough batteries.

You can't build nuclear plants quickly enough. You could build coal plants quickly, and that's what's happening in a lot of the world with a lot of coal plants being built. In North America, it's going to be gas. It'll follow the Pareto rule, the 80/20 rule. Probably 80% of the net demand will be fueled by gas and 20% by other stuff, depending on where the data center is. Some of it might be coal; who knows? A lot of it will be wind and solar. Wind is available, reliable, and backed up with many gas turbines ready to turn on.

Ed Coyne: Right. Let's talk about that because there are so many energy sources now. I guess in the early days of this alternative energy movement, or energy transition, and you talk a lot about that too, about people being delusional about this energy transition, there is a reality we do need it all. Can we talk about that a little bit? What does this energy transition look like? Is it real? How should people think about it?

Mark Mills: I think words and definitions matter. There is no energy transition underway. There just isn't. There is no energy transition whatsoever in the data. By that, I mean pick any timeframe you want, but let's do the last 20 years, where Europe and North America have collectively spent between $5 and $10 trillion on this idea of an energy transition, which is mainly replacing hydrocarbons with wind and solar batteries. Over the last 20 years, the absolute hydrocarbon consumption has increased.

If I did an oil equivalent, hydrocarbon use globally has gone up by the equivalent of six Saudi Arabias' worth of oil production in the last 20 years, even as we spent $10 trillion to avoid using oil, gas, and coal. Coal use, natural gas use, windmills, solar turbines, and EVs are at record levels. To your point, we need it all, except that implies they're co-equal. There's been one energy transition in all of human history, but otherwise, there has never been an energy transition. The world uses more wood today for fuel than it did 100 years ago.

Ed Coyne: Wow.

Mark Mills: Burning wood globally still supplies more than twice as much energy as all the wind and solar in the world combined. We haven't given up burning wood, the oldest non-muscle, other than humans, as enslaved people and animals. Other than the use of that form of energy input, wood is the oldest, and we still use wood. We still burn coal. If you graph this, what you see are additions, not transitions.

This is an opinion because forecasting is always an opinion. In the useful, foreseeable future, there will not be anything vaguely resembling a transition. We're not going away from hydrocarbons. We're reducing the rate of growth in their demand by efficiency, of course. By virtue of new stuff like wind and solar, and now, again, I expect more nuclear energy because of the newfound affection for splitting the atom.

When I was in Canada, I got involved in these nuclear debates in my impetuous youth. The slogan was split wood, not atoms.

Ed Coyne: [laughs]

Mark Mills: Some of us might remember that. I did some calculations at that time on how much wood you have to split in Canada to replace the nuclear plants; Pickering and then Bruce were being constructed at that time, and the others. Of course, it's a crazy number. It's a lot of wood. We still burn wood, not just for aesthetic reasons globally, but because people don't have anything else.

It is a disservice to the word, and it's a dangerous myth to think that there's a "transition" underway or that we can make one happen at any price. It's not technologically or economically remotely feasible, despite all the PowerPoint hand-waving, embarrassingly myopic, simplistic, and naive claims. Batteries will get cheaper when you peel the onion layers away piece by piece on each claim. Yes, they will, but they're on the asymptote.

Windmills will get cheaper. No, actually not. They're getting more expensive because they're so materials dependent, and materials costs are rising because we're demanding record amounts of materials. It takes a long time to build a mine. You don't have to be an economics major to know that makes prices go up. Setting aside volatility, all the key input minerals are more expensive now than 20 years ago, which is a reversal of one century of history.

By silly energy policies, we've managed to reverse a one-century trend of a very slow, continual, about 1% per year real decline in metals' costs averaged out over a century, even as we increase demand incredibly over that century. We're exceeding supply by pushing demand so hard in the last two decades to build wind, solar, and batteries. We're making the metals more expensive, which makes everything more expensive because metals are in everything. It's not like they're in windmills and solar panels and batteries. Every building is full of copper.

Ed Coyne: When you talk about wind and solar, we talk about there being no real energy transition. There are additions but no transitions. I think that crystallizes it. That brings me to the word green energy. Does that even exist, then?

Mark Mills: Of course not.

Ed Coyne: It sounds like it's a myth.

Mark Mills: It is. Of course, it is, and so is the word renewable energy. Let me say quickly that there was an energy transition. The world stopped using whale oil for illumination. They did that because of a Canadian, by the way. Gesner was a physician and a chemist. In 1843, he invented the synthesis of kerosene from coal. That was so much cheaper than harvesting whales for oil that kerosene rapidly replaced whale oil about 50 years before the first productive oil wells were drilled in Spindletop, Texas. A Canadian chemist synthesizing kerosene from coal saved the whales. There has been an energy transition. That's it.

Ed Coyne: Back to green energy. We’re hearing it’s a myth.

Mark Mills: It's a myth.

First, there's no such thing as renewable energy. Energy exists in the universe that we occupy. Whether it's the sunlight, the wind moving on the earth, oil below the surface, or coal, all this is free. The idea is that renewable is better because it's free; it's renewable. Everything's free. The energy is all free. We didn't invent it. We have to build machines from materials to get access to energy in nature. The machines all wear out. You have to build machines to harvest nature's energy. Since they wear out, there's no such thing as renewable energy, by definition.

Every source of energy we deliver to society requires refurbishing and rebuilding machines, land, and money. The distinctions are about money, land use, and material use, which gets us to green. Green energy, windmills and solar plants are supposed to be green. They don't use oil and coal directly, but they use them indirectly to build the machines, to make the steel, and to make the aluminum.

That's an important distinction because carbon dioxide emissions from upstream are much more significant than people realize and very difficult to analyze. After all, it's a complex labyrinth supply chain dominated by China, whose energy systems to produce these things are dominated by coal. The average-sized house in California or Canada with a solar array on its roof uses polysilicon produced in China; 98% is made in China on coal-fired grids.

To make the point about green, one roof, an average-sized house in Canada with solar panels uses 30 tons of Chinese coal to manufacture the polysilicon used to make those panels. It doesn't sound very green to me. More importantly, all the other metals and minerals—as you know—the real green burden lies in the copper, the neodymium, the presidium, dysprosium, manganese, the cobalt, and the lithium.

These metals and minerals require big mines, big machines, and lots of chemicals to extract and refine, and it's the polar opposite of green. It couldn't be further from being "green." If you follow the food chain for a gas turbine upstream to the pipes, drilling, and fracking, yes, there's a lot of sand and chemicals there, too. Regarding scale, roughly 1,000% more rocks and materials are moved per unit of energy delivered by going green than by staying in hydrocarbons—1,000% more.

Ed Coyne: Wow.

Mark Mills: That's not green. I'm not apocalyptic. It's not like we don't have the minerals, and I'm worried about mining. I like mining. I work for a mining company. We both know that mining can be done cleanly and efficiently here but not always elsewhere.

Ed Coyne: Of course, you can consume in a cleaner, more sustainable way, but to your point, you still have to do a lot of things to produce the energy people consume. You have already touched on it, but do you have any further thoughts on energy production versus energy consumption and what our listeners should think about that?

Mark Mills: There’s been a lot of research on this. Especially ever since the oil embargo of '73 and '74, there's been a lot more preoccupation and scholarship on the energetics of systems. What you're talking about, of course, is embodied energy, the energy needed to make something, make a car, make a phone, which is different than the energy to operate it. Then there's energy to dispose of it at the end. Whether trying to recycle or bury it, it takes energy.

The so-called life cycle energy is a rich area of scholarship. It's very complex because you have to decide where you're drawing your boundaries, as you might imagine. When you look at life cycles, one thing I can say that annoys people is that it's also true that everything is touched by oil. Not some things. Oil is the largest energy commodity in the world. It's 10 times more energy delivered worldwide than wind and solar combined. It's the largest traded commodity on the planet. Full stop. Nothing comes close. All the rest combined trade at the level of dollars that oil trades at. Every single product and service involves oil, not some percentage, because we have to transport things.

Back to embodied energy. Think about it this way: when you buy a normal car, by that I mean an internal combustion engine car, the ratios are similar for an electric vehicle, but I'll do it for an internal combustion energy car, roughly 15% of the energy that the vehicle uses over its life is in manufacturing the car. Before you get the vehicle, and you've driven it and used a single gallon of gasoline or diesel fuel, you've used some energy, which has caused emissions and environmental impacts of all kinds upstream.

The ratio flips for a lot of other projects, especially technology products. For a lot of tech things, 85% of the energy used is involved in manufacturing them. Like your smartphone, most of the energy your phone uses in your hand is in its manufacturing. Then, most of the energy your phone uses if you use it other than as a phone; if you do smart services on it, like mapping or shopping, then you're using energy that's not extant in your phone. It's in the networks to transport the instructions and in the data centers to do the actions. If you look at it that way, what you find is fascinating.

I calculated 10 years ago, and the number is now bigger, but 10 years ago, the average smartphone used or the average person who uses it used as much electricity as a refrigerator in your kitchen. Today, if you're using any more sophisticated apps and you're streaming video or watching baseball or a soccer game, you're roughly pushing into the two refrigerators worth of electricity you use for your phone each year versus the one refrigerator you have.

Of course, the average house has one or two refrigerators, but if it has three people living in it, the average home has at least three phones. Of course, the phones themselves are multiplying in the car because the car has the equivalent of four or five phones' worth of microprocessors, which are also increasingly using energy upstream to manufacture them.

When you build a billion-dollar chip factory, you will consume about $300 million of energy over 10 years. That's to make the microprocessors. That's a lot of energy. It takes more energy to manufacture microprocessors per billion dollars than to run cars per billion dollars of cars. Embodied energy is an extremely important concept. Our society is moving increasingly towards high-embodied energy things. The more complex the thing is, the more embodied energy it has.

Ed Coyne: I just had a new dishwasher delivered the other day. I was surprised by the first thing when I got on my phone; I had a question pop up: do you want to connect your dishwasher to the internet? I'm like, "Why? Why would I do this?" We're creating so many hookup points now for electricity, energy, and demand on all these things. I'm blown away by it. I don't know if it's all needed, but that's our direction.

Mark Mills: You put your finger on the key issue here: demand. I'll put it differently. We invent energy demands. Human beings invent energy demands. The only natural energy demand is you have to eat to live and have some shelter. That share of your energy demand that each of us has in an impoverished country is 90% food and fuel to stay alive. In our society, it's 10% or 15%. It's everything else.

Healthcare takes energy. The pharmaceutical manufacturing industry is a bigger energy consumer than the chip industry overall right now. Making pharmaceuticals and inventing them takes energy. Why would we connect things? Interesting question. The dishwasher can now do remote forensics and predictive maintenance, telling you when to replace a filter. We all ignore our dishwasher filters.

Ed Coyne: We didn't even know it had one.

Mark Mills: Yes, but they stop working, and then you say, "Why isn't my dishwasher cleaning dishes?" If the dishwasher can remind you in an app that you volunteer to use, but it's much more than that. The essence of what's different about today and yesterday's demands is that human beings will continue to invent new ways to do things that are interesting, fun, or that make our lives better, safer, and more beautiful. That always consumes energy.

Tourism is one of the biggest energy-consuming industries in the world. It's a big industry because people are wealthier and enjoy doing that. It's about a $4 trillion global industry. Think about that. What is the primary vector for tourism globally? Flying airplanes. You don't have to be a genius to know it consumes energy. If you look at your devices, to your point of connecting your refrigerators and dishwashers, or all manner of things, your car, obviously, increasingly are vital signs or health signs, we will volunteer if it's useful to collect our personal health data to get predictive forensics about our life because we care about that.

If you think about the magnitude of data that evolves, it's staggering. Yes, we want cybersecurity for it, but that's true for everything. You always want security. I will put it this way. It sounds hyperbolic, but for the first time in history, we've invented something for which there's infinite demand. There's no infinite demand for cars because, roughly, population saturates car use. You can't drive three cars at once. You might want to own three for fun: a weekend car, a sports car, and a family car, but you only drive one at a time. You do the equivalent in data computing, owning 10 cars and driving them all simultaneously. That's going to go on for the foreseeable future.

Ed Coyne: Let's talk about the highway of energy or the infrastructure of energy. Do we even have the infrastructure to carry this base load? Forget about the global; let's talk about the U.S. briefly. Where are we from an infrastructure standpoint?

Mark Mills: We're in trouble because the interregnum of slow electric demand growth over the last two decades has created the impression that it's permanent. The reason it's an interregnum is that we are harvesting a whole set of one-time efficiency gains going from incandescent to LEDs. The SEER rating, that is, the efficiency rating of your HVAC system in your house and the building you're in, has improved stunningly. They're 300% more efficient now than they were 20 years ago. Electric motors have a smart drive for dynamic loads and are 30% to 50% more efficient. The average refrigerator uses a quarter of the electricity of 120 years ago. All that stock changed over.

Roughly speaking, half of all the things that were the primary uses of electricity got more than twice as efficient in the last 20 years. North American population has grown 20%. You would say, "Why didn't electric demand go up about 20%? It was flat." It's because the efficiency of half of the stuff got twice as good. The better question to ask is, why didn't electric demand go down?

The answer is that we invented new ways to consume electricity. All those savings were one-time deals. We were hitting physics limits in air conditioning and lighting. There are incremental changes, but no big ones are left, at least in the foreseeable future. That's why everybody's now shocked to see we need lots more capacity as growth happens, both economically—more houses, more buildings, more comforts—and also from data centers and factories.

That is problem one. Then the problem two is that all over the U.S. and in Europe, they're retiring or banning the construction of dispatchable power plants, which are principally called gas-fired power plants that can be fired up when you need them quickly, as opposed to episodic power plants, which are wind and solar, which provide power when nature finds it convenient. This trend is going to destabilize grids. Grids are currently staying stable in Europe and here because we're spending more on them. Electric costs are soaring now all over the U.S. and all over Europe in particular, as the cost to maintain reliability is being realized.

Almost every utility is now sending out red flags about stopping the retirement of coal plants, which is now in the news constantly. Germany has refired coal plants because of this. Of course, China is building them, as are Indonesia and India. You'd have to ask, though, if we have a net new increase in demand far greater than planned, which is already the case, how do you supply it?

My forecast is that it will be supplied because it's just too valuable. There's too much economic value in building these data centers so that they will get electricity. They'll get electricity by paying a premium because they need reliable electricity. Most of that will come from natural gas. Some of it will come from refurbished nuclear plants. A lot of it will come from upgrading nuclear plants because, in both Canada and the United States, the up rates can buy you a lot of gigawatts. It's quite surprising.

Ed Coyne: When you say up rates, what do you mean by that?

Mark Mills: If you have a nuclear plant that's, say, 600 megawatts, it's producing at 90% capacity factors. With modern controls, materials, and design systems, you can take that plant and turn the 600 megawatts into 660 megawatts without building a new power plant. That's 63 megawatts, so to speak. It's not too expensive. If you aggregate that across the North American fleet, it's a pretty big number—a lot of gigawatts of up rates. A gas turbine is the same. You can rebuild and improve. You can do the same.

By the way, racing car engines do it all the time. You buy a car that has a 200-horsepower engine, and you can operate the engine by changing the compression ratios. There are all sorts of tricks you could do without buying a new engine. It's true in nuclear plants. We'll probably see coal plants be built to do this. I don't know about it here in North America, but it's already happening all over the world, with coal plants being built to power data centers.

Ed Coyne: Let's throw a wrench in this whole thing for a second and talk about the hot topic right now: tariffs. How do you see tariffs playing a role in energy speed, cost, reliability, and all that? What role are they going to play in your mind?

Mark Mills: If I were working inside a fund again and trying to figure this out, at the macro level, everyone knows the obvious: tariffs raise costs, but it's not as obvious as it sounds. Let's set aside the politics of tariffs for a minute because everybody has tariffs. Pretending President Trump was the first guy to play the tariff game is silly.

He's a little more pugilistic about it than other recent presidents, but it's hardly a new tool. Anybody who's studied China knows that if you're trying to sell a U.S. car in China, the last I checked was a 200% tariff. Good luck buying a U.S. car in China. Countries do that. China's certainly done that. Europe does that. We have tariffs, by the way, on pickup trucks. The U.S. put a very high tariff on foreign pickup trucks long ago, and it's still in place.

Your question is, will it affect where we are regarding energy supply? Suppose something gets more expensive, like labor or material. In that case, it incentivizes you as the user of that labor or material to check out an automation tool, whether a virtual automation tool, an AI, or a robotic automation tool. It'll accelerate automation, which is net good for productivity and net good for the economy. The midterm effect is, I think, very positive. The short-term effect will cause some price inflation in a lot of markets.

On a specific basis, we'll probably see things like solar panels and wind turbines get more expensive more quickly than, say, natural gas. Steel tariffs are important because most steel is produced in China, and Canada is a big steel producer. Steel coming into the U.S. goes into wind turbines and solar farms because of the structures. It also goes into gas costs because steel pipe goes into the ground for the fracking. If 100% of the steel tariff was passed, gas prices and natural gas might be raised by a few percentage points.

For the producer, 3% is what's gone from the bottom line. If you're running a 20 percentage point business, that's a big marginal change. For the consumer, 3% is not life-changing. On the other hand, if it's 3% on the oil price, nobody likes that because everybody wants gas to be cheaper. It's hard to know how it plays out in a specific industry, except for the obvious ones like batteries. Tariffs on battery materials from China will make batteries more expensive because most materials come from China.

The EV battery dominates the cost of an electric car. If the EV battery became 30% cheaper, it would still make the car 40% to 50% more expensive than its competition. It doesn't dramatically change the economics of EVs, but it does change it. From the viewpoint of what the U.S. and Canada can do to supply their electricity, I think it has almost no effect because the other factors go into the cost to make the power overwhelm the material costs from tariffs.

Ed Coyne: What about tax credits? There seemed to be a theme: "Hey, we're going to get rid of the tax credits on wind and solar, and you'll see no more wind or solar plants be built, period, if that happens." What's your thought on that?

Mark Mills: First, full disclosure, they should be eliminated. They should be canceled.

Ed Coyne: Right. Everything should stand on its own, in my mind.

Mark Mills: Yes. It doesn't mean there shouldn't be any credits in the tax system. Governments can't help but do these things. You have accelerated depreciation for oil and gas wells. You can have accelerated depreciation for wind turbines. You could do the same thing as anybody who falls into the issue knows, but 80% of the wind and solar investments would go away if they got the same class of depreciation and credits that hydrocarbons get. Warren Buffett famously said they're not wind farms or tax farms. They're harvesting tax revenues.

The logic behind the subsidies and the credits was that they were immature industries and needed to be catalyzed. You can make that case, and it's a reasonable case to make. They've been catalyzed. These are trillion-dollar businesses. Please stop it. We'll see how it goes. From here, it may level the playing field so there's an equal class of credits. Even with the credits pre-Trump threat to kill the credits in the Republican House and Senate here, the cancellations in wind plans all over the world have been escalating because the input costs were rising, and the offtake, the buyer, the ISOs utilities, and states just weren't willing to cover that cost.

The idea that these were cheap and getting cheaper was made into a fiction that they were cheaper than they were a decade ago but still not getting cheaper. I think it's an accelerant to something that was already going on. If they disappear or get phased out more quickly, then I guess that will accelerate something already happening. It doesn't mean there won't be more wind and solar. There are many places in the world where wind and solar can stand independently.

If I were guessing at the macro level over the next decade, I think the net contribution of wind and solar to the world's electric supply will at least double. It might triple, which would be a very significant industry. It's a big global market, but it's not a transition away from coal, oil, and gas.

Ed Coyne: Before we wrap it up, I want to touch on two more things. One, quickly, any new tech out there that you're following, you're on the front end of this thing that you're excited about or think could be interesting as it relates to energy or tech?

Mark Mills: This is a tee-up for my book, baby.

Ed Coyne: Oh, there we go. That was my second question, so this is perfect.

Mark Mills: That's all my book's about. My book is about what's already been invented, essentially pre-commercial. You could predict the future based on what's already happened, which is the Peter Drucker aphorism. Two things I would talk about in terms of the supply side. Everybody is talking about small nukes; I don't mean modular nukes, but micro nukes; they're inevitable and exciting. We're still a decade away from anything at scale, but it's coming now. There are 80 different designs around the world. Probably 30 or 40 will work well. A dozen will show up. It's going to be exciting on the supply side.

On the demand side, we're at the end of the beginning of a cloud era, not the beginning of the end. The amount of expansion is now coming in demand for electricity and all the material inputs to making a GPU. We now manufacture semiconductors by the kiloton globally. Chips that weigh a gram, we make them by the kiloton and use 1,000 times more energy per ton to produce that steel. Put differently, we're manufacturing the equivalent of megatons of steel in energy terms to go into the devices and data centers. We're just beginning to do that.

From an economics perspective, the subtitle of my book, which I still stand by, is The Roaring 2020s. I think we're at an economic inflection because of technology of consequence. We will become more energy efficient in the future, fueling more demand, not eliminating it. That's the old Jevons paradox. It's not a paradox. It will fuel more wealth, which fuels more demand for all forms of energy. It will fuel enough wealth that those who would rather have windmills and expensive batteries will afford them.

That's great, but there are billions of people in the world who can't, and they'll be thrilled to have cheap oil and cheap gas made possible courtesy of AI that improves the upstream management and robots to fill the labor shortage in all the industries, especially the industries that build everything. All the mining and oil and gas industries are full of aging skilled trades. Wages are going to soar there, which will attract young people to it. You'll have fewer people amplified by virtual and physical robots.

I write about this a lot in my book: creating more wealth and energy demand. It's a virtuous circle. People think this circle is somehow destructive: "We shouldn't use more energy." No, it's a virtuous circle in which wealth creates demand for energy, innovations that make it more efficient, and allows us to do things more cleanly. There's a one-to-one correlation between clean environments and wealth.

Ed Coyne: You talk about this also. Think about the ecosystem of the U.S., Europe, and other parts of the world that are ever-expanding into third-world countries that are just getting electricity for the first time. We haven't even touched on that. This leads me to an invitation to this fall or maybe the first part of next year; we'd love to have you back on because we've just scratched the surface here. Honestly, I haven't even got through half my questions because we keep going into these different corridors that I think are fascinating. We'd love to have you back if you'd be willing to do that and revisit some of these topics.

Mark Mills: Let's do it.

Ed Coyne: How can our listeners keep track of you, follow what you're doing, and get access to your most recent book? How can we get ahold of you and track you down?

Mark Mills: The books on Amazon, Cloud Revolution, Mark P. Mills. That's the easy way. I have a website called Tech-Pundit, tech-pundit.com, which has the whole canon of my writing. Our National Center for Energy Analytics has a website and tracks what I'm doing now with that new think tank I created.

Ed Coyne: Mark, it was a treat to have you on. Thank you so much. I look forward to part two in the not-so-distant future.

Mark Mills: Thanks for having me. I greatly appreciate it.

Ed Coyne: Once again, my name is Ed Coyne. Thank you for listening to Sprott Radio.