How AI Is Reshaping Energy Markets (in 15 Minutes)

So energy is the single most fundamental asset or commodity to the way we live. I'm recording this on a camera that was charged with a battery. You're probably watching this somewhere where the lights are on above you. That's energy. There's cars outside. People are driving. Energy. Right? The world runs on energy. And AI globally is completely restructuring how energy markets work, the demand for it, and the supply as well. So in this video, I want to take you through that story of how AI has affected the energy markets, what you can take away from that personally and in your career, and then what that will look like in the future. And this one is very personal to me. I work at a startup where we are trying to drop the cost of energy by building power plants specifically for data centers and for AI, at least initially. So I think a lot about this space, and I want to make this more of a friendly just story style, hey, how did we get here, the wide kind of high level context, and then again, where we're going. Now, before we get back to the video, I want to tell you about the sponsor of today's video. And this one has one of the most incredible pivot stories you will ever hear of in the public markets. A year ago, this company went by CleanCore Solutions, and these guys sold ozone cleaning products. And fast forward to September 2025, they launched a $175 million fund, a Dogecoin treasury chaired by Alex Spiro, who you probably know as Elon Musk's lawyer. And then in 2026, they pivoted again, they brought in a new CEO, a guy named Tyler Hasson, who had spent two decades in the industrial and oil fields world, and served in the Department of the Interior. And then they sold all the Dogecoin and on September 1st, the company became Zone Frontier or ticker ZONE on the NYSE. So what does Zone Frontier do now? Well, they build AI data center campuses. And the thesis is pretty simple. It's that the bottleneck for AI right now is not chips or models, it is upstream. It is land, power and permits. And what Zone does is they develop the campuses and lease capacity to AI companies like Hype Scalers who bring their own chips, or they partner with site owners who already have power and bring the capital and the customer. And they have two existing flagship projects. The first is a campus in Minnesota, it is 100% pre-leased to Cerebris Systems, the AI chip company on a 10-year co-location agreement. And that contract alone is worth about $800 million over the term and $3 billion if both of the renewal terms get exercised. There's already about 20 megawatts of power exercised on site, and the rest comes online in Q1 2027, and that's when Zone expects their first revenue. Now, the second is a joint venture in West Texas with HST Technologies. This one starts at 200 megawatts, it can expand up to 500 megawatts. And Zone owns 99% of that JV with room for up to $2 billion in capital commitments. And the deal closed within weeks of the LOI. And these guys are saying that more project announcements are in them. Now, on funding, they closed about a $100 million raise in August. And with the Dogecoin proceeds, that's about $140 million for Minnesota, and management says they do not expect more diluted financing for that project. And that number stands out. While I'm saying this, the market capital of Zone is about $72 million. And that's against, again, an already signed $800 million contract. So that's the story of Zone Frontier, that is ticker ZONE, Z-O-N-E on the N-Y-S-E. Now, their investor page is linked in the description or on the screen here. So you can read their filings yourself. This segment is sponsored by Zone Frontier. And nothing here is investment advice. You should always do your own diligence. And that said, let's get back to the video. So first, what is the actual tie between AI and energy? Well, we know that AI is a function of compute. Computing power. You can think of that like how AI thinks. For us, it's brain power. We need food to convert into energy to think. And what AI or artificial intelligence needs to think is compute. And compute physically takes the form of GPUs, of CPUs, TPUs, all the PUs. But that's things like what NVIDIA, AMD, Google, all of those companies make. And what's in a data center is just that. It is big racks of GPUs and of CPUs that power AI, that are the brain of the artificial intelligence platforms and models and machines. So if we know that AI is essentially compute, what is compute, right? What are the inputs to making that happen? And the predominant one by far, really the only one, is just power, energy. So like, so AI is really just compute. It's a bunch of GPUs sitting in a warehouse somewhere. And a bunch of GPUs sitting in a warehouse somewhere are really just power, right? That is the input that turns them on, that makes them work. And like so, AI is very fundamentally regulated by the energy market. And so the cheaper energy is, the cheaper it is to run AI models and vice versa. And that's kind of the core of why there is such a relationship here, really because they're one and the same. AI is made possible by energy. And now, and I'll get into this shortly, energy is made possible or new energy by AI. But before we go there, there are two opposing forces that regulate how much energy AI needs that I want to touch on. Now, the first of those is new compute or chips, right? Again, your GPUs, your CPUs, your TPUs. They have historically needed more and more power over time. What you can see on this graph are the NVIDIA GPUs and their power consumption plotted over time. And you can see the 2026, the RTX 590, 50 to 60% more energy than 2022. This doesn't even include, I think, the whole Rubin series, which is vastly more than this. So you're getting kind of this exponential curve of just the more time goes on, the better chips, the better compute is made to power AI. And the more energy those chips take. So that's one force, right? Which says, as AI continues to evolve, we're going to need a lot more power, a lot more energy, electrons to power it. But there is a force on the other side that works in the exact opposite way. And what that is, is how efficient those chips are, e.g. that compute. And I think the human equivalent is how much food do you need to perform well? If you need to eat 10,000 calories a day to think at a high level, then you are less efficient than someone in terms of calories to intelligence as a ratio than someone who needs 1,000 calories a day or 100 calories a day. And in the case of humans, we're bounded by these biological limits. Everyone needs roughly that same ratio or the amount of calories you need to eat per day. But in the domain of artificial intelligence, there is no ratio. It's just how good are we at building these chips? And so the chart that you can see now is just that. It's how efficient these chips are over time, this compute. And you can see they have got radically more efficient over time. You're talking like 300% in the past six years. And what that says is, again, the opposite side of this. As AI evolves, chips get more efficient and that pushes down the energy requirements. So over time, how much energy AI uses will be a function of the increasing power demand of compute versus the efficiency of compute. And that's the tech context you need to know to just understand how AI and energy interact and what that future looks like, what the technological developments over the next 10, 20 years are going to look like. But then what's the actual reality on the ground? Well, first, there is not plentiful energy. There is not this super cheap, abundant energy that is flowing around everywhere that makes energy costs super cheap for AI. AI is really new. And what happened is for 40 or 50 years, the US increased their electricity demand by a few percent a year in line with population growth. And then AI came about and your curve that looked like this of electricity demand started to look like this. And power takes a long time to build, to permit, to plan, to finance. And that's created this massive supply demand mismatch where the whole country, right, and the whole world in some ways, although it's really focal for the United States, is working to bring new sources of energy online to power AI and to power data centers. And you can see that with this chart, which is what the demand for generative AI is. We are at the very early stages of a massive ramp of new electricity generation and production, which will go to powering the compute that powers AI. And how much power we need long term for AI will be a function of those two competing forces we talked about. But it will also affect the rest of us in other ways, in how much energy we have and therefore the cost of our residential bill, right, your energy bill that you pay every month. And also how much we pay for things like manufacturing, if you're a business with a factory, or just to keep the lights on in your, say, local business. Now, the sun is getting in my eyes, so I'm going to lower the blinds here. That is much better. So basically, that's why AI needs energy. That's why we're building a lot of new energy now as a country. And what it means is, and the next question, well, how specifically is this new stuff being built? And generally, it's being funded in two ways. Number one are VC dollars, right? So venture capitalists are investing a lot of money in helping new companies and more established companies bring power online. And the number two, the profit or the free cash flow of hyperscalers, so that's your big tech giants for Google, Amazon, NVIDIA, companies like that, is transitioning into the world of hardware. Those companies are financing the build-out. So you have those two different sources, right? And you can throw Anthropic, OpenAI, all the labs into the venture side of it because they are venture-backed businesses. So those two groups of players are the ones who are putting all the money in to build out new energy. That can be things like nuclear power plants, solar, wind, natural gas turbines, in some places, coal, right? A lot of dollars flowing into those types of generation and bringing new ones online to power data centers. So if I had to summarize the three ways in which AI is changing, then the type of energy mix or how we're going about that financing and building. Number one, like I said before, is just a lot more demand. We can expect a lot of demand for energy for a very long time, not just for AI, but for general electrification, robotics, et cetera. And number two, which is a little bit more interesting, I think, is bring your own power. And that is this theme of these companies, your OpenAI, Anthropic, Google, Amazon, when they're building their data centers, they're financing them, they're bringing their own power with them. So you have this line between being a technology company and an energy company that's starting to blur, right? Google is running their own power plant. So are Amazon, NVIDIA, Anthropic. It used to be a very separate market. You have your energy companies and you have your technology or your data center companies. But because of this massive demand, that has done a 180. Now it's the complete opposite, where these companies are becoming power and infrastructure providers themselves. And so either VC dollars or the free cash flow of the hyperscalers of the tech giants, that's all going into making a lot of these companies power providers as well as technology companies, which is a very kind of new concept, at least to the scale on which it's currently operating. And the third and final difference in how AI is reshaping energy markets is community pushback is now a big thing because people are scared of AI. They're scared of data centers. They don't understand a lot of the nuance there, or they do, and they just don't like it. And so that affects how companies go about building energy. For example, people generally like renewables more so than dirty fuels, right? So maybe these companies are starting to build more clean renewable energy versus fossil fuels. It also affects siting or where you're building all this different generation. But that is now a top of mind thing for anyone and everyone building and investing in power, and it wasn't a few years ago. So that is all the context. That's the story that got us to where we're at. And where we're at now is AI adoption across the board is continuing to accelerate. Models and chips and GPUs are still all getting better at a almost increasing clip. That's both in terms of performance and in efficiency, as evidenced by the trend toward open source models, which are still really, really, really good, but also very, very cheap and efficient with energy usage. And then also the earnings of AI companies and general companies in part because of AI are doing really good here in the US, which means stocks are pumping, and that means more money for infrastructure, for energy, this loop that just kind of reinforces itself. And like I said, we're still in the earlier stages of this data center and energy build-out, and we're looking, we're staring down the barrel of hundreds of gigawatts, which is basically the entire US grid, at least half of the US grid that will be built anew over the next few years. And so what's taking place and what will take place over the next five, 10, 15 years is the single largest infrastructure build-out in American history. That's crazy. And a lot of that, again, comes down to not just the fact that you need to build data centers, but that you need to build the energy to power the data centers. And so it's that connect again between AI and energy that is driving these trillions of dollars in investment. And so the world in 15 years might look very different. You might have a lot more data centers and also a lot more electricity, more electrons, more energy. And what that means is more electrification for next generation technologies, electric vehicles, robotics, advanced manufacturing. All of that will be much cheaper if much more energy, many more electrons are brought online. So it's a pretty exciting future. I hope we can all keep that in mind. There are challenges. There are things to discuss, but it also is very exciting. There's a lot of work to be done and a lot of work to be done specifically in that domain of energy as related to AI. That's all for me. See you guys in the next one. Cheers.