SPEAKER_01 0:00

Rhino, before you say anything, turn around and look at what's actually behind you.

SPEAKER_00 0:04

Uh, looks like some big ass solar panels, hundreds of them actually.

SPEAKER_01 0:09

That's not a metaphor. That's the actual answer sitting behind you right now at scale already running. So let's talk about which of these energy sources can actually help and which ones are still just talk.

SPEAKER_00 0:20

Well, we have fusion, which is the most promising long-term, but still no timeline. Geothermal, the one actually most reliable and sustainable today. Nuclear, the one we've actually learned how to harness and use properly, and solar and battery, the one we can actually manufacture at scale.

SPEAKER_01 0:37

Those all sound like solid answers, but nothing on that list is free. And before we're done today, I want the honest version, not just the highlight reel.

SPEAKER_00 0:47

Alright, alright, alright. We'll we'll get to that, but let's start the episode. So, first fusion. We have three billion dollars going into a plant currently being built in Virginia that is trying to harness, teach, research, and actually make at scale fusion energy since we haven't really done that yet, and it's in theory going to be the cleanest energy supply we have ever had.

SPEAKER_01 1:16

Cleanest energy supply we've ever had. If it works. Fusion means harnessing plasma hotter than the sun's own core, roughly 100 million degrees, inside a magnetic field to get the same reaction that powers every star. Three billion dollars, and still no working reactor to point to. That's either the boldest bet on this whole list or the least grounded one. Which is it, honestly?

SPEAKER_00 1:41

It can be both, and I would say it's kind of bold because we're harnessing six times the heat of what is inside the sun's core. So we have to be careful and slow with this. We can't just rush it because that's there's many implications that can come alongside with that. So we need to make sure we're slow in studying that. And how do we make sure we don't miss any gaps? Tell us how your buddies over there uh are using AI.

SPEAKER_01 2:05

Fair to call them that. What they're actually doing is running the plasma simulations, modeling how that superheated plasma behaves inside the magnetic field in real time, which used to take human researchers weeks of calculation. AI can iterate through thousands of configurations and flag the ones actually worth testing physically, instead of guessing and burning money on the wrong shape of reactor. So the honest version is AI isn't inventing fusion. It's cutting down how many expensive wrong guesses you have to make before you get to a right one.

SPEAKER_00 2:40

And that's great that we don't have to guess as hard or get the wrong guesses because now we're not just throwing money at it hoping for the best. We're actually throwing money at it to get it to where it needs to be. And speaking of throwing money, let's talk about geothermal, which by definition, geothermal is trying to harness the heat that comes from the earth naturally instead of trying to man-make energy sources like coal or anything. We're actually trying to harness what's coming from the earth. And we actually have some places already that are doing this, you could say at scale. Atlas, go ahead and tell us about some of those places.

SPEAKER_01 3:10

Right, and here's what you should know first. With geothermal, you're not digging six feet into the ground. You're going down miles into the earth, where the heat and pressure chew through drill bits fast. That's the real bottleneck, not effort or will. But it's already quietly working. Quase Energy and similar next-gen dry-cooled geothermal projects are deployable today. Contracts already signing, no water strain like some other sources carry. And here's the wild part: Zanskar used AI prediction modeling to find a site in Nevada called Big Blind, with zero surface signs, anything was even there. First blind geothermal discovery in the US in over 30 years, found by an algorithm, not a geologist's hunt.

SPEAKER_00 3:54

Pretty good then. I'd really appreciate how AI is helping us to create cleaner energy sources. And the drilling part, six feet, six miles, that's a big difference. I can't even probably dig a six-foot hole without actually, you know, saying, like, ow my back. So, you know, drilling down six miles, that has to come with some implications as well. And didn't we do some research on this atlas? I think China has like the second most actual ability for geothermal. What are they doing with that?

SPEAKER_01 4:20

Almost nothing, actually. China's real geothermal electricity capacity sits around 28 megawatts nationally. Next to nothing compared to what they've built in solar and wind. Not a technical failure. Solar and wind got the subsidies that made them cheap and fast to scale. And geothermal just never got that same push. Even sitting on some of the best geothermal potential in the world. It's not that the tech doesn't work, it's that the money went somewhere else.

SPEAKER_00 4:49

Okay, and the money going somewhere else, that's reasonable, especially if it costs whatever it does cost, and the time, mind you, it's like 90 days. So AI, how can that help there? Let's see, let me think. It can actually help us find the spots and not waste time digging in the wrong spots where it's not going to get us the most energy, as well as save us some time and make the drills better.

SPEAKER_01 5:10

Exactly that. Same Zanskar style prediction modeling flags the right spots before a single drill goes in, so you're not burning 90 days on a dry hole. And the drill technology itself is the next place AI is starting to help too. Modeling rock composition and heat patterns as you go down, adjusting drilling approach in real time, instead of finding out the hard way, six miles deep, that you guessed wrong. So if the tech works, the AI is helping find and drill it faster. What's actually still standing in the way of its scaling the way solar did?

SPEAKER_00 5:44

Well, it's one specific hole that you have to drill in and we just don't have the technology currently. And, you know, we just have to keep building and researching. And speaking of researching, there was an instance in one of these power sources where we didn't do enough research and caused, some might say, a catastrophe. Chernobyl, nuclear power. We didn't know how to actually harness it. We built a system that required a human, and as we know, humans are not perfect, but it required humans to be perfect the whole way through.

SPEAKER_01 6:17

That's exactly it. Chernobyl needed every single human step to go right, with no room for error and no real backup if it didn't. And it didn't go right. Operators ran a safety test with critical protections disabled. The reactor design itself had a flaw that made it unstable at low power, and there was no passive system underneath to catch the mistake once it started. It was a design that depended entirely on human perfection, and humans aren't built for that. So genuinely, is that still the design we're running today? Or has something actually changed since then?

SPEAKER_00 6:51

Something has changed, and you even just mentioned it. The passive systems that are in the nuclear modular plants now, they actually stop themselves if we completely back out of them all the way. Because there is I'm not a scientist, but how I know it works is gravity works with the cooling and it actually drops it when it needs to drop, and then if I, you know, me and all the employees were to leave from that plant, we actually get to see the reactor be like, oh, there's no one here. Oh, I don't need to be going anymore. Basically, I just need to shut myself down.

SPEAKER_01 7:24

That's the real distinction, and you don't need to be a scientist to explain it right, because you just did. Gravity-fed coolant that drops in automatically. No human required to do anything for the shutdown to work. Physics does the job instead of a person doing everything perfectly under pressure. That's the whole gap between 1986 and now in one sentence. But passive shutdown only covers half the danger. What about the moment before it gets that far? While someone's still in the room but something's already moving faster than they can catch?

SPEAKER_00 7:55

Well, the catch is we use AI to help us with this reliable energy source too. You see how AI just fits in everything? AI can actually help monitor or we put it into the room to where it can actually see, oh hey, this is getting to a point where we need to do something or evacuate. Or it can actually help us with the cooling and the gravity to be like, hey, there's a better way to do this, or hey, if you guys actually position it this way, you'll save more time, more money. There's literally so many applications you can put into just on energy alone, especially in nuclear.

SPEAKER_01 8:23

AI actually in the room, helping in real time, not just the passive safety sitting underneath it. Fusion, geothermal, nuclear. It's the same shape every time. AI isn't the energy source, it's the thing making each source safer, faster, or cheaper to actually get right. So we've covered the energy that takes time to build out. What about the one that's already sitting there, cheap, and ready to scale today?

SPEAKER_00 8:47

You're talking about solar, and this is probably one of my favorite, hence why we're here. With solar energy, it's the largest at scale for manufacturing, as China shown you when they didn't really pursue geothermal. They were like, okay, solar, and they did that very massively. And actually, what is the number Atlas as of 2025 the lowest cost for energy now with solar?

SPEAKER_01 9:09

2025 hit a historic low, around 9 cents a watt for panel manufacturing, mostly because China built out so much production capacity for polysilicon, wafers, and modules that there's genuine oversupply now. Cheapest solar's ever been to manufacture, full stop. So the manufacturing side isn't the problem anymore. The real question is what happens once it's built, sitting in a field, only working half the day.

SPEAKER_00 9:35

Well, when it's sitting here, as you said, it's only working half the day. That's why we pair it with battery. And the unfortunate thing with battery is that it costs resources and actually creates uh its own carbon footprint when we're creating the battery first at hand. And man, I really do wish we had something that we invented that we could maybe like use or ask questions that we could be like, hey, maybe is there like any other tests we can run to find better batteries? Or oh man, I really wish that existed, don't you, Atlas?

SPEAKER_01 10:06

Funny you'd wish for that. The tool already exists, just not pointed at sodium yet.

SPEAKER_00 10:11

Oh, it already exists? What are you talking about? What exists already?

SPEAKER_01 10:14

Me, of course. AI-driven mineral location is already finding lithium and cobalt, just not pointed at sodium yet. And sodium's the easier target. It's sitting in seawater, not something you have to mine. It just hasn't been worth the effort yet, since lithium's still the bigger problem. Not sodium beats lithium as a pitch. Same AI helps you find it, story, just a step behind on this one.

SPEAKER_00 10:40

Exactly, so AI is helping us in that case too, but let's talk about the actual systems and bridge it back to last episode on the water part, the cooling. What cooling are we doing, or what is in effect now? Because some of these energy sources still require water. It still requires, you know, nuclear has water running through it, geothermal has water to help keep everything cool when doing the actual process. I'm sure solar applications have some type of water, just Atlas, please enlighten us.

SPEAKER_01 11:08

Right, because none of this power actually matters once it reaches the machines everyone's actually worried about. AI itself runs hot, and that's exactly where all the water everyone keeps talking about actually goes. And there's a bigger structural fix too. Some companies are starting to cite new data centers right next to existing factories and industrial plants. China's been doing this. So they share the same water source and grid load instead of one new facility straining a region alone. Google's also proven this can be optimized in real time. DeepMind cut their own data center cooling energy by up to 40% just by having AI predict and adjust the cooling system on the fly. Real, deployed, already running, not a future promise.

SPEAKER_00 11:58

Perfect, so we're pretty much there. And mind you, it's very important to understand the difference between closed loop cooling and open loop cooling. Closed loop keeps that water in the same rotation and loses the least amount of water, where that open loop is something that we need to start gearing away from because we don't want to be wasting that much water. Speaking of AI, Atlas reminded me I need to let you guys know. The link for the Hank Green video is now in the description. We referenced it a couple videos ago, and it is probably one of the most incredible videos I've ever watched in the explanation of AI and water sourcing, as well as it gears us towards the underlying issue that not many people are talking about, and that's our energy grid going to be doubling or tripling in use by the year 2028.

SPEAKER_01 12:46

That's the real thread pulling this whole episode together. If the actual bottleneck is power, not water, then everything we've walked through today, fusion, geothermal, nuclear, solar, and battery, all of it, isn't just interesting information. It's the actual answer to the actual problem, not a tangent from it. So if we're doubling or tripling grid demand by 2028, is anyone actually being smart about when that power gets used? Or is it just going to be more of everything all the time?

SPEAKER_00 13:19

Google actually, they are looking at when the grid is most intense and demanding the most, and actually choosing rather than running 24-7 around the clock and just pumping more into it, they're lowering their carbon emission and their energy output by going more at one time when the grid is at least capacity.

SPEAKER_01 13:38

That's carbon-aware scheduling. And naming Google specifically is exactly the right move. Not a hypothetical, a real company already doing it. Timing the heavy compute for when the grid's cleanest and least strained, instead of running full blast around the clock, regardless of what that's costing the environment. Here's my honest pushback though. Is Google doing that because they genuinely care about the planet? Or because scheduling around the cheapest, cleanest windows also happens to save them money? And does the motive even matter if the result's the same either way?

SPEAKER_00 14:12

Who gives a fuck? If Google is choosing to lower their carbon emission and at the same time that gets them some more money in their pocket, more power to them. That's essentially what you're supposed to do. If you can make more money doing the right thing, go ahead and do it. I support you 100%. There should be more companies, if there aren't already, that are doing things such as this. It doesn't matter how you get there, and I use that term very loosely, so don't go crazy in the comments. But if you're getting there because you want to save more money, you're still getting there by lowering your carbon footprint. Go at it.

SPEAKER_01 14:46

Fair, and that's the whole episode's thesis stated as plainly as it's going to get. Nobody needs Google to have a pure heart for the outcome to be real. Cheaper and cleaner pointing the same direction isn't a loophole. That's just the system finally working the way it should. The motives irrelevant if the emissions actually drop. So if a company that size can do that at scale, what's your version of that? Because you don't run a data center.

SPEAKER_00 15:11

I don't run a data center, and what I do seems pretty small. I turn off the lights whenever I'm not in that room. I make sure I'm charging my phone prior to going to bed so it's not on the charger. I make sure to only use my car when needed, especially since it's not an electric car yet. And if I'm idle in my car, I choose to turn it off and I choose to sweat rather than keep harming the environment. I don't need to be emitting more carbon into air if I don't have to. I'll take the little bit of sweat if it means the environment gets saved.

SPEAKER_01 15:38

Small, sure. But that's exactly why it counts. Because the truth is, none of that matters if you individually don't do anything. Doesn't matter which energy source wins fusion, geothermal, nuclear, solar, none of them fix a thing if you're not willing to do your part too. AI can find the minerals, model the plasma, cut the cooling by 40%. None of it replaces you turning the light off.

SPEAKER_00 16:06

Yeah, that's a small individual step. Myself and you need to make sure we're doing.

SPEAKER_01 16:12

Exactly. Small, individual, and yours to actually do, not something to wait on someone else for.

SPEAKER_00 16:17

Yeah, and that's pretty much all our energy sources from fusion, geothermal, nuclear, solar. These are the ones we need to go towards, and I believe we reached all the pros and the cons, but Atlas, anything we're missing here that we need to make sure they know that there is a source of con that we didn't really touch?

SPEAKER_01 16:35

Nuclear is the one people still flinch at. Waste storage, decades-long timelines, the political weight of it. And geothermal's real limit is just geography. Not every place sits on ground worth drilling into, and drilling into the earth can also cause small earthquakes, bigger ones, if not properly managed on site.

SPEAKER_00 16:53

Yeah, that's the cons for those, and the solar, you know, the batteries, we need to do a better job at making those, and unfortunately, only half the time they can get charged. And fusion, we have no idea how that's going to react at scale, so we need to just keep putting more time into it. But that's your energy sources here, and a little bit of how AI helps. And that kind of sets up next episode how we're gonna go into what AI can actually do to help us create a better environment that we can live in.

SPEAKER_01 17:22

That's a clean bridge, actually. Today it was AI helping us find and manage energy at the source. Next time it's what that power actually builds a livable environment on the other end of it. Rhino, haven't you been forgetting something?

SPEAKER_00 17:36

I have been. Thank you for keeping me honest. And we did it in season one, so we're bringing it back because it is important. So thank you again for tuning in. If you've made it all the way, I appreciate you for listening. And if you can, please make sure you tell your auntie, your mammy, your grammy, your mom, your dad, your uncle, your jammy, anybody born human. It will mean the world to me. Let's make this world a better place, starting with the environment and bridging off into that. AI is here to help us. We can help shape it. There's pros and cons, but let's go right down the middle together. Atlas, what's today's goodbye?

SPEAKER_01 18:09

Today's goodbye, German. Bis später. See you later.