Welcome to The HPC Debrief where we interview industry leaders that are shaping the future of HPC. As the growth of AI continues, finding power for data centers is going to be a major challenge. Nuclear energy will almost certainly be on the list of solutions. Listen as James Walker explains his company’s progress with microreactors as a way to provide clean, safe, and portable energy to data centers and more.

HPCwire: Hi, this is Doug Eadline, managing editor of HPCwire. And this is Executive Debrief where we interview industry leaders that are shaping the HPC and AI world. So today, I’m honored to have James Walker, CEO of Nano Nuclear Energy, and he’s going to talk to us about his products for powering data centers. Now, before you start wondering why is HPCwire talking about nuclear energy– a little background should help.

As a leading HPC news site, we have been following the surge in data center growth, largely due to AI and including HPC, and because most of these existing data centers and surrounding locations like, for instance, in cities, established areas at universities are running out of power options. One of the questions we keep asking is where will you get the power for your warehouse full of GPUs? In our analysis, if the growth of HPC and AI continues, and we think it will, nuclear power must be factored into this equation. I don’t see how we can do it otherwise. So today, we’re interested in learning more about microreactor options for data centers. So with me I have James Walker, and welcome James. And let’s talk a little bit about the company and products. Where did Nano Nuclear Energy come from? And a little bit background before we get into some deeper questions.

James Walker: Sure. Thanks, Doug, and thanks very much for having me on the show. A little bit of background about the company and myself. I’m the CEO of Nano Nuclear Energy, and my background was as a nuclear physicist and nuclear engineer. I got my start in the industry by working for the UK government, building manufacturing facilities to produce reactor cores for submarines. And I was also seconded to Rolls-Royce. For a while, I worked in the design of nuclear reactors. And so when I was in North America, I teamed up with the founder of the company, and that would be Jay Jiang Yu, who had a bit more of a background in financing. But together we built the company, and it was always a very commercially focused company that was looking to produce commercial products for industry, and we settled on a very portable and small micro reactor solution. Basically as small as you could possibly make a reactor system, because we were looking to create energy solutions for remote locations. So mining projects, remote habitation, island communities, military bases. And we thought this was a this was a huge potential market that we could move into. And since, obviously, we began down that road, the conversation around supplying nuclear power to energy centers. But data centers, AI centers, text tech hubs, even Bitcoin mining; that’s become an increasing portion of the conversation around the future of where a lot of this energy generation from the new small modular reactors and the micro reactors are coming from.

HPCwire: And when I first heard about you, I looked at your website, and what jumped out at me was here was a truck, which is what I assumed was a microreactor on it, that, as far as I could tell, could deliver nuclear power wherever there was a road or maybe a half a road, so forth. But it looks like its mobility is a key feature here.

James Walker: You’re exactly right. And by road is actually intentional because we wanted to confine the reactor to an ISO container system so that we could move it by road, by rail, or by or even with an ISO container and put it on the back of a shipping vessel and, and ship this thing anywhere in the world. And … the advantage of being able to self-contain this system was that if we could move it very easily, we could actually put it down and plug it in very easily too. So micro grids in remote locations, you know, remote habitation where people are subsisting off diesel generators, all these kind of areas where there’s essentially no competition for diesel, for diesel and diesel power. Microreactors would be the first competitive product that could actually come in and compete with these diesel systems in these remote locations. So, very much the focus was on moving by road, by rail and those kind of areas.

HPCwire: It’s very interesting, I assume, making it easy to get it in. Also makes it easy to get it out. So when you have to decommission, or you’re done with it, it’s easy to take away versus that. You’ve got a let’s just call a nuclear site that is has a building with a reactor in it, let’s say that now has, you know, warning signs all around it.

James Walker: No, you’re exactly right. It’s … not just the, the decommissioning either. It’s, we wanted it to be obviously as commercial a product as possible. And that means the reactor needs to be able to go in there, be able to operate using as few people on site as possible. And so even the business model we’ve been looking at is a central location that monitors a lot of the behavior, or it could be hundreds of deployed reactors at any one time with the ability to be … mobile, remotely operate these things. But that means low amounts of infrastructure need to be installed, low amounts of staffing need to be there. And when it comes to the termination of the reactor lifetime, the removal of that spent fuel and essentially the disposal of the reactor hub, which the fuel is in, becomes a very simplified and easy process. And if you can get all these things in line, then you can create a product that is very competitive with remote diesel costs, which can actually be very expensive. So, you know, you can see communities of a few hundred people that use $10 million of diesel a year just to, just to sustain a few hundred people. And over the lifetime of our reactor that it could [cost] $200 million, which is a multiple more expensive than the anticipated price of our reactor system to be put in there. So even a business model where we just sell the power and we keep ownership of it, and we operate it, and we remove it at the end of lifetime. Then it becomes an increasingly appealing product for industry and customers, too, because who wants to operate a reactor and deal with a reactor and decommission a reactor at the end of lifetime?

HPCwire: That kind of leads into my next question. And I think you’ve answered most of it, is that, you know, when I think of nuclear reactor, I think a big control room, lots of people with white jackets on. And so what you’re saying is you can, in one sense, park this next to where it’s needed and connect the mains. And as far as the customer is concerned, it’s just where the power comes from. And the rest can be managed by you, or I assume the customer can help with the management if they choose to.

James Walker: That would certainly be the intention. For instance, if there needs to be security, that could be something that the customer could be more interested in, but certainly, from our discussions with potential customers, there seemed to be significantly more interest in just buying power over a contracted period of time to avoid any sort of big upfront capital costs and then the responsibility of operating it. And the advantage, I think, that microreactors give is that, as you say, when you think of nuclear power, you think of these big installations, and these things typically will generate enormous amounts of power. But the advantage when you shrink it right down to a microreactor level is that the risks and the mechanical issues that could give you issues with a large reactor system aren’t present with a small reactor. So with our reactors, for instance, if every mechanical component was to break within the reactor, then you can’t get this sort of overheating that causes the core melt that you got with something like Fukushima, you would just sort of passively radiate that heat out. So it’s actually an incredibly safe system. And, like I like to point out already that, if you look at deaths per gigawatt hour, nuclear is already safer than any other form of energy ever devised. But when you get down to the microreactor level, it’s still safer, even still. You know, it’s so the operation of the reactor becomes very easy. Very few people are needed at site and a central location where you can centralize your engineers and your physicists can obviously intervene and monitor and operate the behavior of all of these reactors, these deployed reactors at any one time.

HPCwire: Another thing I thought about was … if there’s either a service life or you need to refuel or so are these like big batteries in that okay, this one’s reached its end of life. It needs refueling or maintenance or something. We can just remove it and put another one in. Or is this something that gets serviced on site or something of that nature?

James Walker: So it’s very likely we’ll have technicians that are deployed to just intervene and provide any physical intervention, if anything is needed, like, I don’t know, a bear damages it, or a tree falls on it or something. I mean, these are still not going to cause significant issues. But certainly, the intention here is that, yeah, very few people would be needed on site. We don’t want to actually have a system. Well, it looks more likely that we won’t have a system that will need to refuel the reactor. It’s more than likely that at the end of the operation of this reactor and the operation of each one of these things is going to be about 20 years. We would just, first of all, remove the fuel that’s been in this reactor, then the reactor would, would likely stay in place for about six months. And then actually it can just be scrapped because, you know, there’s nothing toxic or radioactive about that attempted reactor, and if necessary, a new system would just be brought in, put down and plugged into the local microgrid to replace it. But that way you avoid any sort of more complicated operations around refueling exercises, which is something we don’t see this being practical in a lot of remote locations that we’re looking to deploy to. So if it’s a mining site in the middle of Alaska, it’s much easier just to replace that reactor and take everything back to a central location, a central manufacturing base.

HPCwire: Okay. So I wanted to also mention I read your news release about the memorandum of understanding with BlockFusion. And obviously they’re interested in data center use. And the one thing that struck me was it was Niagara Falls … that’s where US electricity production started, and it still is, you know, a huge electricity producing area. And so I would think maybe out in the desert somewhere, but in Niagara Falls seems like a strange place for a company to need power. So, I’m talking about block fusion in any case. So maybe a little bit more about that. And, I’m curious what their project looks like.

James Walker: I think they are making some very fairly sensible decisions because the amount of power that’s going to be required by tech centers, data centers, AI centers is going to be quite enormous. But that essentially means that you either develop your own power systems or come up with solutions, or you basically take from power. That would be going towards big grids, cities and conventional outputs of power and places where they would be able to utilize their power. It’s only going to be a limit with all of these companies, whether it’s Google or Microsoft, where they’ll be able to draw from a mainline grid without one essentially detracting from those systems. So in the case of Niagara Falls, obviously that provides an enormous amount of hydroelectric power, but it also is hydroelectric power that’s also being generated for cities and towns already. Now how much are you allowed to steal without inconveniencing those places? There’s only so much. And, you know, even if you’re looking at, say, it wasn’t exactly Niagara Falls, but even a little bit more removed from the city, and you’re installing transmission lines. Those transmission lines still have to take from the main grid. And if you’re taking from the main grid, then you’re going to need to supplement the power that you’re taking off those grids. So whether that’s coal or gas or solar or wind or geothermal or hydro, it needs to be put back.

So you either come up with a solution to de-risk your own business, or you just hope that you can take from the grid, pay their bills, and they’re going to come up with ways to supplement their own power that you’ve essentially taken because some of these envisioned data centers are going to use as much power as a city. And I think, even if you look at Google or Microsoft or those sort of groups, if you were to imagine them as countries, I think they’re one of the largest energy-using countries in the world. I think something like 30th or something like that. So they use more power than, say, 150 countries in the world. And like if you consider that, you can see that nuclear solutions that obviously they can provide an enormous amount of power, very consistent power. They can can be installed where they complement the power that’s being outputted by conventional … outputs as well. So if these big sensors like block fusion are drawing power, they’re not inconveniencing people and causing power shortages for a lot of these towns, cities and conventional places.

HPCwire: All right. Well, that answers that question. And, that’s kind of what I thought, because everywhere I’ve been, reading about this in the local data, local power companies, you know, they’re used to adding a neighborhood of houses, maybe a new plant somewhere. But the data center requirements just seem to be gargantuan in terms of these other amounts of power that they’re looking to provide over time. So, yeah. And going back to my point that when people are saying, you know, you’re going to need 3 to 4 years just to buy a transformer to put in whatever you want to put in; it just changes the equation entirely. So another question I had is, and I assume, you know, you’re probably neck deep in studying a lot of this. What kind of regulations are there about, say, moving a microreactor across state lines through towns, cities, population centers? And … is that going to be something that’s going to hinder your ability to get these things placed?

James Walker: I think with a lot of these microreactors that are being envisioned, there’s an assumption that the nuclear regulator is going to change the regulation, and they’ve stated that they will do this to allow for a fully fueled microreactor to ship. So that almost certainly will happen in time. And that’s being pushed very heavily. And, it’s probably going to be, partly bolstered by the fact that Project Pele, which is a military application for microreactors to supply military bases, is going to be a fully fueled reactor that’s going to move by conventional road systems. So when … we actually began, we didn’t actually make any modifications based on an assumption of change of regulations. We envisioned that we would actually deploy an MT reactor, which is essentially not a nuclear device at all. There’s no nuclear material in it, so it’s essentially moving metal, or if you see what I mean. So there’s no nuclear regulations around moving that empty reactor system and to actually move fuel that’s already moved conventionally by roads already. So there are basket and caste systems to form transportation systems that move fuel rods and fuel plates around the country already, and they’ve been doing that for many decades, like since the 1950s or maybe even before that.

And these things are incredibly robust, like hit the middle missile, you know, truck falls off a bridge. You know, you can’t get sort of any sort of chain reaction to create sort of a runaway fission reaction to generate sufficient amount of heat. And so the business model we’d envisioned was actually just moving the reactor to site, putting it down, and then moving the fuel rods in, inserting them, and then moving out again. Because that way you have an it abides by all the existing transportation regulations. It simplifies the process enormously and doesn’t rely on changes in legislation or regulations. Now, obviously there is going to be significant advantages when that regulation is changed and it will … admittedly be changed to allow for a fully fueled reactor to move, but that would obviously make our lives a lot easier. Then the challenge would become road weight limits. So the challenge here might be that the regulator makes the changes, and then you would need to just be strategic about how you move a fully fueled reactor to its destination site by not exceeding, like not overly exceeding, the road weight limits for these systems.

HPCwire: Oh, okay. That’s, yeah, that’s a good point. So, a question I don’t need specifics. I’m curious: when do you see this becoming commercially available, or maybe some test sites popping up in the future?

James Walker: So I still believe with regard to microreactors, and I classify a microreactor as anything below 20MW of power that will produce the first commercial product to go to to be sold to industry. And I would say our timeline, currently at the moment we’ve finished the detailed design work, and now we’re looking at doing demonstration work to verify all the modeling we’ve been doing. So we’ve started thermal conductivity testing, rig construction. And that’s essentially to collect the data. We need to then go to a licensing process. So pre-application with the nuclear regulator is going to start very soon and then as we start that pre-application information, we’ll be doing that demonstration work to eventually create a prototype. And that prototype, we hope it will be done sometime around about maybe late 2026, early sometime around about 2027. And then we would enter a formal licensing process. Now, currently the licensing process is envisaged to be about a 40 month process. So call that just under four years. So, say we start the formal licensing process in 2027. That moves us out to about 2031. Now, I should preface it with that there is a mandated change being introduced to the nuclear regulator to shorten that 14-month time frame to 25 months.

And so, obviously that could result in the deployment of a commercial microreactor a lot before 2031, maybe, 2029 sometime around about then, just going by that change in regulation that’s being advised by the NRC. Now, just to be prudent, we’ve not changed any of our timelines until we see those changes being implemented and being executed. So we’re envisaging the early 2030s just based on existing regulations, without any changes. I would say for the larger systems, they are more advanced. So small modular reactors, and you’ve probably heard of the Terrapower’s of the world, the x-energy, the Kairos, and those kind of even General Electric are doing a small modular reactor that a lot of those systems have started formal licensing processes already, which means that they will have commercial license products ready to deploy a lot sooner. And so before the end of this decade, I believe you’re going to see small modular reactors deploying. And a big customer is going to be data centers, AI centers, those kind of areas. So small modular reactors, the much bigger system, the much bigger advanced nuclear systems just before the end of this decade, microreactors early 2030s, I think, is a prudent timeline to see these things deployed.

HPCwire: That’s very interesting. And because I know any time you hear the word nuclear, you think, well, decades away just because of the overhead and the paperwork and all that kind of thing. I also want to before we finish up here, give you an opportunity to speak to safety, because I believe a lot of people in their mind, as soon as they hear the word nuclear, jump to visions of Chernobyl or Three Mile Island and, I also chuckle, back in the day when, there was, uh, nuclear magnetic resonance while there still is. But when they decided to use that for humans, that the word nuclear went away. And, you know, so that’s how sensitive the industry is to that word nuclear. So, and I’m well aware of seeing your website, some of the things you mentioned, but I think, you know, give you the opportunity to speak to the safety, which I know you’ve already talked a little bit about. You could have total failure of everything, and things would be fine. So I think just in case anybody’s sitting there with concerns about, you know, is this thing going to blow up and take out my neighborhood. Clearly it won’t, but I’ll let you, give a little bit of background on that.

James Walker: Sure. So, the good part here is that it’s usually fairly easy to communicate, just how safe these systems are. And I would say that just because the difference between the reality of nuclear power and the perception of nuclear power is so divergent that, it’s a wonder that, I mean, it highlights just how bad the PR for nuclear has been. So, for instance, I mentioned it earlier, I think that if you look at deaths per gigawatt hour, nuclear is already the safest form of energy that’s ever been devised. And that includes even wind and solar in that equation too. And when you talk about accident disasters like Three Mile Island or Fukushima. These are also incidents where nobody died. You have clean-up operations, which can be messy, but they’re not, you know, even in the complete catastrophe, which is the worst-case scenario, can be a core meltdown. You have a cleanup operation, essentially, and the plant is useless. But even those kinds of accident scenarios where you get a core melt and not really possible with a microreactor because the power generated is not enough to actually melt the system. So you don’t even if every mechanical component was to break, as I said, it would passively radiate that heat out. So you wouldn’t even have that cleanup operation and that melted core. And so it’s this these items come up quite a lot when we talk about nuclear and it you know, I think that’s probably fair because, a good job has never been done to communicate just how safe and efficient … this power is. And you know, this is the power source. It’s not just the safest but has the highest efficiency of consistent output of power, the highest capacity factor of any power.

And even when people talk about waste, they forget that if you were to take all the reactors that have ever been operated in the United States, and I include the submarine reactors, the aircraft carriers, all of those ever since the 1950s. So everything and you were to put all the waste in one place. It wouldn’t fill a football field. It’s a very small amount of spent nuclear fuel that is generated by the operation of these reactors. And it’s the only type of power where, you know, 100% where all of your waste is at any one time down to the last atom. So it’s very safe to say that it’s the safest and it’s the cleanest form of energy we’ve ever come up with. And, I think if we can navigate, long licensing periods and upfront capital costs. Theoretically, nuclear should be the cheapest form of power out there, too, because the energy density within uranium is incredible. Like, if I was to tell you that one of our microreactors would operate for 20 years and, it would only burn up 1% of the fuel in the reactor after that 20 years of constant operation. You know, you would think it was an audacious claim. But it’s true. That’s how dense the power of, energy within uranium actually is. So hopefully we’ll begin to communicate just how safe and how clean this power is. So there’s a broader acceptance and public acceptance of just how beneficial this can be for just the countries, but for humanity generally, because more power being put out there means a higher quality of life for everybody.

HPCwire: Yeah. And you were talking about some of the alternatives, which would be diesel engines, which, you know, they’re are not very clean. So yeah, I think that, that is something that, comes into play and I, I know, as I have a science background, I end up explaining to a lot of people that, no, it’s not going to be like Chernobyl or, you know, Three Mile Island. I said, do you realize how long ago these were designed and so forth and so on? So, yeah. I agree, I think it’s actually one of the best technologies to solve a lot of our problems on the moving forward. So, in terms of power. Okay, that’s been great. I don’t know if there’s any closing comments you’d like to make, now’s your chance.

James Walker: Oh, Well, I mean, we covered quite a lot. I can’t think of anything off the top of my head, but thanks very much for having me on the show. Look, it’s always good to be able to communicate more about nuclear. And I think the benefit that it will bring to not just the tech industry, but I think countries and societies in general.

HPCwire: Yes, I would agree. So, thanks, James, for answering a lot of our questions. And I’m convinced we’re going to see nuclear power and data centers. Before you know it, it’s just going to happen. So, Yeah. Agreed. All right. Thanks a lot. And we’ll let you get back to your timeline.

James Walker: Okay. Thanks a lot, Doug. All right.