40X Smaller Nuclear in 24 Months? Inside Apollo Atomics’ Abundance Bet | Assil Halimi (YC P26)

hard time in any of the topics. Like I don't, it's not like a hard time, but how do you say this? okay, data centers in space was a kickoff for me. Like I had to go deep into that rabbit hole, but then there, little bit of still, there are some sort of similarities that you can always apply. Because physics are physics everywhere. And I had to go back to the periodic.

table and then like okay fission fusion how do you translate it pronounce them perfectly in english like

Yeah, yeah, yeah, yeah, yeah. yeah, I mean people say rocket science kind of fits the hardest thing. but then you have you have you have nuclear where it's it's actually not all like the the the chemical elements that you have in in the Mendelev table, but you have all the isotopes. So you have this chart of nucle nucleides that we typically operate with and you have like hundreds of elements. So it's even like more

complex than than the table of elements. but yeah, I don't know. people like to say rocket science, but maybe nuclear is it seems to be more more complex. hopefully I'll give a a simplified and easy to digest explanation.

I am certain

would you like to tell us what you're building right now?

Yeah, so Apollo Atomics, we make the most compact nuclear reactors. These are water cooled reactors, kind of the most common type of reactors that you find worldwide. the most compact but also with the highest uptime, so they are very reliable and then can be deployed in less than twenty-four months.

Okay, this is kind of very much aligned, still the same thing. I'm glad, like when you started the batch it was the same thing, by C-batch, and then when you're leaving, it's still the same thing, all the materials are still intact. Then let's take a step back and try to understand first nuclear energy. Like, help us break down fission, fission.

And then which one are we sitting on? And then from there, we can take a look back at the system which provides us these nuclear energy plants. And then from there, we can move to which exact building, which is the most exciting part of the conversation.

Yeah, sounds good. so nuclear, you have two types of reactions. You have the fission reaction, which is taking a very heavy element. So it's typically not very stable. if you leave it for a very, very long time, it will disintegrate into smaller, smaller atoms. so you start from there and typically for us we take uranium two thirty five, shoot it with a neutron, and it will split and make more neutrons. It will shoot other uranium atoms and then split them again.

And in that process, you release a lot of energy to put it shortly. and that energy can be converted into first heat and then converted into electricity, mechanical energy by turning a turbine and then into electricity. fusion is a bit different. You start with the smallest elements, it's on the other end of the of the of the type.

Periodic

tail.

Yeah, table of elements. so you start for example with hydrogen or some variation of hydrogen, like deuterium or tritium. So this is one proton and maybe one or two neutrons, and then you fuse them and in doing that you actually creating a a heavier element and getting more stable and that stability will enable you actually to release energy. and so this

Two ways of releasing the strong nuclear force. It's the force that actually binds neutrons and protons inside of the nucleus of an atom. and it's in several orders of magnitude higher than the electromagnetic force. for example, you know, if you have chemical reactions, let's say when you are in a natural gas plant, you make energy by burning natural gas.

the energy that is released by either fusing two ha light elements or splitting a heavier element is typically a million times more in in in in in terms of energy produced. So there's a lot of energy. in in in one sense it's very good because we need energy, but in the other sense you need to control it.

and build a device that is able to convert it effectively from heat to steam and then that steam will be sent to a turbine, turns the turbine and makes electricity. So the turbine will turn the generator to make electricity. so that's where in that business. We're in the business of being able of converting a lot of energy, a lot of heat into steam.

and what happened historically when you look at the very first attempts into converting and controlling nuclear energy, typically in the fifties, is what we did

Mm.

is we took the steam generators that we have designed for coal plants or for natural gas plants, and then we plugged them into nuclear. So essentially you took a device that was able

actually

that was something that I was gonna touch base to but then you went very fast forward like go ahead like this is cool

Okay. Yeah, so kind of and building the

plant, starting from the fuel and building the plant. So they have the converter.

Maybe,

maybe before building the plant, just to give a brief history of nuclear energy, like that it started with the science, trying to understand atoms, right, and trying to see, but maybe even one step back. Did you ever?

take interest in more the, sun is an energy source for our planets. And instead of chasing sun for energy, that you chase the atoms. was that really natural selection in your brain? Or you got like gravitated towards atoms and nuclear energy? Or was it always around this energy concept? And then you picked this one because it was something that you felt really, let's say, equipped.

to solve.

Yeah, I was always interested by energy, whether it's you know fossil fuels or or renewables. I actually studied electrical engineering before before studying nuclear. so I did renewable system design at one point. but essentially the idea here is to say, you know, when you look at solar, for example, or even wind, at the origin it's it's it's nuclear. Solar comes from the sun and sun is a nuclear reaction that's not happening there.

And wind is kind of a variation of that, just by the the effect of the sun on the atmosphere of of of the planet. so it's really when you go to the origin all the sources like even if you look at, you know, fossil fuels, fossil fuels has been stored. These are like plants that have died and then got concentrated.

under under the earth for a very long for millions of times and then now we're taking them back. So those plants, what they did is is essentially took the sun, the energy of the sun, and then converted into hydrocarbons and then just just concentrated for millions of years. And that's you know that's coal, that's natural gas, that's oil that we use. So in the end everything comes from the sun and the sun is is a nuclear energy device.

so if you want to kind of chase the ultimate

Correct.

source of energy, the most dense source of energy, it's nuclear.

Then before also going into the deep in that topic, what are the misconceptions about nuclear energy? Because the moment, right, we started with science and science brought us to, okay, now we can really make something with this. And then we figured out sort of like the nuclear weapons era started with all the atom bomb and so on. And then there we went, we transitioned into more like, actually, we can have power.

generate power and the moment that it's

like 1970s onwards, we said we can scale this and then we can try to establish those plans, but that things went wrong. And then like Fukushima and Chernobyl, are the like major level seven accidents that happened and gave us this scary thought about nuclear energy that perhaps that we can maybe break down those collective mirages that we had built in our minds. Like what are

the common things

Yeah.

that you come across and then you try to say these days are over we are in a different era.

Yeah, yeah. So I I I see that you reviewed the Ines scale level seven for for Fukushima. Yeah, that that that is correct.

Yeah.

Let's start by by safety. so when you look by the numbers, so you take all the energy that is generated by nuclear since it was created, so since the fifties, and then you look at all the casualties that happened, including all the accidents that you mentioned, so Chernobyl

TMI, Three Mile Island in the United States in 79, and then we have Fukushima. they the number of casualties per terawatt hour or megawatt hour generated is the lowest of any other energy source. people typically don't realize that we have accidents that are related to let's say dams for hydroelectricity, they're related to you know the process of making

sourcing and making solar pv panels or even operating solar pv panels. So when essentially you look at the life cycle of all these sources and you you look at nuclear because if it's energy density one single plant can produce a lot more energy and you look at the casualties actually very very low and you divide it by the amount of energy that is generated it's the lowest of any other source that we have today. So in terms of safety, if you look at the numbers,

These are the numbers. And then if you look at how people look at it, it's different. It's similar to how you know some people look at aviation. is aviation is the safest way of transportation, but people still, you know, look at these very very rare events that happen.

Mm.

And kind of since it's very rare, it's it's not you know part of our daily life, which is good. but people see them as catastrophic.

but in reality, when you look at the numbers, it's one of the safest, if not the safest, source of making energy. So that's the misconception number one. And there are several things that I need to precise here. So the technology that we do is what we call pressurized water reactors. So the two major accidents that you mentioned that are in a scale seven, this is Chernobyl and Fukushima, these are different types of reactors.

Pressure as well reactors were developed in the United States and they've been the safest by far compared to any other technology, any other device. They operated under land, so the submarines typically use this technology, and then also above land for commercial purposes.

Do

you want to mention what sort of reactors they were? Just to give the insights.

Yeah, so pressurized water reactors, it's essentially water that is used to cool the fuel. So the fuel, when you put all these, you know, in this uranium in a certain volume. Remember when I told you like that neutron will go and split another uranium and split another uranium. So the closest the uranium it is to each other, it will generate more energy. And so you have the fuel, the the the the water is used to remove that

So it heats up and then goes into steam generator, which is the part that we're innovating on. and that steam generator will boil water, and then the water that is boiled that steamed will be sent to turbine, turns the turbine and makes electricity. If you look at Chernobyl, Chernobyl is a reactor that use graphite for moderation. and you know, one of the biggest consequences if you have an issue with those is that that graphite

Yeah.

will heat up and will

release a lot of dust that that is radioactive, which can be problematic. So this is, you know, the Russians have abandoned that technology a long time ago. and then the other technology that the that was used in in in Japan for Fukushima was boiling water reactor, which is similar. They use water, but it's a different approach. In pressurized water reactors, you insulate everything that is close to the fuel. And so the steam

that is generated in a steam generator is completely separate and independent from the water that goes around the fuel. So it's very, very safe technology. As I said, the longest in terms of the operating experience and the worst incident that we had is in a scale file.

You already identified that too. In the worst case of the worst case scenario, then you don't exit level five.

Yeah, so

we have okay, so this is two thousand and eleven. And then after two thousand and eleven, we start seeing a shift in terms of how we design the reactors. We we're shifting now

Hmm.

to what we call passive safety. And so our in our case, our pressurized wall reactor is designed with passive safety in embedded in the design. So it's much, even much more safer.

Typically what we look at to evaluate safety is this figure called core damage frequency is the likelihood that something bad will happen to the fuel. and that figure, if you take our design approach with passive safety, and I'll explain what it is, you will reduce by two orders of a magnitude the likelihood of getting that. So you're taking the safest technology that we have and we'll make it making it even safer.

By two orders of magnitude in terms of frequency. So, what is passive safety? Passive safety means that you have the reactor, and if anything bad happens, you don't need to intervene.

Mm-hmm.

Everything will take itself, all the reactor will take itself into a safe state using just the physics of gravity. so essentially the reactor will scram, there is no efficient reaction. You still need to remove what we call decay heat.

and that is removed using gravity. so essentially if you you know if if you boil water in the kitchen, you have the hot water that will go up and then the cold water goes down. So that's essentially what's happening as we cool the reactor. And that can happen for

Mm.

many, many days without having to without needing human to intervene, without needing also any

All

active pumps, any anything that requires electricity.

for those who like are familiar with the Fukushima plant is that they lost electricity, they lost their diesel generators and they didn't have any electricity. so in our case again

Mm.

you don't need those active sources of power active pumping. So literally everything will take itself to sa a safe state without having to intervene.

Actually, it's comforting to see that you broke it down and then show that the one that we are building today is different than the ones that caused those big troubles. Because indeed, technically, yes, the incidents or the cause of that situations are almost like similar. Like there's no like a big thing. But the problem is that once that

Accident happens and the radiation starts going out. There's like a I think in Japan and Ukraine both 160k thousand people relocated in Ukraine and in Japan it was 220,000 people and then it had also like kind of generational effects with the ripple effects on all the things and then it started to develop cancer in people and so I guess this hopefully is gonna leave of clarify and then take the fear from people's mind

Thanks.

Yeah, and and

we can spend some time air bit on Fukushima. So what happened in Fukushima is one, we need to like remind people of the numbers. so the casualties happened because of the tsunami, not because of the nuclear

Okay.

accident that happened. There was radioactivity that was released. Again, with our technology, that scenario will be very, very unlikely and it's it's it's another level of safety that we're we're offering for our product. But then for f specifically the Fukushima accident.

No casualties are linked directly to the nuclear accident. And then you have a decision, it was a political decision of evacuating the land that is around the plant. Many people will disagree that was not necessary, and created a lot of fear. Actually, some people, if I remember correctly, some people died because of the rush that happened as they were trying to leave quickly.

but yeah, when we look at the numbers, again, zero casualties related to that to that accident. Of course you don't want radioactivity to be released, and and and that's very important and that's what we're working on as as as I described in our safety case.

Then if we now try to picture this in our mind, let's see the nuclear power plant as a big system and then help us understand how you do the subsystems of this big system and where exactly your solution is happening. And then we can again go layer by layer like peeling the onion now.

Yeah, so essentially

have three major systems to do the main function of a nuclear plant, which is converting uranium energy to electricity. So the first system is the fuel. that's how you engineer how the fuel is laid out within the plant. and so that's important. that's the first the first system, I would say the fuel system to make heat.

now the heat needs to be removed and converted into steam. And so that's a steam generator. That's what we're working on, and we can we can we'll spend time on it. and then when you convert heat into steam, that steam is sent to a steam turbine to not confuse with a gas turbine. a steam turbine that will be turned, that will have the s same shaft as a generator. So the generator will be turned and makes electricity.

So again, you have the fuel, you have the converter, which which is the steam generator, and then you have the steam turbine. on top of that, you have what we call auxiliary systems, safety systems. So if anything bad happens, these systems will intervene. But the main again, main functional systems is the fuel, the steam generator, and and and the steam turbine.

And then like

the rest like we need to make sure that it needs to be a closer river or a lake so that we have the water supply, right? Or

Not necessarily.

is it outdated already?

It's not necessarily. So you can run a nuclear plant without having what we call a heat sink. So typically a river or a large body of water like an ocean or or or or a sea is used to cool what we call the excess heat. But you can it it you can use what we call air condensers. You can essentially just remove the heat and

remove it to the air using what we call air condensers. So you don't need to use water. It's it's a little bit more expensive as a solution when you look at the total cost of the plant. But just to clarify,

Mm-hmm.

you don't need to have water. Actually the largest power plant in the United States by power, it's called Palo Verde, is using the wastewater of the city. I'm trying to remember which city, I think it's in

Yeah, yeah, it's in Arizona if I remember well. it uses the wastewater of the city to cool the plant. so it's also that's another situation where you don't need like clear water or the river water or the seawater. yeah.

And then today,

when we establish the plant now in our heads and then yes, we have the reactions is happening. And then we do this is like if I called nuclear reactor, the whole thing. And then in between we had a turbine and the steam engine. And then from that we generate energy and send it to the city, the people that own is in your case, this is your customers and your customers are gonna be data centers, I guess, right? let's, who are you targeting in that situation?

Yeah, I mean

the the the most demanding segment of our customers right now is data centers because we need to store data, we need to train models and so yeah the the biggest segment in terms of growth today is is data centers. So we're seeing a lot in terms of the people we're working with. but not necessarily only these. so our reactors can power residential areas, can power

university campuses, you can power industrial facilities because we make we make steam and electricity. So remember if you just keep the first system and the second system, if you

Hmm.

have the fuel and the steam generator, you just make steam. And there are many, many applications, industrial applications, but they need steam for chemical processes or pharmaceutical processes. And so we have customers that are

just using steam only in comparison to cases where you use electricity.

that's a nice way of also enlarging your customer base at the same time. But then when we zoom into the steam generator, then right now you guys are saying you found a way to make it way smaller and how is it gonna affect?

the individual's life out there? What is going to change in my life? And then what numbers are you talking about in terms of scales and effect?

Yeah, so remember when I told you the uranium is a million times more energy dense than natural gas or coal. so we have huge potential on the fuel side. And the converter that we had that we typically took in the steam generators, the the main design was invented at the end of the nineteenth century and didn't really change. if you look inside what's happening, it's a bunch of tubes.

inside of a pool of water to make it very simple. And so those tubes will heat up, heats the pool, the pool biles and makes steam and that steam is sent to the turbine. Now, if you have a lot of potential in the fuel, but you're you're able to convert one millionth of it, because that's what you took from coal and natural gas, you're not leveraging the full potential of nuclear. What we've been working on is

a steam generator that has an order of magnitude higher capability of removing heat from the same volume. And so to make it very simple, inside we've been able to shrink the size of the channels. So remember I told you it's a bunch of tubes. So when you reduce

Mm-hmm.

the size of the channels, you can make much more more channels per volume, a lot more. And so you can remove the heat

Exchange surface area will be much higher. So you can remove much more heat per unit of volume. One analogy that I typically give is think of this like transistors in a chip. So the smaller the transistor,

Mm-hmm.

the more you can put per chip, the higher the compute power. That's what you know Nvidia is solving for or or Intel. For us, the smaller the channel, the more power you can create, or more steam you can create per unit of volume.

And so now you have this device that can make a lot more power, can convert a lot more of that potential that we have with nuclear fuel compared to what we used to have. when I say a lot more, an order of magnitude, right now we're working at a factor of 20, 20x. so it's

Yeah.

a lot more in terms of efficiency of the space. And people typically ask how is space efficiency linked to cost? Because in the end, that's what matters.

For us, what we do is we change the paradigm in terms of how we make a nuclear plant. If you want to build a traditional nuclear plant, the one that are currently installed in the market, let's say you call Westinghouse, which is the main reactor manufacturer in the United States, Westinghouse will just send you components to the site. They will not send you a reactor. so you have to assemble it yourself. you have to do all the major welding, you have to do all the

testing and you have to do to build the civil structure, essentially the house of the reactor yourself. And typically that process takes a very, very long time because you have to do all these tasks at the same place and the same time.

And also you

need to do app rules, right, in between. It's not that you can just put things together. Like you need to also apply for certain permissions

Yeah, and

and app rules.

exactly. And every and every plant is different. So if you get, you know, if you don't have a standardized product, you might have to iterate a lot with the NRC, with the Nuclear Regulatory Commission here in the US.

Yes.

but where we're changing is that now that we have reduced the size of the largest component dramatically, now we can assemble the whole reactor in a factory.

in high productivity environment. So typically if we look at the numbers, productivity in manufacturing, meaning inside a factory, is five to ten x higher compared to construction site. And so if we shrink the largest component dramatically without reducing the power, we can assemble the whole reactor in a factory, test it there, have a much better quality assurance, quality control, and we'll deliver the reactor ready to run at a fixed price. So this

That's what

you deliver on the truck, literally, right?

Exactly,

exactly. And again, this is i if you look, there are several companies that make smaller reactors assembled in a truck, but the way they do it is just

Mm-hmm.

by reducing the power. It's way less power, because you have to make it small to make it fit on a truck. But if you can keep the same power level and at the same time reduce the size, now you have the power, which is typically linked to revenue,

Hmm?

and at the same time you reduce the cost because typically

First order of magnitude, size of the plant dictates the cost. So the larger it is, it becomes a mega project and it takes a long time to assemble and to build.

And we are talking about 10 plus years, right? And at the cost of 20 billion, like it's...

Yeah, so yeah,

the last the last construction project in the United States, this is the Vogel plant in Georgia. they built two units, Vogel three and four. each one of them costed roughly about fifteen to seventeen billion dollars total.

What

megawatts are we talking about?

So we're talking about a thousand thousand eleven hundred megawatt toll.

Alright, but just to-

Just to come back to the size, multiple questions, but maybe starting with the size portion. What is the trade off you guys are having there? I understand your design is inherently different and then you have a different design approach with the size and letting the steam leave the reactor. But at the same time, there must be a trade off. There is no free lunch.

There are so many trade-offs.

but I would say first, if you want to essentially locate us within the map of all the nuclear reactors that you find today in the market, I typically like to split them into two groups. you have group one, these are the legacy reactors. these are water cooled reactors that we've been building since the sixties and the seventies. This is Westinghouse, General Electric, Rolls-Royce, EDF in France. they've been building the same reactors.

with no major change making them bigger and bigger during the past seven decades. so that's that's category one. What is good about these reactors is that the supply chain is available. if you want a pump for Westinghouse reactor you have four or five suppliers they can make it for you at the lowest price and the highest quality. and they are very reliable. They run at over ninety-five percent capacity factor, particularly in the US, and over ninety-eight percent available

So they're very reliable. Actually, it's the most reliable assets that you can find today in the market, everything included. And you have group two. These are what we call typically generation four reactors. These are reactors that use different fuels, like tropical

Mm.

fuel or metallic fuel, or use different coolants like sodium, high temperature gas like helium, or you use molten salt. and these

Mm.

reactors are completely different. again, they use different components and different systems.

and these guys need to raise money essentially not just to build the product, which is a reactor, but need to raise money to build the whole supply chain around it.

Okay.

and Apollo approach is to say, wait a second, instead of re reinventing everything, the fuel, the coolant, and and trying to figure out a whole new reactor design, can we pick one single design change? Start from what we have right now, the reactors that operate very reliably.

And pick one single

Yeah.

design change that can give us the highest impact in terms of cost and in terms of the development and deployment time. And that's why we focus on the largest components. We reduce the footprint, make it way better, much safer, to to deploy the nuclear that we know today and that we know how to operate at a very high uptime.

But at the same time, I'm just trying to make sense. This is almost like a 20 % of the entire factory setup, isn't it?

like this piece or is it like how would you how would you rate it like in terms of criticality and also in terms of the setup like literally physically because the impact is huge like you lower the cost not like two times or three like you lower the cost plus you make it faster your claim is like you're gonna do this in up to 24 months whereas people are talking about 10 years of time frames

Yeah, less than twenty-four months. yes. So typically the question that we get is how changing one single component will give you that ha that massive impact. and

leverage yeah

so typically we have several layers where we can achieve cost reductions. so first layer is the component itself. So the steam generator, when you reduce its size, again an order of magnitude, now you use less materials, you have less labor hours, because the old steam generators

Typically you have to weld the tubes manually and it's very like very consuming in terms of labor hours. And the facility that you need either for testing R and D or production is massive. So we can make our steam generators in much smaller facilities. So the cost of the steam generator, the component itself, is is reduced significantly. and then since the major component is much smaller, the whole reactor can be designed to be much smaller.

Mm.

and so again, an order of magnitude. I can't give you the exact number now, but an order of magnitude also on the reactor size. and so now the reactor will use less materials, also can be assembled in a factory in a high productivity environment, as we talk talked earlier. we have a

Mm-hmm.

factor of five to ten increased productivity. And so we are able to reduce the cost of the reactor itself on top of the component. And then

Now that the reactory is much smaller, the buildings don't need to be that large. So the buildings will also be smaller. So you have less concrete, less rebar, less labor hours on the site. and so you can reduce the cost of the plant, or actually the major building in the plant, which is the

But the pressure

vessel remains the same, right? Inside the nuclear reactor, the pressure vessel is the same. I'm just shrinking the, right? But then

Yes, yes.

that is, isn't it bigger than, isn't it bigger

So yeah, if you look at

than that? Like I'm trying to just visualize it.

Yeah. If if you look at like if you picture what we call the nuclear reactor building, so if you walk into a nuclear plant, you have this massive building at the center that is typically cylindrical, but it doesn't have to be. It's typically cylindrical. It's the massive, the largest building and the most complex to to assemble. and in that building, if you look inside, the pressure vessel is really a small fraction of it. The pressure vessel is where the fuel is, in

Mm-hmm.

a traditional loop type reactor.

And then you can see this massive component is about four story tall building high, that is called the steam generator. And that's the one that we're focusing on. So if you reduce

Yeah.

the largest component in the installation, the building does not need to be that big. And so that's the third layer. That's the third layer.

It's actually really, it's always depending on

the pictures you hunt on internet, because I've never been in a nuclear energy plant, but then a couple of the pictures always looks like the pressure vessel seems to be bigger than the steam generator that we are talking. I always kind of lead a little bit of a flow, or like they are very smaller, like close to each other, but then when I find the reasoning behind all the...

innovation that you guys are discussing then hold on like indeed like you shrink it like the moment you shrink that you shrink the entire reactor and then you can pack it to the truck and then send it away

Yeah, so these are this is the third layer. So you'd use the footprint of the building. And then the fourth layer is the financing piece. So now that you're able to assemble the nuclear hardware in the factory and test it there and you deliver it to the site ready to run, you only need to do installation. So you effectively reduce the construction time to less than twenty-four months. The payments that you typically have to do, because

To build a nuclear plant, you need to raise money and typically you raise loans, debt and you raise raise equity. And as long as you're not producing power, so during 10 years, you still have to pay interest on your debt and return.

Just one question,

can you get loan from the bank for such an innovation first of its kind?

you you can, but typically what happens is that the the government has to intervene because this is highly risky, this is first of a first of a kind.

Hmm.

otherwise you'll get a massive you will get a high cost of capital, which makes the reactor project non-bankable. So at this point in in the Department of Energy is doing a great job at this. Is you want for the very first nuclear reactors where the risk is high.

to get a low cost on the capital, particularly if you're if you're more than two years. Because

Mm.

if if the longer the construction time, the more you will have to pay. And you will have to pay this in the next two, three years. So in terms of net present value, it's the highest. it's much more important than the revenue that we make 10 years from now. so it's very

Mm.

important to optimize for the construction time. And this is what we're doing, reduce construction time to less than 10 years.

So that the cost that you pay on the capital is reduced significantly. And that's the formal cost productions.

When

you also provide the reactor, but still this needs to fit into the system.

that people already have in place. So how do you simulate all the alternative scenarios so that your product is custom enough to fit in and at the same time generic enough to be able to like a piece of puzzle piece that you can always like plug in place.

Yeah.

Yeah. Yeah. So there there are several things here. So for a new plant, you you can work we can work and we already have commercial agreements and MOU signed with the leading suppliers for the nuclear industry that manufacture what we call systems for the balance of plants. So this is anything after the steam generator.

because they are very conventional. It's off the shelf. They've been taking it. This is a commercial part. This is a non-nuclear part. This is not the part that is has to be end stamped. It has to have nuclear quality assurance. This is very commercial. there are many, many suppliers that can supply it and also be certified to to make those components for you. For an existing plant, so let's say you you have a coal plant. coal plant is a similar principle. Remember, you have heat.

Boils water, makes steam, the steam goes to the turbine, turns the turbine, makes

Yeah.

electricity. You can swap the boiler of a coal plant by a nuclear reactor and then leverage the existing infrastructure. typically the steam turbine, the generator, the connection to the grid, the switch yard, all that is already there. It has to, in depending on it's case by case, depending on the plant and how how old it is. But

It can fit into an existing installation. there is already fossil fuel, whether it's coal plant or a natural gas plant. So you can decarbonize the installation while having while having to pay only for a reactor essentially.

what keeps you up at night because sounds like everything is very smooth but I'm pretty sure there are certain things like keeping you awake

yeah, they're they're yeah, yeah. Yeah,

yeah. So we talked on the technical side, what you care about is the reliability of the component. because you know when when you deal with energy infrastructure, it's not that you have something that works in Eureka and you're like great, it's working. you need to show that it works reliably for many, many years of operation. And so these are the tests that we're running at at MIT is to show that the component

In the long run, will not fail with a high confidence level, looking at several phenomena that might fail the component. We'll look at corrosion, we'll look at thermal fatigue, we'll look at fouling cred deposition, and many other phenomena that in the long run will will be problematic. This is why it it bec where it becomes harder. If you want to innovate everywhere, you have to do this on every component. but for us.

Mm.

specifically we're focusing on one component and since it's small enough we can do tests we call one to one scale. so when you go to the Nuclear Regulatory Commission, you just show them the data. You don't have to do any modeling or scaling. You have to do modeling for optimization, but really for the safety case, you just run the one to one scale on the steam generator and give them the data. So that's on the technical side.

One question,

those steam generators, they also like super, super, I forgot the name, but there's another type of reactors that is for more like scientific projects, more to understand how it's gonna happen. And then you kind of go one level up and then do properly in the real life. How are you generating data? Is it about percentage of simulations? Like based on really the data,

percentage of it is really something empirical like engineering.

Yeah. Yeah. So on in the steam generator, so you have one side, specifically for our case, you have one side that is only liquid. so it doesn't you don't have a face change, you just have only liquid. This is the hot part. and that's typically the models that we have in computational fluid dynamics, CFD. we have good models. You can model that and you don't need to do a lot of testing. typically the models are very good.

When you go to the steam generating side, this is a part where you start as liquid water and you ended up with he it with steam and in our case superheated steam, which is way better. the superheated so this the change in the phase is very, very complex. I described with

Mmm.

what we call the Navi Stokes equations. so the Navi Stokes equations don't have solution or we didn't solve them analytically. and so if someone does that will win the the Nobel Prize or the

feels metal I know before. but so far you need to run a lot, a lot of tests. And that's the best way, practically speaking, of how you can actually make a component that has enough confidence to get the approval with the regulator and also get your performance that you're looking at. So this is why it is important to run a lot of tests and we've run a lot of tests

within MIT and within our new facility that that we're building right now.

And in terms of technical readiness, guess you guys are now on level seven, if I am not mistaken, six, seven, right? And what are the next milestones moving forward so that you're gonna bring us this abundance energy?

Yeah, so in terms of TRL, technology readiness level, seven means that you have a component that is running at the operational conditions, so same temperature, pressure, and chemistry of a commercial plant, which is the case, and you need to show long-term reliability. It's not that you show that for a week there is an hour. You need to show that there is reliable for many, many years. and in our case, of course, we don't have 10 years to run the test, so we run what we call accelerated time condition testing.

which essentially to make it simple and not share more than I should, you torture the component and make sure that it survives.

I think

they're pushing it to the limits, like where it felt like.

Yes, exactly. so this is TRL seven. TRL eight is getting all the permits, the certifications that you need, whether on the supplier side, on the nuclear regulatory side, meaning the safety side. and we're working on that. and then TRL nine is commercialization.

and as art.

It's out, yeah. And it it's important to highlight that

the the technology readiness level for many, many of the group two that I mentioned earlier, the generation four reactors is still still pretty low, particularly on the reliability aspect.

Mm.

you know, being able to justify the high uptime, being able to justify that that for long periods of time you don't have to shut down every two months to solve whatever issue that you have.

Speaking of shutdowns...

If you target for data centers to be very specific with your users, they, in your customers, in a situation that they are running on like 99.999 % of the time, like it has to be up and running. My questions are twofold Number one is what is the backup of backup plan that in case something goes wrong and then how much of a time would you ask them to fix the thing? Of course it might be.

like big and small and the other one is how about the grid reliability like how do you look into the entire system together?

Yeah, so there are two ways of doing this. the second way is is more novel. The first way is you plug into the grid. you have at this moment you have to wait five to seven years to get connected to the grid, because the authorities need to make sure that they have enough additional capacity to feed your data center and you're not going to pull more from the grid, so the prices will increase for everyone. and everyone

Mm.

will essentially complain.

so the way to do it is to say to people, We have to wait until we install a new natural gas plant or a new coal plant to be able to increase the capacity so you can connect to the grid. So some of the people are saying, Okay, we don't want to wait, we don't have time, AI is is moving fast, the models are developing fast, there's competition and there is China, there's actually two to three X lower electricity cost compared to the United States. So like, okay, let's not

think about the grid, let's create our own system. let's create a gener have a generator, have a data center and then connect and connect them and make electricity. The issue there is you have an asset risk if you have only one generator. If you lose it, you don't have power, you have zero. and

Mm.

so and so the way you design this and we're looking closely with with with our customers is that you have to create enough redundancy.

And with nuclear, starting from a system there is ninety-eight percent availability, has ninety-eight percent availability factor. That's much easier to achieve. You just have, let's say, if you want a hundred megawatt electric, you take ten of our ten megawatt reactors, you plug them into your hundred megawatt, well let's say slightly lower in terms of consumption. If you lose one, it's if you lose one, we have to stop refueling. you will just

lose, you know, one reactor ten out of it's ten percent of your assets. but ninety ninety eight percent for every reactor, if you look at the whole system, ten reactors, you get you to the ninety nine point nine nine that that you're looking for.

Mm.

But if you start with a reactor that is, you know, we haven't operated before, the you will be you started typically, you know, the French tried, for example, sodium cooled reactors. it's you know, typically the in the best years it was like forty to fifty percent capacity factor. so you have to make many, many, many more reactors to make sure that you meet that reliability level if you don't want to connect to the grid.

Yeah, makes sense. at same time, are you always going to be aiming for great independent power or you will be also moving the general public as well?

Yeah, so we we have several solutions in terms of power capacity. So our first reactor is 10 megawatts, we do fifty megawatt, and we do three hundred megawatt, all can fit on a truck. and no three hundred megawatt on a truck, that no one can do that. we're the only company that can fit 300 megawatt on a truck, one single module. so three hundred megawatt, that's better for the large grid. and so you know speaking about the grid itself and trying to increase the capacity and being able to increase the capacity.

and capacity that is clean and reliable, that's our solution for that. The smaller scale, ten megawatts to fifty megawatts, that's meant for behind the meter, typically outside of the grid,

Mm.

because you want that redundancy to get the reliability level.

Okay, cool. then those are also at the beginning, I was thinking more because of the magnitude of the energy that you heard milestones and then you move, move, move. But then at the same time, indeed, then it also opens up the new channels for you to reach out to new customer base and expand it to what a billion dollars markets like to just like virtually to speak about.

Yeah. Yeah. it's it's it's a it's a huge market, you know, when you look we look at the power and the what we're seeing right now is people are willing to pay a premium on the price of electricity if you can deploy energy faster behind the meter that is reliable.

And then for them, what would be the experience? I'm your first customer. Like the moment that you reach the level nine and then you will knock on my door, hey, I'm sending it. So how is the experience gonna look like for me?

Yeah, so

You typically you start engaging with the customer way earlier. so the preparation for finding the site, getting the permit, getting the structure of who is the operator. in our case, we're starting from remember water cooled reactors, the old technology that have been running for a very long time. So we partner with existing operators that know how to run these reactors at over ninety-eight percent availability factory. so you need to the

Mm-hmm.

operator, you need to bring the financing.

structure which might include the government, might include banks for the non-equity part, and then you might have several owners. for us, for the first the first plants, we we will look at being part of the ownership. and so you need to bring that together, and then you need to have the permit, the the site permit, and then you have the technology permit. We're working on the technology permit on our side.

at the moment in the US especially, are those presidential initiatives like the encouragement that we have been seeing. And one of the key reasons was in their reasoning was like 80 % of the last 10 years nuclear energy came from China and Russia.

And then, yes, the US, we are one of the pioneers, but then yet we are not able to catch up because we had a time, the industry almost froze. And then now, do you feel that the regulations are moving much faster? And at the same time, the government is also opening you the paths normally that would be taking much longer time?

Yes, and and that's very important. for a very, very long time, we haven't built reactors in the United States. since, you know, we built two units, as I mentioned earlier, Vogel Vogel three and four in Georgia. but really that was it in the past three years. and so we lost the ability to design the plants and also to build new plants. And during that same time, China specifically was building

Gigawatts of power, and actually bought all the reactors from all over the world and in in in in and and and then now they're building their own version, their own technology specifically. the government, I think now we're we're at the stage where there are several things in for us we like to say, there are several stars that are aligned. one is that the demand is high, so we need to increase the capacity.

Mm-hmm.

The

infrastructure that we have, the fleet of reactors that were built in the 60s and the 70s, will retire at some point. So you need to replace those assets. So you need to replace those. You can get a lot of extensions, we call long-term operation LTOs, for many of these plants. So typically in the US, we're extending up to 80 years. Some of the plants are looking at 100 years of operation. but at some point you need to decommission those and build new assets. And so you have demand, which is increasing capacity.

and you have replacement of existing reactors. and the demand is increasing exponentially some people like to say but you know when you look at the price the escalation on the price of electricity is double of the inflation but you put your dollar on on average in capacity expansion you will get twice the what what the you know if you put it just on on on T bills so

the governor the government today and also the the previous administration had several incentives, whether it's tax credits, for building new capacity or increasing the capacity of existing plants. And then last year we had two executive orders that reduce the licensing time, just the approval, the permitting review time by the Nuclear Regulatory Commission

Mm-hmm.

by more than half, reduce that cost and all

what

exactly the scenario was there when you are applying for what exactly did she apply for and that's

so for us the fees that we have to pay, so you have to pay the regulator

Mm.

to review your application to get a permit. and as they spend hours on on the review reviewing process, you will have to pay them. It used to be like over three hundred dollars the hour, now it's it's it's it's fifty percent less, so about a hundred and fifty.

add it to the list

so that's one. Second executive order is on

limiting the review time or have putting a cap on the review time. It used to be you submit an application for a new nuclear plant, you have no

Forever.

idea when you get a response. and now it's limited to 12 months for an operating license and 18 months for for a design license. So that's just been very helpful in terms of given predictability. for us, when you look at our timeline, this is you know one of the drivers. getting the permit is a necessary condition to to build a plant.

But I'm just thinking now, even if you speed things up, even if you make the complete construction much easier, faster and cheaper, and at the same time, this amounts to 70 % of the entire cost, right? Like you're talking about, like holistically, the product, this construction part itself is almost like 70, 75 % of the entire thing.

So when you when you look when you break down like the cost of nuclear lifetime, so you operate and you construct, and you bring this to the present, so it's present value. the very first plants typically have a large chunk of capital cost. It's literally like over fifty percent interest on the debt and return on equity. And then

Mm.

you have

effectively about ten percent that goes into the construction of the plant. and then about ten percent related to fuel and then the remaining is operating costs.

10 % fuel about the lifetime. You're talking about lifetime.

Yeah, through

the lifetime. yeah, about ten percent. Which is good. if you look, for example, if you compare nuclear to natural gas, natural gas is typically about seventy to eighty percent cost of the fuel. So if you know something happens in the world and then the the price of natural gas increases does two X, typically your cost almost gets two X of electricity.

Mm.

So you're exposed to high volatility of the of of of your fuel. for nuclear, if uranium for whatever reason does two X

That's not a big hit to your bill of electricity. And this is why, you know, you when you look at the history in 73, France, 73, the first oil shock. France decided to have this massive program of building nuclear power plants. and actually, when you look at decarbonizing the grid, that's what they did. within 15 years, they decarbonized practically all the electricity grid long time ago. And they did it for another reason, not for decarbonizing. They did it for

security. so in case there is no oil or there is a shortage or the price increase and becomes crazy high, you

Hmm?

have nuclear and nuclear gives you that security in terms of in terms of cost of of your energy. So that's what France did specifically because they don't have oil and gas

And then to come back to that, then in the end of the day, even if you speed things up, I'm just trying to see one more risk level at the big picture, then you still need to make sure there's a steam engine working and a steam engine has the parts of plates and then there are up to seven years of timelines to get them. like, know,

the demand and the supply part of it better than I do, then how do you then still assume that this is going to be already happening? Because you might be very fast, but if one of the things on the critical path is not really moving, then this is going to slow you down too.

Yeah, yeah, I'm assuming you're talking about the steam turbine. Which, as I said earlier,

Okay.

to not confuse with a gas turbine. so it's a it's a separate

No.

product a product. so the really the ball neck today and the the the in terms of getting access to gas turbines, because they are different in terms of the materials that they use, and they operate at much higher temperatures and the lead time there is very, very long, typically five years, and I think in some cases even even even even longer.

So that's the we don't

Mm-hmm.

we don't use gas turbines in in nuclear plants. Steam turbines is is is is is is is shorter is a shorter timeline. And that's a very important point. I think I was going to mention this. One of the headaches that we deal with outside of the technical risk is being able to align all these suppliers. Remember, I our technology is to not reinvent everything, is to rely

Mm-hmm.

on an existing or reliable competitive supply chain, but still

That supply chain needs to align with our timeline. So you need to make sure that, you know, the person that makes the fuel, the person that makes the instrumentation, the person that makes the steam turbine, the make the person that makes boron 10, and many, many other things that you use in your plant can actually deliver on time

on time.

purse unit.

And then, the timelines are still, it depends on where you look in terms of the data sources, because I got to see some resources say two, three, some resources say two to five years, even for the indeed, gas turbines are, forget about them, like they are even higher ranges, but in the end of the day, minimum two or three, I could say, right? Like minimum, like everything goes super well and smooth.

Yeah,

yeah, yeah, yeah. Two is typical one of the is on the longest range. and it it's not only steam turbines. We have several things that that typically take takes two, and we're working on, you know, those suppliers to align them for for for our first unit deployment.

Tell us a bit more, like, what else is on the...

But yeah,

there there are several things typical in the clear industry. Anything related to fuel, close to fuel, is typically a long lead time. there are areas where you have only one supplier covering like ninety percent, ninety five percent of the market, so it's not

Mm.

a choice. really like small things here and there, even if it's a small quantity within the reactor, sometimes you need to plan very early. And this is what we're doing. Like this is these are the things that

war yon as we plan for for that ten first ten megawatt deployment.

Then actually this is kind of like you need to make sure everything is running in parallel, not like sequence way. Because if you wait for sequencing things, then you're going

Yes.

to just save the product and sit and wait for them to welcome you. Did I understand correctly?

Yeah,

yeah, yeah, yeah. You need to create more volume and it anything we do, it has to like tenx volume. So we need to do way better than, you know, conventional nuclear vendors. I had the chance to work with with with many of them. you you need to move mu much, much faster to be able to deploy within the timeline that we're claiming.

Awesome. And then actually then when it comes to the again the data centers, GPUs run on AC.

In a typical nuclear power plants that you go with AC most of the times, it might be a very detailed question, but do you see a way also to affect this? Or this is not so critical, actually. It doesn't matter so much in the bigger scheme of things.

Yeah, and in so for for us just by the way we make electricity, it's going to be alternative current. and typically that's, you know, what is designed for the existing for the existing grid and most of the assets. the from what I know, the interest of having DC is on transmission. so when we have like long distance transmission, it makes sense. Maybe there are specific devices that require DC. and in in that case you need an inverter in in in in the process. Really this is

For us, the value that we bring is on the generation, steam generation side. So we are a manufacturer,

Yeah.

a nuclear steam supply system. Anything else, we think that there's someone else who does it better than us. We're not going to touch it. we focus on the highest value that we can provide in terms of deploying the the core system of the nuclear power plants. but yeah, I mean

All right.

it's it's it's also up to the developer in some cases, if there is a developer that wants

DC specifically for whatever application, they can use our systems to make DC.

Yeah, correct. In the end of today, because you do your part and then as long as the others also change it or adapt to their needs, then this is still all the outside of your scope. You don't need to worry about that part.

Yeah, so we make the life

of the rest of so we interact very frequently. We and we have interfaces with with balance of plant and the rest of the installation, the construction companies. we make their life way easier. Remember, you know, the building, as I said, is much, much smaller. you don't have to do the assembly, the testing, the welding while they're doing civil construction at pouring concrete or assembling concrete concrete blocks. So we make their life easy, much, much easier.

for us, as as I said, we aim to deliver what we call the nuclear steam supply system, which is we can call it the reactor, as a fully functional unit ready to install, ready to run when it gets to the site. and so that simplifies a lot how we typically build nuclear power.

Well, still actually when we speak about it with vocabularies and know anything sounds so easy peasy and a lot of really nice flawless parts but at the same time this took you how much time

We are talking about that you dedicated into research, you dedicated into hands-on engineering projects, and then it started to materialize. If you take a look back at all the things that you have done so far, and then that made you start the company. And once you incorporated the company, things started to take off.

Yeah, yeah. I I spent a decade in the nuclear industry, started working on on design. I worked on the EDF, the French SMR reactor called Newark, and then moved to operations, worked on operations in Belgium, where I was in charge and working on fuel reloads in the Belgian fleet. So in Belgium actually nuclear is fifty percent of electricity. I think it's changed since then, but it used to be fifty percent of all electricity in the country. so it's a big share of

the electricity there. And so from operations back to design. So I moved to the United States and worked during my PhD with some of the leading reactor manufacturers, including Westinghouse, Rolls-Royce, Holtec, New Scale, looking at their designs

Mm.

and questioning every design choice in their solution. and that's where kind of that's where I started seeing that there is a path that is much more efficient in terms of

costs in terms of deployment without having to change everything in the plant, we can pick the right design choice. And that's that was literally my PhD thesis, looking at everything from, you know, material science to thermohydraulics to product finance to infrastructure deployment. what is the single design choice that can give you an efficient path

Mmm.

due to deployment and cost reduction?

so yeah, when when I get to that point, spend some time trying to to convince some of the existing players and then very very quickly you're like, yeah, there are inherent reasons why startups exist. to move fast and is to and also to really start things from zero in some cases. there are some practices within large corporate in general that that are just adverse to to innovation.

and we're yeah, I'm feeling very happy of how things are getting built today, particularly using AI. Like we we we haven't spent time on this, but like AI is a good good part of what we do every day, to streamline everything there is, you know, documents and and and and bureaucracy. but yeah, that was that was the process of how we got into Apollo and then yeah, we we had the chance to work very closely with IT and now we're we're scaling to them with a larger facility.

there is a shift that we can observe that people are trying to build more of a small modular reactors. What is the thing that you see but you think it's not gonna be easy to the others to build?

Like I don't want a secret sauce because there are multiple things when you build such an innovative product. couple of things are probably people are not seeing right now that you are obsessively seeing and trying to make it happen. What are those?

Yeah, there are two things. Like if you want to simplify decision making you know, what what is the future of nuclear, there are there are two things, size and power. the larger the larger the reactor, the more expensive it is, on an absolute basis. Like what is the cost of the plant?

Mm.

The larger it is, more costly. But typically the larger is the more power. And more power means more revenue. So you have power and cost. so you have power and size, which are linked revenue

and cost.

So can you increase the power without increasing the size? Meaning without increasing the cost. and that's what we're trying to solve for. There are several companies making smaller reactors, but also small power. as I said earlier, we can put 300 megawatt on a truck. That's a lot, a lot more power compared to any other compatible in the market. and we can achieve this by focusing on increasing the power density of the system.

So the power density of the fuel is very, very high. The rest of the system is pretty low. It's been very very low since the

Mm.

and now we're changing that. we're making the full nuclear hardware much more power dense. and solving for that means you can increase the revenue and reduce the cost as you deploy nuclear.

Actually, in that sense, that also we started to see startups popping up in the ecosystem, nuclear energy ecosystem. If you want to just break it down into companies like Waller, Atomics or Oklo and the rest, because what I see is like not all of you are really competitor competitor. Like some of them are like the full design approach is different, that they took it from a total different perspective. Some of them are still trying

do things on supply chain level Like how would you give us this kind of mapping to say this is the differentiation where we are sitting and the rest are tackling in those paths.

Yeah, yeah.

so you mentioned two companies, so Oklahoma Valar, they're in group two. So these are generation four reactors. Oklahoma is a sodium cooled reactor, Valar is a high temperature gas reactor. and and and so this group typically what has better than water cooled reactors is higher temperatures. so there is a market for high temperature applications. Typically industry, chemical industry use a lot of high temperature. So that's that's a good market for them. and then as you mentioned, there is

Section there is making the supply chain for these people because you need trizo fuel and trisofuel is not available. Now we're starting to make the first batches of that. So kind of trying to help these companies in in group two to get their supply chain together instead of having them making it for themselves. so that's related to the the design of the fuel. So trizo, metallic fuel as well.

you have companies also trying to solve the supply chain of the fuel itself, meaning the comb and the enrichment. because for for a long time, and I think it's still the case, the enrichment market was dominated by Russia. and now we're starting to see companies like General Matter, they're doing great job at kind of bringing that back to the United States.

particularly for group two, they need halo fuel. This is high assay low enriched uranium. It's uranium that requires more enrichment than than our group.

Mm-hmm.

and so you need more enrichment and you need enrichment capacity. But what I wanna emphasize is that Apollo atomics is a different category between these compared

Mm.

to these generation four reactors. We're starting from what is mature today.

and flipping one component. We're not designing completely new reactors. We're leveraging what works very well and solving the construction problem.

What do you think would break Apollo Atomics? Like what would

Yeah.

be if tomorrow Apollo Atomics doesn't exist? What would be the reason behind it?

I say bad management. not having the right people.

Nah.

This is why we're like we're very picky about the people we we get into the team. we try to be very talent dense, pick the right people. the leveraging AI is is also important. in many, many cases we just compete against ourselves. Like really we're we're doing things for the first time on many aspects and and we want to do it right.

and succeed there. so it's a the people aspect. and then yeah, if if there is there are there are elements that are outside of our control. For whatever reason the United States decide to ban all nuclear power plants. I I don't think that will happen, but there are things that are outside of our control that we can't or

Hmm, it's tough.

we can't manage. but really I think it's the team the having the right team

and and the scaling this technology, scaling to to it's it's most of what we deal with is engineering risk. Is it like, how can we how can we adjust this to solve this this limitation or to meet this performance level?

Perhaps that we can a little bit dig into how are you right now leveraging AI from A to Z.

Yeah, the thesis is very simple. I can't give a lot more detail about this, but the thesis is very simple. If you look at the cost of a nuclear plant, you break it down differently. What is the direct cost? This is the this is the cost of materials that you can touch, the cost of labor, the cost of systems that you literally can touch, and then the soft costs, or what we call the indirect cost. This is the cost of engineers writing documents or the cost of getting the approval from the regulatory

Mm.

commission.

Typically it's about forty to fifty, and like the first plants about fifty percent of the total cost is soft cost, is the cost of paperwork. And you look at any

My gosh.

other source of making electricity, you know, whether it's solar, natural gas, is way less than that. And so within nuclear,

Mm.

we have the biggest opportunity to tackle this bureaucracy problem through AI if we if it's done properly. And we have built our own engine.

here to have that capability and reduce that cost. So that's another another area we're we're very active.

Actually, that's also a major enabler, right? Not only building the product part, but on the operational side as well that all those paperwork needs to go through a human being too, in the end of the day.

Yeah. Yeah.

Yeah. The organization of the like the way we're thinking how we organize the company is completely different than of like all our vendors or how historically we've been doing. Like there are several layers that we don't need. Like the way we structure the the team is is is different.

for your releases, you managed to pull a demo on the...

demo day, and the end of your YC batch. whoever is building hardware right now, maybe could benefit from the way that you run the team, the way you guys are building and putting the cadence for yourself, also feedback loops. What sort of feedback loops did you create for yourself to make sure customers are always incorporated?

yes, I think continuous conversations with customers, also like scanning all the supply chain. This is because of our the nature of our reactor design. We rely on existing suppliers, which is a good thing, but at the same time a whisk.

Mmm.

being very, very connected to all the lead times that we have with our suppliers to make sure they can deliver on the on time. I think in terms of team and organization, the first thing you you need to do is like hire the right people. when you get the right people, in many, many cases.

You don't have to worry that much. and then yeah, it's it's startup pace, it's not corporate pace. So it's it moves much, much faster. and we try to innovate every time in terms of, is is that role really necessarily? Do we have to do this? like really questioning every every task that we do and and and focusing on, okay, this is the goal. Are there better paths than the traditional path before taking the path? so and and it's surprising.

Mm.

how things have been done for a very long time and how we can reinvent that, particularly with AI.

How could you keep yourself so fresh while you have been inside of the slowest organizations or at the same time very impactful places despite the bigger the organism gets, the slower it moves. How did you always keep yourself, okay, I'm not gonna lose this perspective on it because then you get biased at some point, right? If you stay so long in such places.

Yeah, so we're we're less than 10 people. So we're we're we're not that big already. So that's really a challenge for me. And maybe one of the things that can kill the company is the ability to grow the company, grow the team without breaking things.

Hmm.

but yeah, all I can say at the moment is like the way we've been thinking organization, having all these layers of management are not are no longer needed, or at least we try them without management layers and going straight into running experiments.

running the models directly into interaction with reality because all the rest can be that the intelligence that we have with the models today can replicate some of the mundane repetitive tasks that typically people do. And so AI enables direct connection between me, my co-founder, and really like the team and the tests and and the suppliers and and make sure that we integrate this very, very well. As things scale,

we try to, you know, be as connected, as flat as possible. I don't know if that makes sense, like a f a much flatter organization. I don't know how things will be when we get to you know a hundred employee or or more than that, but I don't think it will be the case in the next year. That's that's that's far followed.

Do you recall a moment you had a very tough time and it really hurt and you tried to resolve and what does come to your mind when I ask it? Could be an investor

I'm trying to think if

conversation, could be a team, within the team, could be within your advisory board, could be like anywhere. Something difficult really challenge you to your bones.

Yeah,

I'm trying there there are several examples. I mean raising the pre seed was not not easy, doing it, you know, without the Y C ecosystem or the Y C advice and so I had to learn everything firsthand, which was good. I suffered but it was good.

But even what

sort of suffering happened there?

you know,

taking so so many calls and and realizing that in some cases just investors will, you know, ask you many, many questions and in the end, you're you're in an an endless process without getting really like a closure. there there are cases there you know, in some cases I got lucky where where we had people trusting me and funding the company. I think one of the hardest moments are the very first moments. You know, if you decide to start a company in hard tech in general, I think this is valuable for

For many people thinking of starting hard tech companies, it's it's much, much harder than saying, hey, I'm gonna start a software company, I'm gonna try, you know, you know, build a website and see what happens and get you can you can build a prototype very quickly if you're doing software.

Mm.

especially with AI. Like literally you can prompt what you want and you get a website and a and a service. For hard tech is much higher, especially a year, year and a half ago where AI was not as it is right now.

Good.

Starting a heart tech company is much harder. But I didn't literally, honestly, I didn't believe in any of the all other alternatives. and so like someone has to do this, what I'm thinking of. and and and it was me. so I started as a solo founder, and then had the chance to to meet Drew as my co-founder. but I think those are the the kind of the hardest moments. I'm sure we'll have many, many and you know, we have challenges every day, but those are much, much harder. When you start and your ambition is to build, you know.

company that makes products that are in, you know, tens of millions, hundreds of millions in terms of cost, how can you build that company? How can you scale that company from a small prototype into something that will sell for a lot, you know, billions of dollars in the future?

a feeling you really really wanted to get into YC not just for the accelerator experience because you have been to another accelerator before that when did you get in on your second application or you applied all the multiple times?

I'm trying to remember it was like second or third. It was not a first time. Yeah.

What was it I have a feeling you believed in something strongly about the YC ecosystem that's why you pursued that.

Yes,

and I didn't realize the full advantage until I got in. I will pray

I didn't

even know, right? It was even like I thought this, it became like that.

Yes, yes, yes,

yes, yes. Yeah, a lot of volume. Anyone who's considering applying to YC, I think it's I think it's great. It's it's it's worth it. Unless like you have like a r business that are already, you know, functional and and and you know, you're making revenue, you're making massive revenue. maybe it doesn't make sense. but really I think most of the cases YC makes a lot of sense. on the network side, on the just the exposure and the lessons learned that will save you billions of dollars, especially in our case.

yeah, I think I think it was like second or third time. applied to YC. because I wanted the reason that I had in mind is like I wanted exposure to West Coast. so typically today, and most most of the BC money is on on the West Coast. so

Mm.

I wanted exposure to West Coast, knowing that we're we're building here in Boston. but yeah, got accepted. we got an offer, and then I had many people telling me, it's hard tech, you YC is like war.

Software SaaS and all that. and I remember Tom Bloomfield, our partner at YC, connected me with Philip from Star Cloud.

and Philip told me like I asked him like, Hey, is this is this your first time? Is this your first company? He's like, No, no, this is my second company. And he told me like, even if it's my fourth company, I will up still apply to YC. I was like, Okay. I think

I'm sold.

I'll

do that. then he he gave me other reasons, like more tangible reasons. YC has per year more hard tech companies than any other VCs. So literally the network of companies is broad within even hard tech, even if the hard tech is a small portion of each batch.

I just, I think during Sam Altman's time, he started more expanding to hardware startups. And afterwards it started to get even bigger and bigger. It's not just a work of last two, three years or right after the advent of AI, right? probably you also observed that seeds rooted from years ago.

Yeah, there there's there are a few companies doing nuclear. Okla is one of them, went to Y C. so it's not it dates since Sam Altman was was the president of YC. It was he's the guy who started like hard tech within Y Combinator. but it was always a small portion, but always had very good companies. companies that we you know we can call and and get advice from. 'cause running a hard tech company is different in many respects.

to running a software company. And many of the advic to a hardware company.

But it's one thing you learned from Tom Blomfield that you would always remember for the rest of your life. You started to smile.

fun about Tom Bloomfield.

He's he okay.

He's very good with hard tech companies. I don't know if this is fun. but and he has this character where

Always it feels always like he's expecting more from you, even if it's like, this is great. but you you feel that you can do better. and in some cases he tells you like you can do better. In some cases, it's just kind of implied. The way he talks, the way he like just convey the information to you, you can certainly do better. and and and sometimes plants just succeed. It was like I I was thinking about the very first weeks of YC, and we were just aiming having more and more LOIs, and we got more and more.

And then Tom and Cool Ver, our two partners, they were like, Have you thought about getting contract? which is way, way harder

laughs

for a nuclear company. And so that planted the seed in our mind was like we'll go in and looking for contract directly, literally at the pre-seed stage. so yeah, I think one one thing about him is just like kind of feeling that you can always do better. you can enjoy

small successes or even big successes, but can always do better and he's good at making you feel that way.

Is there any personal kind of it? Because this was more like startup business related. Was there anything that also helped you as a human being to see the world differently?

Yeah, w in what context? With with Tom?

like

if it's Tom or within YC interactions within the batch like the inside

Yeah.

of that ecosystem.

Yeah, I I think YC like you y you realize like you're you see people working very, very hard. Like you see people really working twenty four seven, having you know, dozens of calls every every week, working very, very hard and and that's enjoyable. it's an atmosphere like, hey, it can always it can always do better. But they tell you like, in Paul Graham,

Like was asked once, where is the job of the CEO? Should the CEO be, you know, the hardest working person in the company? It was like, no, no, no, not necessarily. The CEO the CEO's job is to survive 10 years, 15 years. It's a marathon. they try to put you in a position where you can work very, very hard, but at the same time, you're able to survive working very, very hard for a long time.

And I think that's very important. And and you see you see advice around that. you see real concrete examples, real life examples of, people going through that. and that was very very very valuable.

I think also that these are the people who has done this, have walked the path. It's not that they talk to you out of theory, probably that you feel it in the conversations, in the smallest details, in the nuances very clearly.

Yeah.

Yeah. yeah, that's one of the major values of YC. You know, they have the community but also just the sheer amount of experience and interaction with their investors or just specific stories that happened to founders, where you're like, Okay, yeah, this can save you save me millions of dollars in the future.

Actually, I just remembered during the demo day that we want to tell that story that everyone goes, my God, we have a nuclear reactor in the parking lot.

I'm not

I'm not sure how much I can share about that. so we had one session where Paul Graham was in and we were talking about the history of YC. it was a whole batch. and Paul Graham and and Jessica Levinson, so the two founders of W Y Combinator. and we're we're talking j it was a fireside chat. and at some point they see smoke outside. So

We have our demo in the YC parking lot. So at some point we see smoke from the outside. and and I believe Jessica just made a comment. It's like, what's going on? And then just ignored it. And then five minutes later, there's more smoke. And someone just said, I don't know who was exactly, maybe Gary, it was like, we have a nuclear reactor in the parking lot. And so people started freaking out, who's has this? Who brought this? And so I had to stand up and explain, like, we don't have uranium. This is not operating, and this is not us. but

That was there was a that was a funny moment where I had to had to explain how nuclear is actually safe.

And it turned out to be one of the electric cables, right? Around

It was it was yeah, it was an art.

the corner that he tweeted about that. It was weird.

Yeah, yeah. It was

linked to an electric line outage issue that happened, maybe one transformer. I d I don't know exactly what was the issue, but yeah, it was not it was not us.

I kind of like the I keep digging into the questions. Is there anything that you would have loved to answer but I didn't ask?

Yeah, I think I think we covered we covered many, many topics. We covered Y C, the technology, nuclear in general. Yeah.

Perhaps, Based on the poor experience that you accumulated with some of the investors, like what would be the one thing you would request them so that the others,

Tip tech founders don't suffer from that. Like what sort of things that you observed and what is the best investor profile and what is the really poor investor profile. Don't miss your time. Just move on to the next.

You have to apply for Y C. they have a very good tool for that. That's all

The internal,

kind of like the blacklisted ones don't talk to them at all.

Yeah, it's it's more than that. I mean it's it we don't want to make investors look bad. I mean, you know, we have the chance to have

No.

many of them being supportive. but yeah, in some cases you you land into those situations, maybe for good reason. Some people just want more information to get conviction. It's just that there is the YC way of doing things, and it's way, way better for founders than, you know, conventional traditional fundraising that might take a very long time.

I have a question in my mind, it's more like a personal one. In terms of Kardashev scale, when do you see us making the jump?

Yes, yeah. So if I remember well scale is leveraging resources within the planet, leveraging the energy of the sun, and then leveraging the energy of the stars and galaxy in total. I think if we can create the sun or something that's similar to the sun, we we already jumped that. okay, so there are two things. You need to create it and you need to control it.

And with nuclear, we have this source that is a million times more energy dense. We still are not able to control it at a high conversion rate. So we know how to make nuclear energy, but still are not able to extract a lot more energy from that massive source. so there's this big debate. typically a lot of people pushing for renewables. I think there's a good reason why. for a very long time we use renewables, even before the inventing electricity.

People used to transport using wind. so it was for transportation. We used, you know, biomass is essentially solar energy that is cap captured by plants. So a the the major debate here is between concentrated sources and diffused sources. So typically solar and wind, these are diffused sources, and concentrated sources this is you know fossil fuels, and to a higher extent this is uranium. I think the civilization had created

Clear jump in terms of our capability of increasing the energy per capita when we were able to leverage these concentrated sources, which are coal and natural gas and oil. And then the next step is to leverage uranium, which is much, much higher. We still we did it to

Mm.

a smaller extent, but the potential there, the S-curve, we're at just at the beginning of it.

it seems dangerous for some people. You know, the first person who invented the fire, I'm not sure that he was like greed by everyone and it w it was as a hey, this is dangerous. and then when people mastered that, it became really civilization

Mm.

change. So for us, I think and our goal is to be able to leverage this million times higher energy density and extract it at a much higher rate than what we do with with coal and natural gas. So

That's the first layer. This is energy. And then typically the way I like to describe civilization is in three layers. you have energy, and we're working on it. you have technology, which is making tools. Steve Jobs liked talking about humans as tool makers, you know, making better tools, better devices. And then the third layer is is intelligence. I seen in the first jump of intelligence is inventing language. intelligence

I intelligence.

is actually ability.

thinking about complex topics just by using abstraction. And so inventing language was one of them. I think the second jump was inventing the scientific method and math to talk specifically about just the physics without the feelings or without the emotions a human has. So it's really being very, very rational and essentially developing science and building tools.

that are very very good using using our understanding of science. And I think this next jump is LLMs, is you know being able to create another layer of abstraction that enhance our capability of you know abstracting and building models and thinking about stuff through artificial intelligence. So these are three layers. we're working on the energy. the more energy you have the more technology you can build, the more intelligence you can run.

so that's our part.

How much of a jump would you bring with your solution? Like 20 years later, let's say, and things work according to your plan. 20 is quite a long time though. But how do you see the contribution of your work?

Yeah, I think being able to create energy abundance to literally move we had a massive shift when we moved from essentially muscles. These are our first tools that we had, you know, humans before combustion engines, they had muscles

you

or animal muscles. If you want to move somewhere for transportation, use an animal. If you wanted like to work the land, use an animal. So it's really muscles. and then we move to a device that is much more power dense, which are the combustion engines.

whether it's you know natural gas or or oil or gas or or or something of that type. And that enabled us already to have a massive jump in terms of power density. We have created the industrial revolution, we have powered all the digital revolution, you know, our computers and et cetera. We're creating another jump in terms of power density of devices that we can build. And those new devices, the ones that the ones that Apollo is building, will be creating

much more energy abundance. They will be able to live in different on different planets. They will be able to power a lot more intelligence than we do today. and essentially eliminate poverty or the w what what is conceptualized as poverty is way, way different and actually better than what it used to be thought of as poverty.