💡The Heretics Who Made Fusion Small: What Avalanche Energy Teaches Us About Impossible Problems

Please see the corresponding Substack episode

When the orthodoxy says “it can’t be done,” someone inevitably proves them wrong—but only if they’re willing to fail fast, think sideways, and trust the data over doctrine.

There’s a peculiar comfort in impossibility. When experts across an entire field agree that something fundamentally cannot work, we get to stop worrying about it. The case is closed. The limits are real. We can move on to problems that might actually have solutions.

This is precisely the psychological space that electrostatic fusion occupied for decades in the minds of plasma physicists. Not controversial, not debatable—simply impossible. The math was clear: space charge limits would prevent sufficient density, and Coulomb collisions would drain energy 25 times faster than fusion could release it. Building a small fusion reactor using electrostatic confinement wasn’t just hard; it violated basic principles.

Electrostatic Fusion Orbitron

The Company Reinventing Fusion

Avalanche Energy

This is Heliox: Where Evidence Meets Empathy

Independent, moderated, timely, deep, gentle, clinical, global, and community conversations about things that matter.  Breathe Easy, we go deep and lightly surface the big ideas.

Thanks for listening today!

Four recurring narratives underlie every episode: boundary dissolution, adaptive complexity, embodied knowledge, and quantum-like uncertainty. These aren’t just philosophical musings but frameworks for understanding our modern world. 

We hope you continue exploring our other podcasts, responding to the content, and checking out our related articles on the Heliox Podcast on Substack. 

Transcript

Speaker 1: 0:25

When you think about fusion power, what usually comes to mind, I bet it's these enormous facilities, right? Like buildings the size of warehouses, costing billions, timelines stretching out forever. That scale, it's sort of been the defining limit.

Speaker 2: 0:38

It really has.

Speaker 1: 0:39

Yeah.

Speaker 2: 0:39

I mean, when it takes five years and maybe half a billion dollars just to build the next version, you just can't iterate quickly. You can't experiment fast enough to solve the really hard science problems.

Speaker 1: 0:50

Exactly. So today, we're taking a completely different path, a hard left turn. You could say we're diving deep into a company called avalanche energy. They have this well profoundly contrarian idea with their compact fusion concept the orbitron mm-hmm our mission here is to really unpack their source material. Look at the tech the engineering challenges they face the data science behind it all and show you how they're aiming to build what might be the world's smallest fusion reactor.

Speaker 2: 1:19

When we say small, we mean really small. The actual plasma core. You could hold it in your hands. The whole integrated reactor system is designed to fit in the bed of a pickup truck.

Speaker 1: 1:28

That's just wild. And that small size completely changes the game, doesn't it? They're not talking about replacing massive coal plants one for one.

Speaker 2: 1:35

No, not at all. Their focus is totally different. Think mass-produced units. mobile power, distributed energy, you know, micro grids for remote places like the Arctic, maybe long haul trucking, potentially even aviation down the line.

Speaker 1: 1:50

And the timeline they're talking about aiming for commercial operation in under six years. I mean, that's incredibly ambitious, almost audacious, really. which means we absolutely have to look closely at the science.

Speaker 2: 2:01

Agreed. That kind of claim demands scrutiny.

Speaker 1: 2:04

So let's get into it. The orbitron. They call it an electrostatic and magnetic hybrid. What's the core idea here? How does this architecture let them shrink fusion down so much.

Speaker 2: 2:14

Okay, so the orbitron cleverly combines two existing ideas, trying to capture the best parts of both. The first piece is inspired by something called the orbit trap. That part is purely electrostatic confinement.

Speaker 1: 2:25

Right, electrostatic. So trapping particles with electric fields, not magnets. What does that mean for the ions?

Speaker 2: 2:32

Exactly. You have a central rod, the cathode, that's highly negatively charged, hundreds of kilovolts negative. The positive ions like deuterium or tritium are strongly attracted to it. But instead of hitting it, they get trapped in these stable, high energy orbits around it. Think of it like a satellite orbiting Earth, but voltage is providing the pull instead of gravity. No magnetic field needed for the ions themselves.

Speaker 1: 2:57

Okay, makes sense. But you said hybrid. So where do the magnets come in? That must be the second part, the magnetron aspect.

Speaker 2: 3:02

That's right. The magnets are there to handle the electrons. The Megatron concept uses a relatively weak magnetic field, maybe 0.6 Tesla, which is pretty low for fusion running along the axis of the device. It works together with the electric field. It's what physicists call an E cross B configuration.

Speaker 1: 3:19

E cross B. Let's just quickly break that down for folks. What does that do to the electrons?

Speaker 2: 3:23

It basically means the electric field E points radially inward, and the magnetic field B points axially, along the machine. These perpendicular fields force the electrons into a very specific kind of motion. They drift around the central cathode, co-rotating with the orbiting ions. It's less about simple trapping and more about forcing them into this controlled spinning cloud.

Speaker 1: 3:47

So the real innovation is getting both things happening together. Ions orbiting due to voltage, electrons spinning due to E cross B, all crammed into this tiny core just tens of centimeters wide.

Speaker 2: 3:58

Precisely. Co-confinement. And having those electrons spinning alongside the ions is absolutely critical. It helps them tackle two major historical roadblocks for purely electrostatic fusion ideas.

Speaker 1: 4:08

Okay, let's unpack those roadblocks. What's the first crucial role the electrons play? You mentioned something about space charge.

Speaker 2: 4:15

Yeah, space charge. So imagine you're only trying to trap positive ions using voltage. As you pack more and more positive ions in, Their combined positive charge starts to repel any new positive ions trying to get in. The electric field gets shielded, you hit a density limit, the trap essentially fills up.

Speaker 1: 4:31

Ah, okay. So the plasma itself fights against becoming denser. How do the electrons fix that?

Speaker 2: 4:37

By adding negatively charged electrons into the mix, you cancel out that positive charge buildup. The plasma becomes quasi-neutral, roughly equal positive and negative charges. This allows them to push the ion density much, much higher than you could in a purely electrostatic system, way past that traditional space charge limit.

Speaker 1: 4:57

Got it. That clears one hurdle. And the second function, you mentioned it's more about energy, something like giving the ions a tailwind.

Speaker 2: 5:03

Right. It's a bit more subtle. Since the electrons are also hot and spinning in the same direction as the ions, the idea is they help maintain the ions' energy. They provide this sort of rotational support, a tailwind effect. This is meant to help the ions burn longer, meaning stay at high enough energy for fusion to happen instead of quickly losing energy through collisions and just scattering off each other.

Speaker 1: 5:25

Which brings us neatly to the controversies. Because, let's face it, electrostatic fusion, especially non-thermal approaches like this... They've often been seen as, well, maybe heretical by mainstream plasma physics, largely because of the very problems you just described.

Speaker 2: 5:41

Oh, absolutely. They knew going in that the Orbitron design flies in the face of two major longstanding criticisms.

Speaker 1: 5:49

So let's dig into how they confronted those head on. Critique number one, the space charge limit we just discussed. How did they prove their quasi-neutral approach could overcome it?

Speaker 2: 5:59

This is where computation comes in.

Speaker 1: 6:00

Yeah.

Speaker 2: 6:01

Traditional, simpler models suggest you just couldn't get the density needed for fusion this way. So Avalanche invested heavily in these really high-fidelity computer simulations called particle in cell, or PIC simulations. They track millions of individual ions and electrons interacting.

Speaker 1: 6:16

And what did those PIC simulations show?

Speaker 2: 6:18

They showed pretty definitively that in this specific hybrid orbitron geometry, the old space charge limit assumptions didn't hold up. When you load ions and electrons together carefully, the way they interact and couple allows the density to climb dramatically higher than predicted. It validated their quasi-neutral concept. So that was critique A tackled, at least in simulation.

Speaker 1: 6:42

Okay, one down. But critique B Coulomb collisions. This one feels like the bigger killer for non-thermal fusion concepts.

Speaker 2: 6:49

It's arguably the existential challenge. Yeah. The basic argument is really simple and powerful. In any plasma hot enough for fusion, particles are constantly bumping into each other. That's Coulomb scattering. The problem is for fuel like deuterium tritium, DT, the rate at which particles scatter off each other and lose energy is way, way faster than the rate at which they actually fuse.

Speaker 1: 7:10

like 25 times faster. 25 times faster. So particles lose their energy 25 times before they even get a chance to fuse. That sounds impossible to overcome for net energy.

Speaker 2: 7:19

That's the standard critique, yes. If particles are constantly cooling down, how can you ever get more energy out than you put in, especially in a small device without the sheer scale of a tokamak to compensate?

Speaker 1: 7:31

So how does the Orbitron design possibly get around those 25 to 1 odds?

Speaker 2: 7:36

This is where the hybrid nature is key again and having independent control. They ran their own collision transport models tailored specifically to the Orbitron's conditions. They realized they have a unique knob to turn. They can set the ion energy using the high voltage largely independently from setting the electron temperature using the magnetic field strength.

Speaker 1: 7:56

Ah, decoupling those two variables. That's not something most fusion concepts can easily do. Did the models show a specific operating point, a kind of sweet spot?

Speaker 2: 8:05

They did. The models pointed towards this potentially very interesting low-loss regime if they could keep the electron temperature controlled within a specific narrow band, around 10-20,000 electron volts.

Speaker 1: 8:17

Okay, 10 to 20 kV for the electrons. What's special about that range?

Speaker 2: 8:21

In that specific window, something crucial happens. The rate at which the high energy ions lose energy by scattering off the heavier fuel ions, like tritium, gets balanced out. It's balanced by the rate at which they gain energy from scattering off the lighter hot electrons that are acting like that tailwind we met.

Speaker 1: 8:37

Whoa, okay. So instead of just fighting the energy loss from scattering, they're trying to use the electron scattering to pump energy back into the ions, keeping them hot.

Speaker 2: 8:45

Exactly. It's like setting up this dynamic equilibrium, a constant energy balancing act. If they can maintain that specific electron temperature, the net energy loss from the ions due to collisions could be dramatically reduced, far lower than what traditional theory would predict for a non-thermal plasma. It suggests a pathway to efficiency.

Speaker 1: 9:05

So the Fisk simulations look really promising. They had potential answers to the big critiques, but then... reality hits. The engineering bench, let's talk about that series A milestone crisis back in 2024. They hit their voltage goals, right? 300 kilovolts, beam injection worked, but plasma density, they hit a brick wall. Yeah, that was a really tough period.

Speaker 2: 9:28

They achieved the high voltage, proved they could inject the ion beams into the core, But the plasma just wouldn't cooperate. They were aiming for a density of 10 to the 11 particles per cubic centimeter, and they just couldn't get there. The core was fundamentally unstable.

Speaker 1: 9:43

And the whole project basically stalled because of this.

Speaker 2: 9:46

Pretty much. You can have high voltage, you can have injection, but if you can't hold a dense enough plasma stable for long enough even to measure it properly, you're stuck.

Speaker 1: 9:55

They called the problem a rotating mode instability. Can you give us a picture of what that actually looked like inside the reactor core? Right. An analogy maybe?

Speaker 2: 10:04

The analogy they use, which is pretty good, is an unbalanced washing machine. Imagine all the wet clothes clump up on one side of the drum during the spin cycle. What happens?

Speaker 1: 10:13

The whole machine starts shaking violently around. It's unstable.

Speaker 2: 10:17

Exactly. Inside the Orbitron core, the plasma wasn't staying uniform. It would clump together into this dense region on one side. This clump would then rotate really rapidly like the wet clothes, and that rotation would basically just sling the high-energy particles right out of the trap and into the walls. It acted like a super-efficient vacuum cleaner, constantly sucking the plasma out before it could get dense. Devastating for their goals.

Speaker 1: 10:41

Wow, so they're stuck. billions potentially on the line facing a fundamental instability. That sounds like immense pressure. But it also led to this really fascinating breakthrough moment, a cross-disciplinary insight in April 2025. It really did.

Speaker 2: 10:58

It sounds like one of those classic necessity is the mother of invention moments, born out of sheer desperation. The CEO was apparently thinking hard about this rotational problem, and it reminded him of fluid dynamics, specifically his Ph.D. work on turbulence modeling.

Speaker 1: 11:11

Turbulence, like in air or water flow, how does that connect to plasma?

Speaker 2: 11:16

Do you remember this phenomenon called relaminarization, where under certain conditions, applying rotational shear, meaning different layers rotating at different speeds, can actually break up turbulence and make the flow smooth again? He saw a parallel. Maybe strong shear could break up the rotating plasma clump.

Speaker 1: 11:34

Okay, interesting idea. But how do you apply shear to a plasma?

Speaker 2: 11:39

That's where another team member, a plasma physicist, made the connection. They were called specific Russian fusion experiments from back in the 80s and 90s, particularly a device called PSP-2. Those researchers had experimented with using a pulsed voltage on the central cathode.

Speaker 1: 11:55

Pulsed voltage? How would that help?

Speaker 2: 11:57

The idea was that pulsing the voltage would rapidly spin up the plasma layers near the cathode, creating exactly that kind of strong rotational shear different layers, spinning at very different speeds. And those Russian experiments had apparently shown some success in using this shear to break up similar rotational modes.

Speaker 1: 12:13

Incredible. So they connected fluid dynamics, old Soviet fusion research, and decided to try pulsing their cathode to introduce shear and rip apart the instability.

Speaker 2: 12:22

That was the plan. They even gave this new experimental approach a codename.

Speaker 1: 12:27

Jin.

Speaker 2: 12:28

Jin. As in Jin Erso from Star Wars, the one who stole the Death Star plans.

Speaker 1: 12:33

Apparently so.

Speaker 2: 12:34

Yeah.

Speaker 1: 12:34

Which probably tells you something about how risky or maybe even rebellious they felt this whole pivot was.

Speaker 2: 12:40

It definitely paints a picture of their mindset and it sounds like they didn't waste any time trying it out. They talk about engineering resilience.

Speaker 1: 12:47

Right. Their philosophy wasn't about cautious slow steps. It was rapid iteration. If an experiment using the Jin pulsing failed on Monday. They'd literally be rebuilding parts of the reactor core over the next few days, maybe swapping components on Friday, ready for another shot the following Monday. Intense cycles.

Speaker 2: 13:05

And did this high-risk, fast iteration approach pay off? What happened when they ran the defining GIN experiment with the pulsing technique?

Speaker 1: 13:12

The results were frankly spectacular. The instability was suppressed. And the plasma density they achieved. It shot up to four times 10^12 particles per cubic centimeter.

Speaker 2: 13:22

4 times 10 to the 12. Wait, their original milestone was 10 to the 11. Exactly. They didn't just meet the milestone they had been stuck on, they blew past it by a factor of 40.

Speaker 1: 13:31

40 times the target density, that's a game changer.

Speaker 2: 13:34

Absolutely. In their view, that single experiment, born from that cross-disciplinary insight in rapid engineering, fundamentally changed the nature of the project. It moved the Orbitron from being primarily a science risk, questioning can this even work, to what they now consider mostly an engineering project, figuring out how to scale it and make it reliable.

Speaker 1: 13:55

That's an amazing turnaround story. Okay, so the core clasmophysics seems to be on much firmer ground now. But there's another crucial piece we need to touch on. None of this works without reliably handling extremely high voltages in a compact device. hundreds of kilovolts. That's a huge engineering challenge in itself, right? Especially with large surface area electrodes in vacuum.

Speaker 2: 14:17

It's absolutely critical. It's a foundational requirement. When you're dealing with high voltage across broad electrode surfaces in a vacuum, a major limiting factor is something called field emission, or dark current. It's basically unwanted electrons leaking off the negative electrode surface. This leakage current can cause all sorts of problems. It loads down the power supply, can trigger arcs, and ultimately limits how much voltage you can reliably hold.

Speaker 1: 14:43

And engineers have had ways to estimate this for decades, right? Using models like the Fowler Nordheim equation from the 1920s or Murphy Good from the 50s, Why weren't those good enough here?

Speaker 2: 14:55

Because those classic models were really developed for and work well for. Emission from a single idealized microscopic point like the tip of a sharp needle. But on a real world broad area electrode you don't have one emitter. You have potentially hundreds or thousands of microscopic bumps and irregularities across the surface all emitting electrons.

Speaker 1: 15:13

Ah, so the models require knowing the exact shape and location of every tiny bump?

Speaker 2: 15:17

Essentially, yes. They need parameters like the exact tip radius and the field enhancement factor for each emission site. And trying to measure that across a large surface is practically impossible. So those old equations become non-predictive for real systems. Engineers could basically only use them after a system failed to try and figure out why it failed by fitting the data. Not very helpful for design.

Speaker 1: 15:41

Okay, so trying to model the exact geometry was a dead end. Avalanche took a different approach, moving from physics-based geometry to data-driven statistics.

Speaker 2: 15:49

Exactly. Instead of trying to measure every microscopic bump, they decided to train a machine learning model to predict the total dark current based on statistical properties of the system. They built this massive comprehensive data set. It included over 259 hours of actual current voltage measurements, from their test systems.

Speaker 1: 16:07

- Wow, 259 hours of data? What else went into training the model besides just current and voltage?

Speaker 2: 16:12

- They supplemented it with four other key streams of statistical data. One, results from electrostatic simulations, giving them statistical features of the electric field distribution across the surface, specifically where it was high, above 5 megabolts per meter. 2. Data from optical microscopy, providing statistical measures of the surface roughness like the typical height and density of those microscopic projections. 3. The material's work function, which is a known property affecting emission. And 4. The total surface area of the cathode.

Speaker 1: 16:43

Okay, so a really rich dataset combining actual measurements with statistical info about the field and the surface. They fed all this into various ML algorithms. How did that turn out?

Speaker 2: 16:54

They tested a range of models, from simple linear regression up to neural networks. But the clear winners were ensemble methods.

Speaker 1: 17:01

Ensemble methods, like combining multiple models.

Speaker 2: 17:03

Yeah, techniques like random forests or gradient boosting. The best performer by far was a gradient boosting regression model, or GBR. They measured its performance using the coefficient of determination, R squared.

Speaker 1: 17:14

R squared, right. That measures how well the model's predictions match the actual data. What score did the GBR model achieve?

Speaker 2: 17:21

It achieved an R squared value of 98.2%.

Speaker 1: 17:25

98.2%. That's incredibly high. Almost perfect prediction, statistically speaking. What does that actually mean for their engineers day to day?

Speaker 2: 17:34

It means they've transformed high voltage design from being largely empirical guesswork into something much more predictive and scientific. Now they can design a new high voltage component, run the design parameters through their GBR model, and get a highly accurate prediction of what its dark current performance will be before they even build or test it. They can predict failure points much earlier in the design cycle. It's become a vital tool for accelerating the development of the Orbitron's high-voltage systems.

Speaker 1: 18:03

That's fantastic. So let's try and synthesize this. We've covered a lot of ground. Two major achievements stand out. First, this data-driven ML model that gives them predictive power over high-voltage field emission, a critical enabling technology.

Speaker 2: 18:15

Right. Moving from curve fitting after failure to predictive design before building.

Speaker 1: 18:20

And second, the massive breakthrough in plasma density using that GIN pulsing technique

Speaker 2: 18:29

Yeah, demonstrating that their compact electrostatic magnetic hybrid can achieve fusion-relevant densities, directly addressing long-standing criticisms.

Speaker 1: 18:38

So the big-picture takeaway seems to be that the fundamental, can this possibly work? Science questions have largely been answered, or at least strong path forward have been demonstrated.

Speaker 2: 18:49

That's certainly how they frame it. They see the primary risk shifting from fundamental science to engineering execution. It's now about scaling up higher voltages, stronger magnetic fields, integrating all the systems, and pushing towards Q greater than one net energy gain using deuterium-tritium fuel.

Speaker 1: 19:05

All aiming for that disruptive vision. Mass-produced, mobile, distributed fusion energy.

Speaker 2: 19:11

And it's interesting how they embrace that fusion heretic label. They consciously challenged the prevailing assumptions that fusion must be huge, complex, and decades away, and that electrostatic approaches were inherently flawed.

Speaker 1: 19:23

Which really brings us to a final thought for you, the listener. Here we have a team that seems to have unlocked a potentially viable, radically different path in a field dominated by massive decades-long projects. They did it by questioning core assumptions, applying engineering resilience, using novel data science methods. So it makes you wonder.

Speaker 2: 19:43

Yeah. What revolutionary ideas or technologies maybe in your field are currently dismissed as impossible or heretical? What might you be able to achieve if you stopped accepting the fundamental limits as absolute gospel and instead rigorously challenged them with data and creative engineering? Food for thought.