The Unicorn Tears Problem

How a quantum computer helped crack fusion's tritium crisis — and a 110-kilocalorie mistake revealed why nature refuses to be chopped into pieces

What we don't love talking about is that the "clean unlimited energy" of the future currently runs on a fuel so scarce it might as well be harvested from unicorns.

So the reactor has to make its own. 

A classical supercomputer, playing head chef, handled the bulk of the simulation: hundreds of atoms, the overall shape of the chaos. But for the one delicate reduction sauce that required molecular perfection — the razor-thin quantum correlations between fluorine's squishy electron cloud and a newly formed tritium ion — they handed the dish to a sous chef: an IBM quantum processor, taking a million measurements a second and letting nature's own randomness find the answer, the way rain finds the lowest point in a mountain range without anyone walking it.

And it worked. Spectacularly. Across nine different molecular clusters

The error lived somewhere else entirely: in the act of chopping the problem up in the first place. To make the math small enough to compute at all, the researchers had drawn an artificial boundary around each fragment — a line that said, essentially, we will only pay attention to the electrons close enough to matter, and ignore everything past it. But in a fluid this interconnected, nothing is really past it. 

They pinpointed, with real precision, exactly where the blind spot lived — in the fragmentation, not the hardware — and they're now teaching AI systems to draw smarter boundaries, ones that preserve the whisper of connection between distant atoms instead of severing it. The hardware, they now know, is ready. The intelligence being built to wield it more wisely is catching up.

There's something almost tender in that. A thousand-degree radioactive salt bath teaching us, gently and at enormous expense, that everything is more entangled than our tools were built to admit — and that the next real leap forward might not be a bigger machine, but a better way of paying attention to the whole.

Quantum Computations on Fusion Blanket Molten Salts

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.

Transcript

0:25

You know, when we normally think about a major scientific breakthrough or even just like medical diagnosis, there's this comforting expectation of sheer precision. Right. Like it should feel like engineering. Exactly. Like if you break your arm, you go to the hospital, they take an X-ray and the image just shows that stark, jagged white line across the bone. And the doctor just points to the screen and is like, there it is. It's entirely binary. I mean, the classical world operates on this level of tangibility that just makes sense to our human brains. You have an object, you measure the object, and you define its state. Right. It's either broken or not broken. It's clean, definite. Which is exactly why we lean so heavily on classical physics and traditional diagnostic tools. we fundamentally crave a universe that is visible you know a reality that can be neatly categorized mapped on a grid and measured by instruments that we just completely intuitively understand but then you take a massive step down in scale yeah you step past the cells past the molecules right down into the fundamental world of quantum mechanics and atomic behavior and advanced thermodynamics. And suddenly that metaphorical x-ray machine just completely short-circuits. Yeah, exactly. We find ourselves staring at a diagnostic landscape that is, well, honestly, it's murky. We're trying to look at the absolute foundational building blocks of the universe. And the classical tools we've relied on for decades are fundamentally starting to fail us. It is the absolute definition of diagnostic muddy waters. I mean, when your task requires you to look down at the level of individual correlated electrons dancing around inside a highly radioactive fluctuating molten salt bath at a thousand degrees, our classical ways of seeing don't just blur. They actively blind us to the reality of the physics. And that profound technological blindness is exactly where we're heading today. So welcome to the Deep Dive. We are thrilled you're here with us. Today we're looking at a stack of sources that, I'm not going to lie to you, looks incredibly intimidating at first glance. It's definitely dense material. Oh, absolutely. But once you peel back that dense academic language, it's actually one of the most exciting paradigm-shifting scientific narratives we've seen in years. Yeah. Okay, let's unpack this. So the centerpiece for our discussion today is a groundbreaking paper published in July 2020. The title is quite a mouthful. It's called Quantum Computations on Fusion Blanket Molten Salts. A very catchy title. Right. Very radio friendly. But we aren't just looking at this one paper in isolation. Surrounding that highly technical document, we have a really fascinating stack of supporting intelligence. We're diving into news reports detailing the recent activities of the 119th U.S. Congress. Specifically, legislative bills regarding the National Quantum Initiative Reauthorization Act, right? Exactly. And we're also looking at international defense postures, focusing pretty heavily on a massive new strategy rolling out of Canada. Right. So our mission for this deep dive is to essentially connect the dots across vastly different scales. We're going to trace the behavior of literally a few dozen microscopic atoms floating in a chaotic radioactive salt bath. And then we're going to draw a direct line from that microscopic dance to the macroscopic geopolitical race for limitless clean energy, global quantum supremacy, and national security. It's a journey from the smallest imaginable mathematical fraction to the largest possible global stakes. It's a remarkable narrative through line of, But to really grasp how strange and wonderful this story is, we kind of have to start by looking closely at the people and the institutions that actually authored this July 2026 paper. Because the collaboration itself tells a story. It absolutely does. Yeah, because right off the bat, as I was reading through the author list on this deeply complex nuclear physics paper, there's a detail that is just incredibly quirky. Oh, I know exactly what you're going to say. Right. So you have scientists from Oak Ridge National Laboratory, which makes total sense. I mean, they're the Department of Energy Titans. They run the biggest supercomputers. They do nuclear physics. Sure. And you have IBM Quantum. Obviously, they're bringing the bleeding edge quantum hardware. You have Michigan State University bringing heavyweights in theoretical chemistry. And then you have the Cleveland Clinic. I have to pause on this. The Cleveland Clinic. As in the world-famous hospital in Ohio, the place where you fly in if you need a cutting-edge heart bypass or advanced oncology treatment. That's the one. I'm genuinely confused here. Why in the world is a hospital's medical research wing co-authoring a definitive paper on simulating radioactive nuclear fusion materials? Did I miss a memo where they started building a tokamak reactor in the basement of the cardiology wing? I know, it sounds totally absurd on the surface. But the answer reveals the entire trajectory of how modern computational science is evolving. The Cleveland Clinic hasn't suddenly pivoted into the commercial energy system. sector. Okay, good. That would be a weird pivot. Right. Their involvement has absolutely everything to do with the tool they helped forge. They helped build the computational microscope. The computational microscope. What exactly are they looking at with it? To understand that, we have to rewind just a few months to early 2026. IBM and the Cleveland Clinic formalized a partnership to achieve something monumental in computational biology. They used these exact same quantum-centric computing techniques. The highly specific hybrid workflows we're going to dissect today to simulate a massive, incredibly complex biological protein. When you say massive, what scale are we talking about here? We are talking about a molecular structure consisting of 12,635 atoms. Oh, wow. Yeah. In the world of quantum chemistry, that is a leviathan. The Cleveland Clinic wanted to understand exactly how that specific protein folded, how it interacted with other molecules, and the fundamental electron-level chemistry of it. And the ultimate goal there, I'm assuming, is drug discovery. Of course. To find new pathways for disease treatment by understanding the biology at the most fundamental physical level possible. Okay, so they were essentially trying to build a highly advanced quantum-powered way. to look at extremely complex atomic structures inside the human body. Precisely. They built a methodology, a highly specific algorithmic workflow designed to handle chemical problems that are simply too mathematically dense and complex for normal supercomputers to chew through. And the researchers at Oak Ridge National Laboratory, who are desperately trying to solve a massive real-world bottleneck in nuclear fusion, said, essentially looked at that biological success over in Ohio and said, hey, can we borrow that microscope? That's amazing. They took a tool designed for fighting human disease and aimed it at a miniature star. They pivoted that exact same quantum technique from cracking biological proteins to cracking a critical bottleneck infusion energy. because at a fundamental level, the math doesn't care if it's calculating the shape of a protein in a human liver cell or the behavior of a radioactive salt inside a fusion reactor. Right. It's just math. Exactly. It's all just the complex many-body interaction of electrons. That is wild. The algorithms are completely agnostic. It's just about managing complexity. But before we can really dive into the quantum magic that IBM and the Cleveland Clinic brought to the table, we kind of need to understand the fusion problem that Oak Ridge is actually trying to solve. Right. We need to set the stage with the ultimate goal, the holy grail of energy. Let's talk about nuclear fusion. Because for the listener who might just know it as the physical process that powers the sun or, you know, perhaps cynically as the technology that scientists are always promising is, quote unquote, 30 years away. The running joke in physics. Yeah, exactly. What is the actual tangible promise here? Why are billions of dollars pouring into this? Well, the promise of nuclear fusion is the holy grail. It's limitless baseload electrical energy. But we really need to distinguish it from nuclear fission, which is what all of our current nuclear power plants use. Fission is splitting atoms, right? Correct. In fission, you take incredibly heavy, unstable atoms like uranium or plutonium, and you split them apart. It releases a lot of energy, but it also leaves you with highly dangerous, long-lived radioactive waste that we literally have to bury in mountains for tens of thousands of years. Right. And fusion is the exact opposite process. Entirely opposite. In fusion, you take incredibly light atoms, variations of hydrogen, and you smash them together with so much force that they fuse to form a heavier atom, usually helium. And that releases energy. An absolutely massive amount of kinetic energy. The beauty of this process is that there are no greenhouse gas emissions, and it doesn't produce the long-lived radioactive waste profiles associated with fission. Furthermore, the base fuel can theoretically be sourced from seawater and lithium. I mean, it sounds like the perfect energy source. And reading through these government documents, it is clear the U.S. is taking it incredibly seriously. Oh, absolutely. The sources mention the U.S. Department of Energy's Genesis mission. This isn't just basic lab research anymore. They are explicitly aiming to integrate exascale computing, artificial intelligence, and quantum architectures to finally get commercial gigawatt-scale fusion pilot plants connected to the actual power grid. The ambition is staggering. The ambition is staggering. But, as with all things in physics, the universe does not give away limitless energy for free. There is always a catch. The reality check here is immense. Who is the reality check? What's stopping us from building one tomorrow? It's twofold. First, the physics of the reaction. To get these incredibly light hydrogen atoms to overcome their natural electromagnetic repulsion and actually fuse... you have to essentially recreate the conditions found in the center of the sun right here on Earth. Right. Which sounds impossible. It's difficult. You have to create a high energy density plasma, heat it to hundreds of millions of degrees, and then confine it using massive, incredibly complex magnetic fields in a machine called a tokamak or a stellarator. Containing a miniature star in a magnetic bottle. Yes. And honestly, over the last few decades, we've gotten significantly better at the plasma physics part. We are setting records for how long we can hold the plasma. But the second challenge is arguably much harder, and it's a logistical nightmare. And what's that? It's the fuel itself. Right, because you can't just pump regular hydrogen gas from a tank into the reactor and expect it to ignite. No, you cannot. The easiest fusion reaction to achieve on Earth, and I'm using the word easiest, very relatively here, meaning it requires the lowest temperature to ignite, requires a very specific mixture of two different hydrogen isotopes, deuterium and tritium. Deuterium and tritium. Now, deuterium, I understand, is relatively abundant. It's stable hydrogen with one extra neutron, and you can extract it from ocean water fairly cheaply. Right. But tritium, This is where our story really accelerates. Let's talk about the tritium crisis. Tritium is the bottleneck. It's a radioactive isotope of hydrogen with two extra neutrons, and it is incredibly vanishingly rare in nature. Rare almost feels like an understatement. When I was reading through the sources detailing the supply chain for this, I actually had to laugh out loud. Because it's sort of a gentle satire of our current energy logistics. It really is. We're talking about building a revolutionary, global, limitless energy grid to power human civilization. Right. And the specific fuel we absolutely require to run these reactors. The entire global stockpile of tritium across the whole planet right now sitting in storage facilities is roughly 25 kilograms. That is correct. About 55 pounds for the entire world. You could literally fit the global supply in a large suitcase. That is just wild. And the absolute kicker is... according to Genesis Mission Projections, a single one gigawatt commercial fusion reactor, just one plant providing power that may be a mid-sized city, will guzzle about 0.5 kilograms of tritium every single day. Yeah. So a single power plant would burn through the entire planetary stockpile in roughly 50 days. It's a devastating mathematical reality. It's completely absurd. It's like designing a revolutionary new global airline industry, building these incredible supersonic jets... but designing the engines around a specific chemical fuel that you can only harvest from unicorn tears. That is actually a very apt analogy. You have a magnificent, theoretically perfect engine, but absolutely no way to fill the tank long term. Because, well, tritium is radioactive. It has a half-life of about 12.3 years. Meaning it literally just disappears over time. Right. It decays right into helium-3, so you can't just mine it out of the ground. And you definitely can't stockpile it for decades because it physically vanishes. You have to manufacture it. We need a unicorn farm. Exactly. And this brings us to the proposed solution, which is the specific focus of the Oak Ridge and IBM paper. If you want a commercial fusion reactor to work, the reactor itself has to breathe its own fuel. It has to manufacture the tritium at the exact same rate it's consuming it. It has to make the unicorn tears while the engine is running. How does the tokamak do that? Through an ingenious piece of engineering called a blanket, the idea is to wrap the interior of the fusion plasma core, the artificial star, in a thick physical blanket of material. And the specific material they're focusing on in this study is a high-temperature molten salt called FLIBE. FLIBE-B? Spelled F-L-I-B-E. What exactly is that? It's a chemical mixture of fluorine, lithium, and beryllium. Specifically, it's formulated as two parts lithium fluoride to one part beryllium fluoride. Okay. This salt mixture is remarkable. It has excellent thermodynamic properties. It remains stable under high magnetic fields. and it handles extreme high-temperature environments beautifully. It's actually a leading candidate for cooling advanced small modular fission reactors as well. So it's a great coolant. It is. But in a fusion reactor, it has a very specific secondary job beyond just absorbing heat. Walk us through the mechanics of that job. The plasma is burning at 100 million degrees in the center. The flib salt is circulating in the walls. What happens in the blanket? It starts with the fusion reaction itself. When a deuterium atom and a tritium atom fuse in the center of the plasma... They produce a helium atom and a highly energetic, fast-moving neutron. Now, because a neutron has no electrical charge, it's completely unaffected by the massive magnetic confinement fields holding the plasma together. It shoots straight out of the core, flying outward like a microscopic bullet, and slams directly into the fly blanket circulating in the walls. Okay, so we have this incredibly high-speed neutron rocketing into a wall of thousands of degrees of molten salt. Right. And when that energetic neutron hits a specific isotope in the salt specifically, lithium-6, a nuclear transmutation, occurs right there in the liquid. What happened? The lithium-6 atom absorbs the neutron, becomes highly unstable, and instantly fragments. It breaks apart into a stable helium-4 atom in a newly minted, precious tritium ion. Boom. A unicorn tear is generated right inside the wall of the reactor. Problem solved, right? The reactor makes its own fuel. What's fascinating here is it's not just the breeding of the fuel that's the issue, but the chemistry of recovering it. Making it is only half the battle, because now you have a freshly minted tritium ion floating in a chaotic, fiercely fluctuating, thousand degree radioactive soup of fluorine, lithium, and beryllium. And tritium doesn't just sit there politely waiting for us to scoop it up with a net. No, it interacts. It's highly reactive. In this specific philibon environment, the newly formed tritium ion tends to bind very strongly to the highly electronegative fluorine atoms. It often forms tight, complex, multicenter bonds, specifically what chemists call bent or stretched FTF bonds, where the single tritium ion is essentially sandwiched and trapped. Between two fluorine ions. So the fuel gets locked away in the salt. But we need that fuel back in the core. Exactly. If we can't extract that tritium efficiently, rapidly, and continuously, the reactor starves and shuts down. Right. We have to design complex chemical extraction systems. perhaps using gas bubblers or chemical sponges to pull that tritium out of the salt. But to design an extraction system that actually works, engineers must know precisely how tightly that tritium is holding onto the fluorine. They need to know the exact strength of the grip. Yes. And the molten salt is not a static, predictable environment. It's a wildly fluctuating liquid order. So squishy. Very. Very. into this absolute chaos. You introduce the tritium, and it gets locked into these shifting chemical bonds. Which brings us to the computational roadblock. Because if we're going to spend $10 billion building a commercial fusion reactor, we probably shouldn't just guess how the chemistry works based on a whiteboard sketch. No, that would be a very expensive mistake. We need to model it. We need to simulate that chaotic molten salt environment on a computer to see exactly how hard it is to pull the tritium out. And that requires monumental computing power. To fully appreciate the quantum breakthrough in this paper, we first have to examine the journey the scientists took and the massive brick wall they slammed into when using the most powerful traditional computers on Earth. Right, because Oak Ridge didn't just instantly jump to quantum computers. They started with the heavyweights of the classical computing world. The sources specifically mention systems like Frontier at Oak Ridge and Perlmutter at the National Energy Research Scientific Computing Center. Yeah, these are the big ones. To give our listeners some context, these aren't just fast laptops. These are exascale and pre-exascale supercomputers. They take up entire warehouses and can perform quintillions of calculations per second. They are the pinnacle of classical computers. and the Oak Ridge scientists used these supercomputers to run massive, incredibly intensive molecular dynamic simulations of the F-leap salt. Specifically, they utilized a workflow combining ab initio molecular dynamics with machine learning force fields, often abbreviated as AMD-MLFF. Okay, let's step back and explain the mechanics of that. For someone who isn't a computational chemist, what does an AMD slash MLFF simulation actually look like under the hood? Imagine setting up a highly complex virtual sandbox inside the computer. You mathematically define a box, and you populate it with a few hundred virtual atoms of lithium, beryllium, and fluorine, carefully balancing the ratios to match the felib recipe. Got it. Then you tell the supercomputer, heat this virtual box to 1300 Kelvin, which is about 1900 degrees Fahrenheit, and apply the fundamental laws of physics to see how these atoms push, pull, and move over time. So the computer calculates the position of everything, steps forward a tiny fraction of a second, and calculates it all again. Exactly. The supercomputer calculates the forces between all the atoms at one unimaginably tiny fraction of a femtosecond, moves them slightly based on those forces, recalculates everything, moves them again, and iterates this millions of times. This generates a molecular dynamics trajectory. It's like generating an incredibly high-definition, microscopic, slow-motion video of the liquid bubbling, shifting, and bonding. Yes. And I want to be clear. These classical supercomputers are actually fantastic at simulating the bulk thermodynamic properties of the liquid. Like what? Well, if you want to know the overall density of the salt at $1,500, degrees frontier can tell you if you want to know the general coordination numbers meaning on average how many fluorine atoms are dynamically dancing around a single beryllium atom at any given moment classical supercomputers handle that brilliantly but clearly they hit a dead end or we wouldn't be talking about ibm quantum today if frontier could do quintillions of calculations a second Where exactly did the classical supercomputers fail? They failed at a very specific, deeply quantum mechanical concept known as electronic correlation. Electronic correlation. Let's dive deep into that because this seems to be the linchpin of the whole problem. It is. When you're trying to figure out exactly how tightly a tritium atom is bound to a fluorine atom inside that chaotic soup, you can't just look at the atoms as solid, rigid billiard balls bouncing off each other. You have to look at the electrons. You have to look at their electron clouds. Electrons are negatively charged particles, which means they constantly fiercely repel each other, but they're also entirely constrained by the laws of quantum mechanics. Meaning they don't just act like tiny magnets pushing away from each other in straight lines. Not at all. The movement of one single electron is instantly and dynamically correlated with the movement of all the other electrons around it. It's an incredibly complex many body bands. Right. In an ionic salt like Flib, as we discussed, the fluorine ion is highly polarizable. It doesn't have a rigid shell. It has a big squishy cloud of electrons that gets actively pulled, distorted, and warped by the highly charged beryllium ions. nearby. So the environment isn't rigid. It's a constantly shifting, squishy, pulsating quantum dance of repulsion and attraction. And every time one electron shifts, every other electron has to instantly adjust its position. Precisely. And calculating that many body quantum dance, finding the exact mathematical energy state of all those instantaneously interacting electrons requires a level of computational complexity that scales exponentially. Right, because every new particle adds a massive amount of new interactions. Exactly. As you add more electrons to the simulation, the math doesn't just get a little harder, it explodes in size. Even classical supercomputers like Frontier simply run out of memory and processing power. They literally choke on the math. So how do classical computational chemists normally get around that? Yeah. They can't just throw their hands up. They use brilliant mathematical approximations. The most common one is called density functional theory, or DFT. Instead of trying to track the exact instantaneous correlation of every single individual electron, DFT essentially averages them out. Oh, okay. It treats the electrons as a smeared out continuous cloud of charge densities. It simplifies the math dramatically, allowing the supercomputer to actually finish the calculation. But a simplification is still a compromise. It's a blurry lens. And why does that microscopic mathematical compromise matter to a nuclear engineer? I mean, if density functional theory gives the supercomputer a 90% accurate picture of the tritium bond... Isn't that good enough to build a pump? Actually, it's catastrophically inadequate. In the unforgiving math of thermodynamics, a 10% error margin means your billion-dollar commercial fusion plant might simply fail to operate. Walk me through that. Why is 10% so disastrous? In most engineering fields, 90% accuracy is a solid passing grade. Because of what they're trying to calculate. To know if you can actually pull the tritium out of the salt, chemical engineers need to derive a specific metric called the excess chemical potential. The excess chemical potential. What does that mean in simple terms? In simple terms, this is the exact thermodynamic cost, the energetic bill you have to pay, to break that tritium free from the fluorine. If you want to design a chemical sponge or a vacuum system to extract it, you need to know exactly how tightly it's holding on. The previous studies that relied purely on classical DFT models deviated from the true experimental free energy by roughly 10%. Okay, so the energy bill is off by 10%. It sounds small, but chemical equilibrium equations are exponential, not linear. A 10% error in predicting the underlying binding energy can mean your macroscopic prediction for how much tritium you can actually recover is off by a factor of 10 or a factor of 100. Wow. It might even mean your extraction process runs backward, mathematically speaking, and pushes the tritium deeper into the salt. In computational chemistry, to build real-world systems, you need a standard called chemical accuracy. Chemical accuracy. What is the threshold for that? You need to know the energy of the system down to one kilocalorie per mole or one kecal per mole. Is that a lot of energy? To put that in perspective, a kilocalorie per mole is an incredibly tiny amount of energy on a macroscopic scale. But chemically, it's the difference between a reaction happening and not happening. If your simulation is off by more than one kikaltamol, your macroscopic engineering designs the pumps, the pipes, the extraction filters are essentially built on thermodynamic quicksand. So, to summarize the immense roadblock Oak Ridge was facing, they have Frontier, the biggest supercomputer in the world. They run simulations of the molten salt. But because they cannot calculate the exact squishy quantum repulsion of the correlated electrons because the math scales exponentially enter day, they have to use density functional theory. Yes. DFT blurs the image. The blurry image results in an energy estimate that is off by more than one kecal per mole. And therefore, they absolutely cannot guarantee to the Department of Energy that the tritium recovery system will actually work. You summarize it perfectly, and this is the pivot point of modern science. When the world's fastest classical supercomputers hit an insurmountable wall of exponential quantum complexity, the team couldn't just build a bigger classical computer. Right. They had to change the rules of calculation entirely. They had to call in the quantum cavalry. Which brings us to this specific revolutionary experiment detailed in this July 2026 paper. Enter IBM Quantum and the Cleveland Clinic methodology. I love the mission architecture here because it's crucial to understand that it didn't just throw the classical supercomputer in the trash. It's about a highly specialized, deeply integrated partnership. Absolutely. The specific hardware they used is the IBM Heron R3 processor on a system designated as in Boston. And it operated as part of what they call a hybrid quantum-centric supercomputing workflow. The hybrid approach is the only way forward. We have to be realistic about hardware. A quantum computer in 2026 is not designed to run an entire molecular dynamics simulation of a thousand atoms over time. It just can't do it. It doesn't have the architecture for that. It would be completely overwhelmed by the basic bookkeeping. You have to be incredibly strategic. You have to carefully decide what specific mathematical tasks you give to the classical machine and what tasks you reserve for the quantum machine. You know, it makes me think of a high-end Michelin star kitchen. Let's see. This analogy holds up. Okay, let's hear it. The classical supercomputer frontier is the head chef. The head chef handles the massive bulk of the recipe. They're organizing all the ingredients, managing the timeline, doing the heavy lifting of figuring out where the hundreds of atoms are generally positioned in the liquid. They're running the kitchen. Okay, sure. But then, for this one specific dish, there's an incredibly delicate, wildly complex reduction sauce that requires absolute molecular perfection. The tricky, squishy atomic bonds of the fluorine and the tritium. The head chef knows how to make it, but they literally cannot do the math fast enough or accurately enough without burning it. Right. The exponential scaling. Exactly. So the head chef outsources that one specific mathematically dense sauce to the incredibly specialized sous chef. And the sous-chef is the IBM quantum computer. That is actually a phenomenal way to visualize the hybrid workflow. The head chef, the classical supercomputer, looks at the whole chaotic thousand-degree salt bath and realizes it can't cook the whole thing perfectly. So it extracts tiny, highly representative samples. And how did they do that? To do this, they used a technique pioneered heavily in that Cleveland Clinic biological protein study we talked about. It's called the embedded wave function method, or EWF fragmentation. Let's deeply unpack EWF fragmentation. Yeah. Because reading the paper, this methodology seems to be the absolute core of how they managed to pull this off. You can't simulate the whole ocean of salt, so you have to scoop up a cup of water. Exactly. Exactly. From their massive classical AMD simulation of the bulk liquid, they essentially hit pause and extracted instantaneous snapshots of the atomic positions. So they freeze time. Yes. From those frozen snapshots, they pulled out tiny localized clusters of atoms. Specifically, they isolated nine different structural configurations. They looked at a 21-atom-neutral Phleib cluster, a 22-atom-negatively-charged Phleib cluster, and crucially, a 23-atom cluster where a tritium ion was actively bound to the fluorine. So they're dealing with just 21 to 23 atoms at a time. Intuitively, that sounds small. But quantum mechanically, if you're trying to calculate exact electronic correlation, 23 atoms is still way too big, right? Yes. For exact quantum chemistry, without using the blurry approximations of DFT23 atoms is still a mathematical monster. So the embedded wave function method takes that 23 atom cluster and partitions it even further. It slices the cluster into atom-centered fragments. Atom-centered fragments. What does that mean? Imagine taking that 23 atom cluster and shining a spotlight on just one single atom at a time, say one central lithium atom, and isolating its immediate surrounding bath of strongly entangled electrons. You create a unique mathematical fragment for every single atom in the cluster. Okay, so we chop the problem down in stages. From a massive whole liquid down to a 23-atom cluster, down to an individual atom and its immediate, deeply entangled neighbors. Now we have all these fragments. What is the routing logic? How does the system decide who cooks what? This is where the physical chemistry of the salt dictates the computational routing strategy. The system analyzed the size of the electron clouds, specifically the number of active molecular orbitals for each individual fragment. For the small, predictable fragments, which were mostly the lithium-centered ones, possessing under 13 molecular orbitals, the head chef kept the job. They solved those classically using a highly accurate, non-approximate method called full configuration interaction, or FCI. Let's pause on FCI. Why is the classical computer allowed to do FCI in the small fragments? Because full configuration interaction is the absolute gold standard of classical quantum chemistry. It calculates every single possible electron correlation perfectly without the blurring of DFT. But it scales exponentially. Exactly. Because it scales exponentially, classical supercomputers can only run FCI if the system is incredibly small. Under 13 orbitals, the classical computer can chew through the SCI math perfectly. What, the big stuff? The big stuff. The large, highly complex fragments. These are the fragments containing 13 up to 33 active molecular orbitals. This represents mostly the dense, highly polarized, squishy fluorine environment. and the absolutely critical tritium-fluorine bonding regions. So these are the hard tickets. Yes. These fragments are simply too mathematically massive for the classical FCI solver to handle efficiently across thousands of required calculations. So any fragment with 13 or more orbitals was instantly dispatched to the IBM quantum hardware. The quantum sous chef gets all the hard tickets. Yes. And the way the quantum computer solved those dense fragments is fascinating. They didn't just ask the quantum computer to spit out a single magic number. They used a technique called extended sample-based quantum diagonalization, or X to SQD. This is where my mind was completely blown reading the methodology. Okay. Because they didn't just run a standard algorithm from start to finish. They used the quantum computer in a very specific, almost statistical way. They describe it as doing quantum on the fly. Right. In X-DesQD, the quantum computer is not being used to rigorously solve the entire Schrodinger equation from start to finish all by itself. Instead, it acts as a highly advanced, unimaginably fast configuration generator. How does that actually work mechanically? Well, you load the mathematical parameters of this complex fragment into the quantum processor using a parameterized quantum circuit. In this paper, they use something called a local unitary cluster Jastro-NSATS. Which is basically... It's essentially a set of instructions. You run the circuit, the quibits entangle and represent the electron state. And then you measure the quibits. You take a shot. Each single shot collapses the quantum state and gives you one candidate electron configuration. It gives you one highly probable way that the electrons could be arranged at that instant. And they don't just do this once. Not at all. They do this thousands or even millions of times in rapid succession. For the fragments under 20 orbitals, they took 100,000 shots. For the massive fragments over 20 orbitals, they commanded the IBM Heron processor to take 1 million shots. Here's where it gets really interesting. They're rapidly sampling the quantum state, pulling out the most likely physically relevant ways these electrons are dancing around each other. I want to build another analogy here to really grasp the sheer scale of the math they avoided by doing this. Go for it. The paper notes that if you tried to solve a 33 orbital fragment using classical brute force methods, the Hilbert space dimension which is the number of possible electronic configuration you'd have to calculate, is roughly 540 billion possibilities. 540 billion for one single fragment out of hundreds of fragments they needed to process. That is mind-boggling. Let's visualize that. Imagine the classical supercomputer is a guy with a flashlight trying to find the absolute lowest point in a massive, completely dark, 540 billion acre mountain range. That's a lot of walking. Exactly. The classical approach means he has to walk every single square inch of those 540 billion acres, measure the elevation, and compare it just to be sure he found the bottom. It would take a lifetime. But the quantum computer using XSQD doesn't walk the mountains. No, it doesn't. Taking a shot is like the quantum computer instantly flooding the entire mountain range with a massive rainstorm, and just watching where the puddles form. The water naturally finds the lowest points instantly. That is a phenomenal visualization of quantum sampling. The quantum computer uses the physical laws of nature superposition and entanglement to naturally zero in on the configurations that actually matter energetically. It generates a massive pool of highly probable candidate configurations, the puddles. Then it hands that specific pool of data back to the classical computer. The sous chef hands the reduction back to the head chef. Yes, exactly. The classical computer takes those million shots, cleans up the data because current quantum hardware still has physical noise, so they apply air mitigation techniques, identifies the absolute most dominant configurations, and then performs a final, much smaller classical diagonalization to get the true ground state energy. Wow. It navigates that exponential 540 billion space efficiently, and it did it incredibly fast. The large fragments requiring a million shots ran in about four and a half minutes on the I-B. 4.5 minutes instead of centuries, which brings us to the results of this monumental effort, a stunning success, and then a 110 kilocalorie per mole elephant in the room. Yeah. Let's start with the triumph, because we shouldn't gloss over what a leap this is. Did the Quantum Sous Chef actually deliver the perfectly measured sauce? It delivered spectacularly. This is the primary celebrated victory of the July 6, 2026 paper. Across all nine complex clusters, for all the dense fragments dispatched to the QPU, the quantum hardware workflow reproduced the absolute gold standard classical math to within 0.7 kilocalves per mole. Wait, let me make sure I'm hearing this right. The quantum computer, dealing with the inherent hardware noise and the sheer complexity of the fluorine-tritium bonds, got the exact same answer as the theoretical classical ideal down to a fraction of a single kilocalorie. Yes. The mean absolute deviation was just 0.3 kcal per mole. And keep in mind, to verify this, they had to run the classical gold standard, the full configuration interaction... on those large fragments, using massive classical supercomputers like Frontier, taking thousands of GPU hours just to generate a benchmark to prove the quantum computer was right. That's crazy. The quantum workflow matched that extreme chemical accuracy using a fraction of the computation time and energy. That is a massive historic validation. It unequivocally proves that quantum hardware can solve incredibly complex, strongly correlated charged ionic systems with strict chemical accuracy. It proves the hybrid head chef and sous chef approach works fundamentally. It is a profound milestone. To our knowledge, as the authors state, it is the first such demonstration for a charged ionic system in an inorganic molten salt. they proved beyond a shadow of a doubt that the quantum fragment solver works. But in science, there is almost always a but. And as I was reading the discussion section of the results, I hit a massive plot twist, a real wait a minute moment. And the elephant in the room. Exactly. The quantum solver worked perfectly. It calculated the exact energy of the individual fragments it was handed. But when the scientists took all those perfectly calculated fragments, added them back together, and calculated the overall binding energy of the tritium to the entire salt cluster, The final assembled answer was wildly different from the unfragmented whole molecule classical models. It was off by a staggering amount. The paper explicitly states the difference was 110 kilocal per mole. Let me push back on this because I'm really trying to wrap my head around it. We just established that chemical accuracy requires being within one kilogal per mole. Correct. If the quantum computer did its job perfectly, if it got a 0.7 kilogal per mole accuracy on the pieces, why in the world is the final assembled answer of the whole molecule off by 110? If I measure all the ingredients perfectly, the cake shouldn't explode. Did the kitchen analogy fail when they tried to plate the meal? Did we chop the molecule up too much? You've deduced the exact problem. Did we chop the molecule up too much? Yes. The error was absolutely not in the solving of the fragments. The IBM Heron quantum processor did exactly what it was mathematically asked to do. The error was in the setup. The fundamental flaw was the algorithmic bias of the embittery wave function fragment construction itself. Explain that to me. How does the setup fail? What did they physically lose when they chopped it up? Think about what the EWF method actually did to create those fragments. it isolated a central atom like a fluorine atom and its immediate entanglement bath the nearby electron orbitals it interacts with most strongly but to make the math manageable even for the quantum computer the algorithm imposed a rigid cutoff threshold it truncated the bath it essentially drew a hard artificial line and said we are only going to care about the electrons that are this physically close and this strongly entangled all the other electrons further away in the cluster we will simply ignore for the purpose of this fragment They put artificial blinders on the sous chef. They told it to ignore the rest of the kitchen. Exactly. But in a highly polarizable, fluctuating liquid salt like Phleib, everything is subtly, profoundly connected. By truncating the entanglement bath, by artificially limiting how the isolated fragments talk to the rest of the molecule, they lost crucial physics. Specifically, they lost long-range dispersion interactions. Dispersion interactions. What are those in a physical sense? Imagine the electron clouds are like water balloons. Even if two water balloons are not directly touching, if they're floating in the same small pool, pushing on one sends ripples that gently push on the other. Dispersion forces are the subtle long-distance pushes and pulls between fluctuating electron clouds that span across the entire 23-atom cluster. By chopping the cluster into strictly isolated fragments, they deleted those ripples from the mass. So the quantum computer perfectly calculated the energy of the isolated pieces. But because the pieces were artificially isolated from the holistic reality of the molecule, they missed the total energy of the entire system by 110 kilogal per mole. The butterfly effect of an electron on the far left side of the cluster pulling on an electron on the far right side was totally lost. And as we established earlier, a 10% error is bad. 110 kcal per mole is a fundamental catastrophic failure for predicting actual real-world tritium chemistry. From an immediate engineering standpoint, yes, you cannot hand that number to the Department of Energy and tell them to build a reactor based on it. But from a broader scientific standpoint, this specific failure is actually a massive accelerated success. I am going to need you to justify that. How is 110 kcal per mole error a success? Because in advanced computational chemistry, identifying exactly where the error lives is more than half the battle. This experiment brilliantly isolated the variables. They proved definitively that the quantum solver is ready, the hardware works, the algorithms for extended sample-based quantum diagonalization work flawlessly. Therefore, the primary bottleneck is no longer can the quantum computer do the incredibly dense correlation math. The bottleneck has been successfully shifted entirely to how do we better define the fragments we feed to it? The hardware is caught up. The algorithms just need to get smarter. So what is the roadmap forward? How do they fix the fragmentation so they don't lose the dispersion ripples? The authors outline it very clearly at the end of the paper. They need to build chemically aware fragments. Instead of just using a blunt, rigid numerical distance threshold to chop up the molecule... They need much smarter algorithms that actively account for the chemical bonding nature and the long-ringed correlations without ballooning the computational cost back up to impossible levels. And they explicitly point to integrating artificial intelligence to do this. AI entering the chat. Exactly. Machine learning models could analyze the bulk liquid and automate the intelligent construction of these chemically meaningful fragments. The AI would learn which distant electrons are actually secretly communicating with each other, ensuring that the critical long-range physics is preserved in the fragment, while still keeping the overall problem sizes manageable for the quantum hardware. So the classical head chef needs to use AI as an executive planner. to get much smarter about how they prep and slice the ingredients before handing them to the quantum sous chef. This is absolutely fascinating. We are watching the iterative, grinding battle for quantum accuracy play out in real time, step by step. We really are. But this isn't just an academic exercise happening in a vacuum for the sake of writing papers. The ability to simulate these advanced materials, to crack the fusion energy code, to fundamentally master quantum computing, the world's major superpowers are watching this exact timeline, and they are actively legislating it. Yes, the geopolitical stakes are immense. Which requires us to pull our focus back from the microscopic world of flibes, salts, and kilocalories, and zoom out to the macroeconomic geopolitical landscape. The context in which this IBM and Oak Ridge paper was published... is vital to understanding its true impact. This research does not exist in a silo. Right. Let's look at the news sources regarding the 119th U.S. Congress and the National Quantum Initiative Reauthorization Act. This is a major bipartisan bill led by Senators Cantwell and Young in the Senate and Representative Weber in the House. For our listeners, we're going to carefully and neutrally outline exactly what this legislation does because it represents a massive tectonic shift in how the federal government views this technology. We're not taking sides on the politics here. No, simply reporting on the explicit strategies laid out in the bill's text and the surrounding defense coverage. And it is a profound shift. If we connect this new legislation to the bigger picture, you have to look at the original 2018 National Quantum Initiative Act. In 2018, the legislation was primarily focused on basic foundational lab research. It was about funding universities and national labs to essentially figure out if quantum computing was even viable long term. So pure research. Exactly. The 2026 Reauthorization Act is a fundamentally different beast entirely. How so? What changed in the text? It represents an aggressive, unapologetic pivot toward commercialization, supply chain resilience, and explicit geopolitical competition. The text of the reporting specifically notes the aim is to maintain a technological edge over international competitors, primarily China. The era of viewing quantum computing purely as a theoretical physics experiment is officially over. It's applied science now. It is now explicitly viewed by lawmakers as a critical pillar of both national security and future economic dominance. And you can see that urgency in the specific mandates they're handing down to government agencies. For example, the bill now requires the Defense Advanced Research Projects Agency, DARPA, to measure the real-world performance of emerging quantum applications. They don't just want white papers and theories anymore. They want working tools that solve actual problems. Exactly. Which ties in perfectly to why this exact IBM and Oak Ridge Flive experiment is making headlines and getting funding. It is exactly the kind of applied science solving a real-world critical energy bottleneck that Congress is aggressively demanding to see. Precisely. They are demanding utility. Furthermore, the Reauthorization Act formally adds NASA to the initiative for the first time. They're authorizing specific research and development funding for space-based quantum sensing and satellite-based quantum communications. They're looking at deploying quantum sensors in orbit. That is incredible. And there's a huge focus on logistics, right? Yes. They're directing the Department of Commerce to thoroughly map global quantum supply chains. They want to identify any critical foreign dependencies, like where are the rare earth metals coming from, who makes the specialized cooling equipment, and build defensive strategies to aggressively protect the domestic quantum industry. It's a full governmental mobilization. Okay. And it's vital to note that it's not just the United States making these moves. Our sources detail a massive, parallel, and highly aggressive move by Canada. Canada has rolled out a comprehensive dual-track strategy, their national quantum strategy, paired with a brand-new, heavily-funded $900 million CAD defense industrial strategy. Canada is pivoting just as aggressively, if not more so, in certain focused areas. They recently launched the Canadian Quantum Champions Program, which comes with an initial $92 million CAD designed specifically to anchor domestic talent and rival the U.S. DARPA initiatives. And I found their funding philosophy really interesting compared to the U.S. approach. Instead of spreading small research grants around to dozens of different universities, they're anchoring heavily into a few designated domestic champions. They granted up to $23 million CAD each to four specific, highly advanced companies, Xanadu, Nordquantique, Photonic, and Ennion Systems. Yeah. The explicit goal is to build scalable, bolt-tolerant quantum computers right there in Canada and keep the intellectual property, manufacturing, and the hardware strictly domestic. The security angle is paramount in the country. Canadian strategy. A massive portion of that $900 million defense strategy is explicitly earmarked to unleash quantum technology for military use. They're focusing heavily on quantum sensing for GPS-free navigation. Why is GPS-free navigation so critical? Because classical GPS relies on satellites, which can be jammed, spoofed, or shot down. If you have a highly sensitive quantum sensor that can read minute changes in the Earth's magnetic or gravitational field, a submarine can navigate perfectly underwater for months without ever needing to surface for a satellite ping. That's a huge tactical advantage. It is. They're also investing heavily in quantum communications for anti-spoofing and entirely jamming resistant military networks. This all sounds incredibly futuristic, but there is a looming deadline in the Canadian documents that really caught my eye and brought the stakes crashing back to the present. It's the mandate regarding post-quantum cryptography, or PQC. This mandate is arguably one of the most critical drivers of the entire global tech industry right now. The Canadian Center for Cybersecurity has legally mandated that all federal IT infrastructure must fully migrate to post-quantum cryptography by the year 2031. 2031. That is essentially tomorrow in terms of massive government IT procurement and overhaul. Let's really dig into why this is happening. Why are they suddenly terrified of quantum computers breaking encryption? How did the quantum computer actually do that? To understand the threat, we have to understand how our current encryption works. Almost all of the classical encryption that protects our bank accounts, our power grids, our personal emails, and military communications, specifically RSA encryption, relies on a very specific mathematical trick. Prime factorization. Explain prime factorization for us. It's easy for a classical computer to take two massive prime numbers and multiply them together to get an unimaginably huge number. That is easy. But if I give a classical computer that massive final number and ask it to figure out which two primes were multiplied to create it, it's incredibly difficult. Right. For the size of numbers we use in encryption, it would take a classical supercomputer millions of years to guess the right answer. That computational difficulty is the lock on the door of the Internet. So classical computers are just too slow at guessing to pick the lock. How does a quantum computer pick it? Through something called Shor's algorithm. A quantum computer doesn't guess one by one. Because of superposition and quantum interference, a powerful enough quantum computer can run Shor's algorithm to essentially find the periodic structure of the number and extract those two prime factors exponentially faster. What takes a classical computer a million years could take a fault-tolerant quantum computer a matter of hours or days. It just completely shatters the fundamental math the block is built on. Exactly. And the intelligence community is deeply concerned about a strategy called Harvest Now, Decrypt Later. Even though a quantum computer powerful enough to run Shor's algorithm doesn't exist yet, adversaries can harvest massive amounts of encrypted data today, stored on hard drives, and simply wait 10 years until they build a quantum computer to unlock it all retroactively. That is terrifying, which explains the massive rush to post-quantum cryptography. So how does PQC actually defend against a quantum computer? If they can crack primes, what math do we use? PQC moves away from prime factorization entirely. It relies on completely different mathematical foundations, most commonly lattice-based cryptography. Lattice-based cryptography. Let's visualize that. Imagine a massive multidimensional grid, a lattice with thousands of intersecting points. The encryption essentially hides the secret key at a specific point within this incredibly complex multidimensional geometric structure. To break it, the computer has to find the absolute shortest vector, the shortest path, to that point from the origin. And quantum computers are bad at finding paths and grids. Precisely. Unlike finding the period of a prime number, finding the shortest vector in a massive multidimensional lattice is a problem that does not currently appear to have an exponentially faster quantum shortcut. It's a math problem that is equally hard for both classical supercomputers and quantum computers. So governments like Canada are mandating that by 2031, all their systems must be upgraded to this multidimensional lattice map to prepare for a world where quantum hardware becomes fully fault tolerant. Yes. And this intense geopolitical and security pressure is directly, forcefully shaping the private tech ecosystem. The hardware companies are racing to meet these massive government and commercial demands. Look at the strategic partnerships forming in the private sector. Hewlett Packard Enterprise, HPE Cray, which builds massive classical supercomputers, is now building unified, full-stack, hybrid quantum supercomputing platforms. They're building the commercial off-the-shelf versions of the head chef and sous chef kitchens we talked about earlier. They want to sell these hybrid systems to corporations. Exactly. They are partnering with an array of quantum companies like Intel, Regetti, Quantinuum, and Quera to integrate diverse quantum hardware modalities directly into classical supercomputing networks. They know the hybrid workflow is the future. And the quantum hardware itself is evolving rapidly to meet this demand. D-Wave Quantum, for instance, recently announced a roadmap explicitly targeting 100 logical quibits by 2032. We absolutely need to clarify the term logical quibits. Because normally when you read a press release from a tech company, they're boasting about having hundreds or thousands of physical quibits. What is the distinction and why does it matter? The industry is maturing and they're moving away from just counting raw physical quibits because raw physical quibits are inherently noisy, fragile and error prone. They lose their quantum state if a stray cosmic ray hits them or if the temperature fluctuates by a fraction of a degree. Right. D-Wave is focusing on a scalable, superconducting dual-rail architecture that is specifically designed to suppress and lower error rates right at the foundational hardware level. So a physical quibbit is fragile. What is a logical quibbit? A logical quibbit is a construct of error correction. It's essentially a group of many physical quibbits that are mathematically linked and working together to act as one single, perfect, incredibly reliable quibbit. If one physical quibbit in the group flips or makes an error, the others detect it and correct it. It's like having a choir of singers. If one person hits a slightly flat note, the collective sound of the rest of the choir drowns it out and keeps the melody perfect. A fantastic analogy. Getting to 100 logical, fully error-corrected quibbits by 2032 would be a monumental, paradigm-shifting achievement. It would allow scientists to run incredibly deep, complex algorithms reliably without the calculation degrading into noise. They're focusing entirely on fidelity. They're chasing absolute accuracy, which brings this entire narrative full circle, doesn't it? Because fidelity chemical accuracy down to one kilocalorie per mole is exactly what the Oak Ridge and IBM scientists... so desperately needed to calculate the behavior of that fall-life salt. Exactly. They needed a quantum computer that could provide incredibly accurate, reliable answers about how tightly that tritium was bound to the fluorine. The hardware is getting there. As the IBM paper proved, the extended sampling algorithms are getting there. And the immense, undeniable geopolitical will driven by energy crises and national security fears is providing the unlimited fuel to accelerate the timeline. It really is an incredible convergence of disciplines. We've journeyed all the way from the microscopic, desperate shortage of a radioactive isopope. The elusive unicorn tears required to power the global energy grid to the massive classical supercomputers that literally melted down mathematically, really, trying to simulate it. It's a huge journey. We explored how the ingenious hybrid quantum systems, acting as a Michelin star kitchen, successfully modeled the exact fragments of that chaotic salt. It proved the hardware works, even while dropping the plot twist that we need smarter AI to chop the molecule up correctly so we don't lose the long-range physics. And finally, we zoomed all the way out to see the geopolitical machines in Washington and Ottawa aggressively rushing to fund, protect, and weaponize the next iteration of this exact technology. It's all connected. So after all of this, what does this all mean for you as a listener, as a taxpayer, as someone simply watching the future unfold in real time? I want to leave you with a philosophical extrapolation, a thought experiment based on the mechanics of what we've discussed today. We are currently pouring billions of dollars into utilizing the most advanced mathematically bleeding edge technology in human history, quantum computing, to solve a critical roadblock in creating a miniature star on Earth. But remember, the central plot twist of the Eflide paper. The 110 kilocalorie elephant. Right. The embedded wave function fragmentation method forced the classical computer to chop the molecule into isolated, artificially separated pieces just so the quantum computer could digest the math. And by doing that, by forcing an artificial separation, they lost the 110 kilocal per mol holistic paper. Yeah. individual atoms, individual fragments, individual components on a spreadsheet. It's how we're wired. But if a quantum computer fundamentally operates on the principle of interconnected entanglement, where every single particle is simultaneously, instantaneously relating to and affected by every other particle in the system... Perhaps the next great scientific breakthrough isn't just about building a quantum processor with more logical quivits. Perhaps the true revolution is developing a completely new holistic mathematics, a framework that stops trying to chop the universe into bite-sized classical fragments, and finally allows us to compute nature as the fully entangled, unified, breathing hole that it actually is. a completely new way of looking at the universe no more murky diagnostic waters no more broken classical x-ray machines just the entire deeply entangled picture viewed all at once that is certainly a thought to mull over as you go about your day thank you so much for joining us on this incredibly deep dive from the chaotic molten salt of a fusion reactor to the halls of congress we will see you on the next deep dive