Hi everyone, this is Charles Hoskinson broadcasting live from warm, sunny Colorado. Always warm, always sunny, sometimes Colorado. Every now and then, something comes up that is super cool, and it’s the result of a brilliant, long, brutal slog. I have to hand it to Microsoft; they’ve done something absolutely remarkable and incredible. I wanted to make a quick video about it. I’m going to spend quite a bit of time looking through this, and the paper is very dense and hard to read but quite interesting.
So, this is the paper here. Basically, what Microsoft did is figure out how to build a topological quantum computer. We’re going to format the background a little bit and change it to squares; that’s a little easier to see. A topological quantum computer is a long-theorized but previously impossible thing to build. In general, quantum computers have different states that allow you to get a sneak peek at a large, expensive calculation, giving you some clarity on how to solve it. That sneak peek is efficient, and as a result, there are all kinds of categories of things that make these super useful to society. For example, you might want to break cryptography, simulate biochemistry, or predict weather. The long and short of it is that if you can build one of these things that give you that sneak peek capability, it’s super useful because many calculations in these categories take more time than the universe has to solve.
For the cryptocurrency industry, we care a lot because you could break the cryptography of most cryptocurrencies. However, not all cryptocurrencies are vulnerable; for example, Starks are immune to quantum computers. A quantum computer is useful for solving the discrete logarithm problem.
Now, what is this whole topological thing about, and why is it interesting? In general, the power of a quantum computer is called a qubit, and more is better. As you increase the qubit count from one to n, you start getting the capability to solve these problems. The problem is that as your qubit count goes up, the rate of errors also increases, and those errors can destroy your calculations. Eventually, you reach a threshold; currently, that’s somewhere around a few hundred qubits, which is not particularly useful for where we want to be. There are physical and logical qubits, and many things beyond the scope of this video, so we’ll keep this brief.
Everybody in quantum computing has been trying to figure out a hardware approach. Some approaches use photons, called photonic quantum computing, while others use trapped ions. Topological quantum computing is another one of those approaches. Here’s what makes it interesting: it shows the dedication of Microsoft. They spent 17 years on this—17 years! An enormous amount of time. They first did something called molecular beam epitaxy, which is a top-shelf, high-end way of building semiconductors. Basically, it looks like this: you build a chamber with pumps that create a vacuum, and you have diffusers inside it. You have a substrate, usually a crystal, and a cryopump. You shoot atom by atom materials into this vacuum chamber, growing a thin film atom by atom of whatever you want, typically at a rate of about three micrometers per hour. It’s super slow.
This allows you to build any exotic material you want. Microsoft first took indium and arsenide, typically used in the semiconductor industry, and blended it with aluminum, which is used in the superconductor business, to create a hybrid exotic material they call a topological superconductor. The only way to build this topological superconductor is with one of these molecular beam epitaxy devices—super expensive, super slow, and it has to be built atom by atom. This alone is pretty hard to do, and this is what your guys at TSMC or other fabs are really smart about—top-shelf superconductors.
Then, they configured it in a way where they made nanowires in a U-shaped configuration. They combined the nanowires with quantum dots and a microwave reader. Here’s where it gets crazy: they used a supercooled apparatus, typically a helium-3 and helium-4 diffuser, because it has to get super cold—basically to absolute zero. You freeze all of this in a chip and then use a magnetic field. If you do it just right, you can create a brand new particle that doesn’t exist in nature, called a Majorana fermion. Normally, every particle has an antiparticle, creating a duality of nature. A Majorana fermion’s antiparticle is identical to its form; it’s like a half particle.
Now, why do you do this? Because you get nonlocality for your qubits. That’s a big word, and I’ll break it down for you. Nonlocality means that both parts of the qubit have to be affected for the computation to fall apart. Even if one part gets hit, which happens all the time, the other part has to be affected too for the thing to fall apart. If I can do that for two, eventually I can do it for n pairs, and that’s called the Majorana surface code, invented around 2015.
What does this mean? It means you have a stackable way to exponentially reduce the error rate in a computation. This technique can be used to put many of these U-shaped nanowires together in a single chip. If you’re measuring these quantum dots, we’re pretty good at that. When they entangle, you can use microwaves to read the output, giving you a super accurate way of measuring state.
So, all this technobabble—what does it mean? It means that I think Microsoft has a scalable platform for a real quantum computer. They invented a brand new way of making a brand new material; it’s the very first topological superconductor. It’s a new state of matter they invented just for this. They had to create it atom by atom, which is just crazy. Then, they had to figure out how to put it into a chip and manufacture it in a very specific way so they could stack things. They also had to figure out how to get magnetic fields just right to create what are called Majorana zero modes. Basically, they create these quasi-particles that get entangled and share state.
They had to build this material in a way that it has a property called topological material, so you have these braided wave functions that resist changes. It’s almost like putting Teflon on the surface of a pan; things don’t stick to it. You prevent disruptions to the calculation. Then, they still had to figure out how to cool it and have all that work at near absolute zero. That’s what the aluminum is doing; it’s a superconductor inside the thing, meaning there’s no resistance.
They had to put all of that together and figure out how to read it, linking it with quantum dots and a microwave reader. Then, they had to write a programming language for it, which they did—Q# (Q sharp). They have an actual programming language today that you can look up and use. Finally, they had to figure out how to deploy it, and they’re going to put it in Azure. Holy moly! This is a monumental achievement for quantum computing in general.
What does this mean for Cardano? The good news is that Cardano’s research headquarters is at the University of Edinburgh, which is also a world-class university for quantum computers. Right now, the brilliant minds there, who know far more than I do about this, are going to take a look and deconstruct this. We’ll try to develop a horizon of what we think it’s going to take to productize this, if it’s even possible. Microsoft has to show some more; there has to be some reproducibility in this, although they claim the chip they’ve created already has these properties and is running at 8 qubits.
In general, we need to speed up our post-quantum hardening for Cardano. We have a group of people who think about these things, and there are different techniques like hash-based crypto, lattice-based crypto, and others that make your crypto immune or resistant to quantum computers. The reason we haven’t embraced them is that there weren’t good standards yet, and typically, these solutions are about five to ten times slower and five to ten times larger in terms of transaction size. We know how to do it; it’s just that we haven’t embraced it because we’re waiting for optimizations and standardization.
The good news is that NIST has standardized these things. We now have a suite of them, and there are a lot of new capabilities coming on. In the budget, we’re going to add a proposal for accelerating the post-quantum hardening scheme and adding new capabilities and features to Cardano. This can be done in the two to three-year horizon. I do not believe the Majorana fermion will be product-ready within two to three years. I don’t think this will create an immediate impact; it’s going to create a revolution with a new paradigm of quantum computing, and many people will start working on topological conductors. You’ll see massive improvements; the TSMCs of the world can do this far better than Microsoft can.
Fundamental physicists are going to get really curious about this, and they’re going to massively improve the design. Now that these things are real and they see them, Microsoft is certainly going to work with partners to figure out all these different mechanics. In five years, we have a problem; in ten years, we certainly have a problem. But in two to three years, we won’t. We need to model a quantum adversary, rewrite the papers from the perspective that the adversary has a quantum computer, and then work our way backward.
Post-quantum VRF, post-quantum oracles, post-quantum signatures—these are what most people refer to when they say they are quantum resistant. We also need to look at the hash algorithms and ask if we’re okay with hashes. We probably are; Grover’s algorithm is not a problem there. It’s pretty crazy when you think about it.
So, there’s a lot of post-quantum work we need to do now. Thanks to Microsoft, who made my life a little harder. On the other hand, it’s pretty magical what you can do with all this. As a final point, they recently released something called MatterGen. MatterGen is a large language model for material science. You just tell it what you want, and it figures out how to make it—how to build things atom by atom. If you pair MatterGen with a quantum computer, you can basically do miracles.
You could say, “I manufacture cars, and for my engine, I want the following properties: it melts at this point, bends in this way, and bonds with these materials.” MatterGen, paired with a quantum computer, would sit and think about that and come back to you with a recipe, telling you how to make a new material that has never existed before, which is perfect for what you’re looking for, and how to make it at scale. It’s just insane—absolutely insane. It’s literally a god creation kit when you think about it.
The same goes for biochemistry. I have a biotechnology group, and there’s this one thing called plasmids that can make any chemical you want. A company called Minicircle uses them, among many others. They’ve been around for a long time and are currently using them to make Fistatin, which is used to build muscle. It’s like the muscle pig stuff; they give you the plasmid, and you build a lot of muscle—sometimes 10 to 15 pounds.
What if I tell this model, “Hey, people have microplastics in their bodies, and I want you to invent a molecule to bond with them, and that molecule will be water-soluble”? What does that mean? It means you just pee out your microplastics, detoxing your entire body. I can then make that in a plasmid inside your body and inject it, lasting for a year as an on-off switch. The Minicircle ones do it; it’s pretty cool.
Okay, the model just tells me how to do that, and I say, “Oh, by the way, it can’t bond with anything else, so it’s chemically inert to your body chemistry and creates no side effects.” How about that? It’s just nuts. That’s what this does; it’s going to change everything. When people say, “Move fast and break things,” Microsoft, a $3 trillion company, proved that the opposite paradigm is prudent. They worked for 17 years to create a topological superconductor. Congratulations, Microsoft; you knocked this one out of the park!
1:07:35
26:05
26:19