Overview
Johannes Pollanen, co-founder of EeroQ and a professor of physics at Michigan State University, explains how EeroQ builds qubits from single electrons floating on a film of superfluid helium, and why he thinks the semiconductor chip is the only proven way to scale a quantum computer.
He covers the different kinds of quantum computer, why helium never freezes, how much helium-3 a dilution refrigerator needs, how a computation runs on a qubit, and what EeroQ’s 10,000-qubit goal means next to IBM’s plans.
This video is divided into 14 chapters as follows:
- 0:00 Quantum computing at a wedding
- 3:57 Reverse engineering quantum computing
- 5:15 Quality or quantity of qubits
- 6:23 The modalities of quantum computing
- 12:36 You’ll need liquid helium
- 19:01 Transistors and silicon
- 20:21 Is that thing behind you using helium-3?
- 26:15 How do you cool quantum systems?
- 31:08 How do you make more qubits?
- 35:15 How do you do a computation on a qubit?
- 40:05 Faster quantum computers
- 43:45 Can IBM make a useful quantum computer?
- 46:00 The quantum lessons of nature
- 48:33 Carry over question
Transcript
Quantum computing at a wedding
Mark Fielding: Disruptors and curious minds, welcome to the good ship thinking on paper. In a world increasingly vibe-coded by 20-year-olds and run by billionaires, this is where you can think for yourself about the impact of technology. Keep listening, and the algorithm can’t touch you, at least for 40 minutes. Sadly, Jeremy can’t be with us today. He’s been called away on a top secret data center mission. He signed a lot of NDAs, so the very fact I’ve mentioned this might get me disappeared. So if I’m not here next week, big AI. But we’re not talking about AI today, we’re talking about quantum computers. We’re talking about scaling quantum computers. And we’re doing this without a PhD. The idea is to combine my beginner’s mind with an expert’s mind, meet somewhere in the middle, probably near me. We’re going to be talking about trapped electrons, trapped ions, we’re going to be talking about scaling quantum computers. We’re going to be talking about the biggest challenges facing the industry and quantum mechanics at the moment. And we’re going to end with our handover question from our last guest, which is a doozy. And today’s guest is Johannes Pollanen, CSO of EeroQ. Johannes, welcome to Eric. Johannes,
Johannes Pollanen: EeroQ, yeah. Yeah.
Mark Fielding: welcome to Thinking on Paper. Thank you for thinking on paper with us. And rather than read your incredible background, I want to ask you you’re at a wedding, you’ve sat by someone you’ve never met in your life. They say, Johannes, nice to meet you. What do you do? How do you answer that question?
Johannes Pollanen: Yeah. I, that’s a good, it actually happens.
Mark Fielding: What’s your go to?
Johannes Pollanen: Yeah. My go to is you know, I start with I’m a physicist by training, but what I do for a living is try to build essentially a next generation of supercomputer using quantum physics and then follow up with like, have you ever heard of quantum computing is sort of what it starts out with. And yeah, that usually gets the ball rolling pretty quickly because people have usually heard of some amount, it’s in the zeitgeist, it’s out there and folks are thinking about it and want to know more. And yeah, that’s how I lead in and
Mark Fielding: You hope that whoever’s wedding it is, they know what your background is and they put you by somebody who can actually who’s like, not whoa. What in those situations, what is the most common go to question they ask you when you say that?
Johannes Pollanen: Yeah, I often get one of two, I guess I get a handful of similar responses. I’ll often get like, physics, that was my worst subject in high school or college. Which often then heads the conversation off into the other part of my life, which is I’m also an academic. I’m a professor of physics at Michigan State University, which is where we spun out EeroQ from. We’ve developed some of the technology there and then spun it out as a company in Chicago. But often they’ll ask me, like, yeah, I’ve heard of quantum computing. I heard that, you know, when are quantum computers gonna exist, you know? Which I always think is a fun question because actually even my academic colleagues often ask me, like, when are quantum computers gonna exist? And my answer is,
Mark Fielding: Don’t they already exist?
Johannes Pollanen: Exactly. That’s what I say to them. I say they already exist. You could get on Amazon’s Braket service and connect to one via the cloud right now if you wanted to. You know, they’re not the ultimate machines that we hope to build as a species sometime, but they exist. Like we know that the basic principles work. Yeah.
Mark Fielding: Yeah, well last week we were playing quantum games using a quantum computer with moth quantum and
Johannes Pollanen: Exactly. Yeah.
Mark Fielding: that was actually kinda cool making quantum music with quantum computers.
Johannes Pollanen: Yeah. And it totally works. And you can do even more. Not just playing games, right? You can people can simulate small scale molecules and you know, if you’d asked me that five ten years ago, like when are we gonna be actually simulating nature with small scale quantum computers, I don’t think many people would have predicted that we’d be doing that now, but we are.
Reverse engineering quantum computing
Mark Fielding: well let’s start there then. So on the EeroQ website you speak about reverse engineering quantum. And so I like reverse engineering things. So let’s really work backwards by going forwards. If you had a million qubit quantum computer for whatever that’s description is worth, what are you doing with it?
Johannes Pollanen: Yeah. I think the thing I’m most excited about is associated with chemical simulation and drug discovery. I think if you have millions or tens of millions of physical qubit machines, lots of logical qubits, then you can get to the point where problems that are intractable now in molecular simulation become very tractable. I think that’s the thing that keeps like when I get excited about it, that’s what I get excited about. You’ll also have the opportunities, you know, for cybersecurity and those kinds of things, like, you know, Shor’s algorithm kind of efforts. Those are great. I those don’t get me out of bed in the morning. I think usually it’s the things that can become life-changing, you know, for many people that I find most compelling. You know, sort of related to your first question about the, you know, what do you say to people when they ask you what you do? As I said, I’m an academic. For a long time, the work that I did in physics and in quantum science was really just for the sake of science, for trying to understand about nature. Now that we’re building technologies based on what we know, it’s the first time in my career where we can actually, you know, the things that we’re doing in the research facilities or in the academic labs or national labs in the world actually has the potential to make an impact outside of physics. You know, we can address something like, you know, curing diseases with, you know, computers that can simulate new medicines. And that I find extremely satisfying. I never was an applied person, but I understand once you realize like you can like the things that you’re doing are actually good for the world, it can be very exciting. So I think that’s the thing that I find most compelling. But I do think that for the biggest applications, you’re not just gonna get away with, you know, 100 logical qubits. You’re gonna need a lot. You’re gonna need a lot of qubits. That’s gonna be the big issue. I mean, the bigger you can make the machines, the more powerful they’re gonna be. And for those most impactful things, I think you’re gonna need big machines. Not necessarily large in space, but large in their information processing density.
Quality or quantity of qubits
Mark Fielding: So if Jeremy was here, he would ask you now on the qubit question, do you want quantity or do you want quality?
Johannes Pollanen: I think you need both. I think you really do need both. You need, you know, I can tell you a little bit more about our technology specifically, but I think the reason I like our technology is because I think we can get both. We can get high quality qubits, we can get a whole bunch of them, and we can leverage all of the most state-of-the-art advances in algorithms and error correction. And I think you need those things. I think as a community we’ve learned that none of the modalities is perfect in some sense. They all have their challenges. I think all of them can potentially be really successful, at least to some level. And but finding the one that can really like, you know, leverage sort of all of the benefits of all the different modalities, that’s the tricky part. And that’s what I really like about our system. But we can get into the details. Like I said, we have our own challenges too. Yeah. Yeah.
The modalities of quantum computing
Mark Fielding: Right. Well, let’s go back to the wedding talk. So I’m dangerous at a wedding. So modalities, I know what that means. Many people listening to Thinking on Paper will think, a quantum computer is a quantum computer. What do you mean modalities? So what are the five big modalities and where does EeroQ fit into that modality network?
Johannes Pollanen: Yeah. So by modality, I really just mean like what are you, what is the actual hardware that you’re gonna build the quantum computer out of? Because you can, you know, as you say, a quantum computer ultimately is a quantum computer. If you can program it the way a quantum computer should be programmed, it doesn’t matter what’s under the hood in some sense. Just like for a regular computer, it could be transistors, it could be punch cards, it could be vacuum tubes, it could be whatever you want under there. As long as it allows you to process information the same way, fundamentally, it doesn’t matter what the computer is.
Mark Fielding: Yeah, but a punch card and a vacuum tube aren’t gonna give you the same results, are they?
Johannes Pollanen: Yeah, I mean, they will if you can figure out how to make them big enough, but that’s the big issue, right? Like you’re not gonna make an iPhone out of punch cards or vacuum tubes, nor are you gonna make a laptop. And so we are now facing we’re sort of in the world where there’s all these different competing ways of building computers, quantum computers that are figuring out different parts of how to do it. And but we haven’t yet found, you know, the transistor of quantum computers. So the different approaches to answer your question directly, you know, there’s the superconducting circuits of IBM and Google. There are trapped ions like the ones that Honeywell and IonQ use. There’s now you’ve probably heard, I don’t know if you all have had anyone talk to you from the neutral atom community. They’re very much like ions except without the charge. Yeah.
Mark Fielding: We’ve had Infleqtion on a couple of times.
Johannes Pollanen: Right. So Infleqtion, or atomic, logical, all of these. There’s a it’s they’re the new kid on the block in some sense. And there’s electrons trapped in semiconductors, so like the things that Intel and HRL used to do, which is now part of IBM. Those they’re using the spins of electrons in semiconductors. There are folks who are working on trying to use particles of light as the qubits, like PsiQuantum and Xanadu. And then, you know, there’s us, which we’re a little bit different, but I think we combine some of the benefits of these others. So we are an electron spin based quantum computing company. We ultimately want to use an electron and its little bit of magnetism. Each electron is kind of like a little north and south pole magnet. And you can make a quantum bit out of that magnetism. And that is what we’re working toward. And I can say more about we take those electrons and instead of having them in the silicon like the semiconductor folks do, we actually just pluck the electrons out and place them above the semiconductor control chip and hover them in vacuum, kind of like an ion trap or a neutral atom system. And the way we do that is by actually coating the underlying chip with a layer of a very thin layer of superfluid helium, and then the electrons float above that surface and are bound to it in vacuum. Yeah, it’s pretty wild, but it’s kind of amazing.
Mark Fielding: Right. Right. Of course they do. Right. We’re gonna get into this in more detail. Yeah, it sounds pretty wild.
Johannes Pollanen: You know, one of the things that I think is really important about all of these quantum technologies that I think is sometimes lost. As a species, we’ve known about quantum mechanics for like over a hundred years. But it’s only really been in my lifetime that we’ve got to the point where we can take individual quantum objects, whether they’re atoms or photons or electrons or engineer circuits that behave quantum mechanically and control their properties. You know, to like fully deterministically move them make them do what we want and then measure them. But once you have that level of control, then the engineering starts. That’s when you can really build a technology. And that’s where we are and that’s the exciting part, you know. It’s kind of amazing. Quantum mechanics was what I got excited about when I started getting into physics. But now we get to just manipulate it, you know, and do what we want.
Mark Fielding: Right. Right. You mentioned the magnetism and of the electron. Before we get I wanna go over everything you said again. Before we get to that, can you speak about magnetism, why you said the magnetism of the electron, does that matter? And then the role perhaps of magnetism beyond that in the quantum realm?
Johannes Pollanen: So for us, like you know, you imagine an electron and it’s a little charge, a little negatively charged object, but the electron has a property, a purely quantum mechanical property called spin. But spin is just a word. What it really means, what it really is physically is there’s a magnetic moment, a little north and south pole associated with you can think of each electron as having its own little bar magnet that it carries around with it. And whether those north and south poles point up and down, or whether you create superpositions of that magnetism, you know, in combinations of up and down, you can encode quantum information in that. And that’s actually the thing that, you know, the folks like Intel and HRL and the various startups, Quantum Motion, all of these companies that work on electron-based qubits, they’re utilizing that magnetism of the electron as their quantum bit. And we do the same thing.
Mark Fielding: When you said imagine each one carrying a little bar magnet around. Is that the right way to think about that? Because they’re not really spinning.
Johannes Pollanen: They’re not spinning, but there is truly a magnetism associated. You can measure, you know, there’s a magnetic interaction just like there would be for two bar magnets. It’s like the electrodynamics of that is the same. It really does boil down to that. But the spin is sort of a bad word in the sense that spin really makes it seem like there is something spinning, which, you know, that’s a very classical way of viewing the world. But ultimately we don’t. You know, sometimes people get hung up on the like we end up using language to describe things that we don’t really have intuition for. So we use words from the classical world to try to describe the quantum world. And there’s not really necessarily a one-to-one mapping on that. We understand how they work in terms of like what we measure and what we do and how we describe it mathematically, but the language often ends up being what do you say, insufficient to fully capture the underlying phenomena.
You’ll need liquid helium
Mark Fielding: I wanna build the stack. I want to understand the stack here. And I stack might not be the right word, but the ingredients, the ingredients is a better word. I want to understand the ingredients here. So you mentioned the electron, you mentioned helium. Liquid helium.
Johannes Pollanen: Yeah, liquid helium.
Mark Fielding: Had a look at liquid helium. There’s some pretty gnarly quotes about liquid helium. Is liquid helium the physical state of the element, helium at extremely low temperatures, specifically when it is cool to about minus 269 degrees, making it the coldest liquid on earth. There was another one which was what was the other. There’s something about being called the cleanest state which what’s that all about?
Johannes Pollanen: Yeah. So helium at so if you take we have one of our dilution refrigerators back here that goes down to, you know, very close to absolute zero. It goes to like ten millikelvin in temperature. So we build our quantum chips ultimately live inside of machines like that. And very much like what IBM and Google do for their circuits. And at those temperatures, as you say, liquid helium, helium is a gas at room temperature. But if you cool it down, like any gas, it will eventually condense and form a liquid. And helium, you just have to get very, very cold. Forms of liquid. But interestingly, helium has a very funny property. Normally, if you take a liquid and you keep cooling it, eventually it freezes. That is not true for helium. It’s true for everything, every other element known, but helium remains a perpetual liquid all the way down as cold as people have ever gotten. And there’s good reason to believe
Mark Fielding: The only one.
Johannes Pollanen: The only one, there’s one example. And it’s a purely quantum mechanical effect, actually. Because as you cool these things down, the helium atoms actually become little waves. They become quantum mechanical probability waves. And those waves overlap with one another to the extent that the system can never crystallize. And it forms instead a perpetual liquid. And it actually forms a very funny liquid called a superfluid, which flows without any viscosity.
Mark Fielding: This is where my silly questions come in and I just have to ask. Why does helium behave like that and nothing else does though? Even though you might have just answered that question.
Johannes Pollanen: Yeah. There’s two I didn’t really I alluded to it, but I didn’t totally answer it. There’s two reasons. One is the helium’s crazy light. So helium other than hydrogen is the lightest of all the atoms. And so that means that when you try to you’ve probably heard of quantum zero point motion. So if you try to confine a quantum particle, it jitters around more and more. That happens more if the underlying atom is very light. So helium has a very large quantum zero point motion. So when you make it a liquid, you’re trying to essentially confine it to a single place, you know, next to its neighbors, but it doesn’t like that quantum mechanically, and so it jitters around. And so that’s one aspect that keeps it a liquid. The other one is if you bring two helium atoms close to each other, they don’t really interact very strongly with each other. They’re kind of, you know, you can’t do chemistry with helium. All of the electronic shells, if you remember from high school chemistry, the electrons on helium are completely filled up in their electronic shells. They can’t interact with one another or form bonds. And normally you need those kinds of bonds and connections to form a solid. And helium can’t do that. And it’s those two properties. It’s chemically inert and it’s very light that allow it to be a liquid forever.
Mark Fielding: Okay, and for those people who are listening not watching this, A, go to YouTube and watch it or click that button on Spotify to make it video or behind Johannes, that is happening in the chambers that we can see.
Johannes Pollanen: You have like super microphone photos. Yeah. I was just gonna say this property of superfluidity isn’t new, you know. We’ve known about it like since 1904, I think was when helium was liquefied and probably the first demonstrations of superfluidity were achieved. So it’s a state of matter that’s quite exotic and strange, but we’ve known about it as a you know, like we understand that system quite well. But there was a second part to your question, which was the cleanliness.
Mark Fielding: Yes.
Johannes Pollanen: So if let’s say that you wanna have an electron and you want it to be a qubit, so you wanna shield it from all kinds of noise and decoherence and sources of disturbance. So if you have, say, a material substrate onto which you wanna put the electron or wanna put the electron into that substrate, those things have chemical impurities and crystal defects and all kinds of vibrations and all kinds of stuff that’s terrible. So what you would really like to do is take that electron and you want to hide it away from the rest of the universe, you put it in vacuum very far away from everything. But you need to somehow hold it in vacuum. Holding an electron in vacuum is actually not so easy. But if you bring it near the surface of liquid helium, it actually naturally gets stuck about 10 nanometers above the helium surface and floats there. And then helium at those temperatures is liquid. Everything else is frozen. There’s no other atom around except helium. There’s no magnetism, there’s no charge noise, there’s none of these other things that plague other qubit platforms in the solid state. And so in that sense, it is the cleanest system for hosting one of these electronic qubits. And that’s one of the reasons we like this system. So very much the other natural qubits benefit from this as well. So for example, atoms and ions are hidden away in vacuum and then you interrogate them with lasers. And the reason that they’re so high coherence and so perfect is because they’re in vacuum very, very far away from other things.
Mark Fielding: So no lasers for you.
Transistors and silicon
Johannes Pollanen: No, yes, no lasers whatsoever. We use so actually I’ll mention the other thing that really we love about this technology is that so the other important part is that we you know, you asked this question earlier about how do you what do you need, quality or quantity? And I said both. So we kind of addressed the quality issue just now.
Mark Fielding: Yeah.
Johannes Pollanen: But the quantity issue is very important. And you have to we built quantum computers out of a variety of things and but we’ve never figured out how to build a lot of quantum bits onto a platform. In fact, in all of human history, we’ve never figured out how to make lots of anything on the computational side, except for transistors in silicon. So like the things that give you iPhones and laptops and the best supercomputers and GPUs, that is built on a semiconductor manufacturing process that only exists. We’ve only ever invented it once in all of human history. And so what we do is we actually build the underlying substrate beneath the helium out of that semiconductor material, the same way that chip builders build iPhone chips and Intel builds their quantum computers, their quantum chips, except we then combine it with this pristine helium surface with electron qubits trapped above. And I have, you know, you asked about show and tell earlier, I have one of our chips here. In the very center of this board is something about, you know, three by five millimeter chip and that can house, you know, thousands or tens of thousands of electrons. And being able to make it tiny like that is really important. You want to jam as many qubits onto a single monolithic chip as possible. And that is sort of our philosophy. You gotta get quality and quantity in as small a footprint as possible.
Is that thing behind you using helium-3?
Mark Fielding: Right. I want to lead up to that. I want to know how many qubits you can put on that and how many there are. First of all, just another helium question because at Thinking on Paper, we connect all technologies with tech, agnostic, we speak about all things. Not so long ago, we had a former NASA scientist who wants to go to the moon to mine helium three. And when we were speaking to him, one of the reasons he said that we need to go to the moon to mine helium three is because the quantum industry uses it for their cooling. Is that thing behind you using helium three?
Johannes Pollanen: Yep. Yeah, this machine uses helium three. The helium that I’m talking about for our chips is helium four. It’s the more standard one. But there is all of these behind our director of operations here is that dilution refrigerator and that thing has helium three in it. Like these cooling systems have that. You know, in an ideal world when we that system has something like 20 liters of helium three gas. And each dilution refrigerator needs that amount and you never lose it as long as you, you know, their safety is in place to kind of recollect it. Yeah, it keeps it.
Mark Fielding: Okay, it circulates around. Okay.
Johannes Pollanen: Yeah. And but it is still, you know, to make more of these, helium three is a resource that isn’t very naturally occurring. It is not so to give you a sense, you know, helium isn’t around on Earth, that’s about a part per million out of the air, because most of it when it’s in the air escapes out into outer space. It’s actually so light that it can get enough speed to get itself out of the atmosphere. So there’s about a part per million of the air is helium and about a part per million of that is helium three. So it’s a tiny fraction. And finding new sources for helium three is really important. It is extremely rare.
Mark Fielding: Very rare isotope. Well, we know there’s a show. I’ll put it in the link for our show on Helium Three because it’s all about nuclear weapons as well. So it’s a very interesting. Yeah, yeah. It’s
Johannes Pollanen: Yeah, when you said it went That’s right. Yeah, there’s an interesting history there on how low temperature physics intersected with some other innovations that we made as a species.
Mark Fielding: Right. Could you do the show and tell again? Could you hold up the chip to the camera again, Johannes?
Johannes Pollanen: Yeah, absolutely. Yeah. So this is our Wonder Lake chip. Am I kidding?
Mark Fielding: Wonderlake.
Johannes Pollanen: Yeah, we named all of our chips after Lakes. Wonder Lake is a lake nearby in Chicago that we named this one after. And this one was prototyped. We design the chip and then it gets manufactured by a commercial semiconductor foundry using the standard processes that are made to like I said, make the chips in your iPhone or in your laptop. And this device was this Wonder Lake architecture was built so that we could demonstrate that we can move electrons around to do error correcting codes. So like the neutral atoms or ions, we can move our qubits around. And this was built so that we could have something like 2500 electrons on that chip.
Mark Fielding: Does twenty five hundred electrons equate to twenty five hundred qubits?
Johannes Pollanen: No, so actually our qubits ultimately will be so we can get into the details of various things, but ultimately our qubits are built from two electron objects. So we make what are called singlet triplet qubits out of two electrons. And that harkens to some of the things that the silicon spin people do, like us. We share similarity with them. And so it equates to something like, you know, half that many qubits. We don’t have thousands of qubits on this chip. This was really just designed to show that we can do the choreograph the error correction that we’ll eventually need to do. But to give you a sense our ultimate goal is to build chips that are not that much bigger than that, than the thing I just showed you and to fit, you know, millions or tens of millions of electrons and tens of qubits ultimately qubits on that kind of footprint.
Mark Fielding: So this is how we scale. This is the famous this is how we scale quantum computing.
Johannes Pollanen: This is our thesis for sure. Because again, I think, if you’re gonna set out to scale quantum computers, you have, I think, two options. You’re limited to two choices. One, use the one scaling architecture that we have ever built in all of human history, which is CMOS manufacturing and semiconductors, or you have to invent something equivalent. And the second of those sounds really hard to me, given that all of humanity has invented this once. And so I’m picking that particular battle. Like we have work to do on the quantum front and on the error correction front and all this stuff, but I think that’s a more tractable problem than inventing some other technology that we can scale to the level of, you know, ten to the ten transistors that we can do in your laptop.
How do you cool quantum systems?
Mark Fielding: Right. So we have the wafer, we have the liquid helium, we have the two electrons that are floating on that.
Johannes Pollanen: They’re floating. They’re bound. It’s actually quite so the electron is to give you a sense of why, you know, like why on earth would it get stuck there? So if you have this layer of liquid and you bring an electron nearby, the helium underneath polarizes because of the electron charge. So the helium is a dielectric. So if you bring a charge near a dielectric, it polarizes a little positive charge on the other side. And so then this electron sees a positive image charge in the liquid and gets stuck to it. It sees it and wants to go toward you know, if you have a negative charge and a positive charge, they try to get close to each other. And so the electron creates its own like fictitious positive image charge in the liquid and tries to get to it. But it can’t go in because the helium is completely inert. There are no open electronic states, so it just gets and those two things compete and it ends up finding itself, you know, 10 nanometers above the helium surface. And this is yeah.
Mark Fielding: Just gonna pull on another thread there because we spoke about dielectric fluids a few weeks ago and we were speaking about cooling AI data centers. So there’s another thinking on paper thread for everyone to pull on. But sorry, continue.
Johannes Pollanen: No, that’s a good point, actually. One of the things that we spend a lot of time one of the questions that comes up in some quantum technologies a little bit in the weeds, but how does one get heat out of the chips? You know, there can be power dissipation problems on the chip that one has to solve, particularly at scale. So we think a lot about these kinds of problems. And the dielectric helium actually is the it has at those temperatures, you know, very near absolute zero, it has the highest thermal conductivity of any known liquid, any known substance, correctly. So it pulls heat out very efficiently. So we benefit naturally from that aspect. But what was I saying? But yeah, that’s why the other thing I wanted to emphasize is all of this sounds crazy exotic. Actually, if you talk to anybody about their quantum technologies, it always sounds exotic. But all of these things have been known for a long time. So like I think the first trapping experiments of electrons above the surface of helium happened well before I was born. They happened in the boy, late sixties, early seventies.
Mark Fielding: Yeah, okay.
Johannes Pollanen: You know? And people have been building up from there. You know, it’s and it started out as just
Mark Fielding: Was that Bell Labs or something?
Johannes Pollanen: Yeah, the first experiments were done at Bell Labs and some universities. But yeah, it was it’s pretty amazing that, you know, you build up from these Bell Labs at the time had a real knack for laying the foundation for all kinds of technologies. I don’t think none of those ideas that they originally had were intended to be technologies at the time. They were really just like interesting aspects of nature that they wanted to realize and then people have built off of that, us.
Mark Fielding: You have the wafer, you have the helium, you have the electrons. At the moment, how many qubits do you have? Right. So why why is that challenge? Why are you stuck there or not? Why are you there? And why can’t one just be ten? Why can’t ten just be a hundred? And why can’t a hundred just be a thousand?
Johannes Pollanen: One. So this is now you’re getting to our challenge. Yeah. Yeah. So this idea of making qubits out of electrons on helium has been around for a while. It was actually the first proposals were made back in the nineties. But then nobody really it’s an artifact of history in the sense that, you know, people back in the nineties didn’t actually even like the first ion trap experiments were really going on then. Nobody really invented superconducting qubits or the techniques for measuring them. Silicon spin qubits didn’t exist, neutral atoms definitely didn’t exist. And so you had this community of people that had been working on electrons on helium, and then there were other communities working on quantum computing, and they just didn’t talk to each other. And then this sort of started to change near the end of my PhD. Our colleague Steve Lyon, who’s an electrical engineer at Princeton and the chief technical officer here at EeroQ, had this idea of using electron spins for quantum computing. But again, nobody was really trying to make it work, honestly. The community was doing other things. And then some academics started saying, you know what, we could use some of the techniques that people develop for superconducting qubits to do this kind of readout. But they mostly worked on it in academic environments, kind of one graduate student at a time, you know, and things go slow in those kinds of environments. But you kind of had this ecosystem building around quantum computing generally. And so when we came onto the stage, when I started my faculty position, my background before coming to MSU and before EeroQ was I was at the Quantum Institute of the Institute for Quantum Information and Matter at Caltech. And that’s where I really cut my teeth on quantum computing and thinking about like what systems could be really good. And I felt like this one had a lot of the benefits. Steve and I had talked for years about, you know, how to scale these kinds of systems. And I thought, you know what, this actually needs a real effort. Somebody needs to try. And so what we did is we set out to try and we developed essentially it took us about 18 months to build the first qubits, the first charge qubits out of electrons on helium and get that working. And now we’re developing the spin qubit part. And it just needed somebody to try and to utilize the techniques that people had built up over years in other parts of quantum computing. And that’s really what we’re leveraging to catch up. But this gets at the heart of, you know, I said everybody has their challenges. I’d say we’re a little earlier than most. Not everybody, but we’re a little earlier than most on the quantum part. But we’re way ahead on the scaling part. I feel very confident in that part. I don’t think many of our colleagues have ideas about how to really build large-scale systems and we’re catching up. I mean, that’s the sort of two-minute version of the history. But it’s an interesting random walk through history and science and technology development that has led us to where we are and why it’s taken long. Yeah.
How do you make more qubits?
Mark Fielding: But what does it take to go from one to ten to a hundred to a thousand though?
Johannes Pollanen: Yeah. Good question. So what we’re doing, what we do, the way we do our device development is we build the quantum devices in-house and test them here. So we when I say in-house, we use our nanofabrication engineers use academic clean rooms to build prototype devices that we then rinse and repeat and test. And things work sometimes, sometimes they don’t, but we have a very clear roadmap on how to go that. We have a recent the paper that we published recently that demonstrated this first coupling to a charge qubit of an electron on helium is in nature physics from earlier this year. And it’s just the rinse and repeat process until we get the materials right to do the readout and get everything working appropriately. And, you know, I’d say that we’re working faster than an academic environment ever could. You know, like it took us about 18 months to get this charge qubit readout working. That idea had been around for more than 10 years before that, you know, and we just jumped on it. Instead of having one person work on it, we had, you know, a fab team, a design team, a theory team, and a measurement team, and it cracks in 18 months. And we’re doing the next step is to build these two electron spin qubits, measure their fidelities, get all the two qubit gates working. So that’s the process. But in parallel, we work with the semiconductor foundries to build these chips. So we’re not leaving scaling as an afterthought. We’re doing both of these things, you know, full speed ahead on that. And then actually, so we I used to say we build the quantum devices in-house, we build the scaling architectures with the semiconductor foundries, and eventually we have to start integrating. Well, actually now we’re at the integration stage. So we’re taking our single qubit kind of devices and starting to integrate them with the stuff that comes from this commercial foundries.
Mark Fielding: So do you have a timeline? Do or no timelines are impossible, aren’t they?
Johannes Pollanen: God. Yeah, they are. I mean, predicting the future is always strange. But I think there’s some interesting I always think it’s gonna be faster than people I mean, it’s been faster than I would have expected with quantum computing. You know, I started this earlier, I said like if you’d asked me five or ten years ago, would we be simulating small scale molecules on quantum computers that are connected to the cloud? I would have like, that’s crazy talk. You’re just speaking like it’s nonsense. And that is exactly where we are now. We take it for granted, but if you had a time machine and could go back five years, people would have been like, What are you talking about? That doesn’t really work that way. A really good yeah, I mean our goal is to have in the next couple years a ten thousand true qubit system built on a chip like this, that we can connect to the cloud or sell to people on premise. That machine I will mention is still even at 10,000 qubits, it’s essentially a NISC machine. So it’s one that you can do little error correcting experiments on, but it’s not the full-scale machine that we need to attack the biggest problems in the world. That needs, I think, millions or tens of millions of qubits. And then we go from the 10,000 to that million. So we’re working on the 10,000 qubit chip now. Because I think we’ve got these pieces worked out and we’ve figured out from these other systems and how things work. We figured out what works and what doesn’t, and now we’re just putting it together.
How do you do a computation on a qubit?
Mark Fielding: When I finish recording this, I’m gonna go see some friends and family. And I’m gonna say, I’ve just recorded a podcast on quantum computing with EeroQ and Johannes was he’s a quantum physicist. And they’re gonna say, Yeah, but how does a quantum computer work, Mark? What is it? And we have these notions that we can understand gates, ones and zeros, ons and off. It doesn’t matter if there’s a trillion of them, okay, we can still kind of see that and understand it conceptually from a very high level. When people talk about doing computation on qubits, then my brain stops working and I can’t visualize how that works. Like how do you do a computation on a qubit?
Johannes Pollanen: So I mean there’s two kind of answers, right? There’s the abstract answer, which is what you already alluded to. You run gates. You know, you start with an input string of information that you know you have some number of qubits, you encode a certain sequence of zeros and ones in that. Then you let it evolve in time where you get qubits to interact with each other and you run these gates in a very abstract way. And then at the end and these that’s right, they’re quantum gates.
Mark Fielding: These are quantum gates, not the regular gates that we are familiar with, and they’re not the same.
Johannes Pollanen: They’re not the same. And depending on I once we go from abstract to physical, I’ll get to that point.
Mark Fielding: Okay, sorry.
Johannes Pollanen: No, but that’s a good point. But it is quantum gates, and then at the very end you perform a measurement of all those qubits again, and that gives you a data point for your answer. And then you have to repeat that. So in that sequence of gates that you run, the quantum gates, you’ve encoded an algorithm. There’s a process for moving information around. And then at the end, the measurement tells you what the final answer is. And that is what every quantum algorithm looks like. And there may be other pieces where you’re doing error correction in the middle, but that’s what the algorithm looks like. And then you have to somehow, that’s the abstract level. And then for a given modality, a given kind of hardware, you have to realize how do you make that happen in reality on a chip, you know? And for us, it involves moving electrons close to one another, far apart, close to other close to wires that are on the chip to cause single electrons to change their state, and then to move them around to readout zones to do the readout. So we’re essentially our algorithms look like shuttling electrons around. Actually a lot like the ion traps or neutral atoms. And instead of lasers for readout, we use on-chip circuits for the readout. Does that help at all? Or does it make it worse?
Mark Fielding: Well it probably makes it worse before it makes it better, but like I said in the green room, we do these shows that one day we speak to enough people, it will click and it will become clear to everyone who’s listening. I’m sure some people will know exactly what you’re talking about. Some people are like, What? Briefly quantum algorithms, are they agnostic to the modality?
Johannes Pollanen: Not entirely. So for example, error correction is a type of algorithm. So if you try to do if you’re trying to fix errors that happen in a quantum computer, that is an algorithm that you’re running. It can be mapped abstractly onto a series of gates and measurements, and that is a process that is an algorithm. So some algorithms, like some error correcting algorithms like the surface code, which is the most famous of all the quantum computing error correcting algorithms, was developed with the qubits being connected to their nearest neighbors. So that’s a level of connectivity to the circuit. You have qubits are connected to the ones that are the left, the right, just in front of them and behind them on the circuit. And that allows you to do the surface code. There’s a new class of error correcting algorithms called quantum LDPC codes that require a different kind of connectivity. So you need to be able to like take a qubit that’s over here and connect it to a qubit over here somehow. That means you could either do information swapping between a whole chain of qubits, or you can actually move the qubits around to get them to talk to each other. But those two algorithms are like the surface code and these quantum LDPC codes. They are not agnostic to the underlying hardware. Because if your hardware can’t be reconfigured, if you can’t move your qubits, it limits what you’re able to do algorithmically.
Mark Fielding: That makes a lot of sense to me, actually. Like I can visualize that, so that does help.
Johannes Pollanen: Yeah. So it’s not and that’s true. I picked error correction as an example, but similar things could be true for other algorithms, right? Like depending on how the circuit is actually architected and laid out and how the connectivity is between qubits and what operations are easy and hard, then that determines like how easily that algorithm can be run on one machine or another. In principle, it could always be run you can come up with some way to do it, it just won’t be it will be wildly not ideal to run on a particular kind of modality. And a big one is connectivity, this issue of how well do you connect. This is where ions, atoms, and us really shine. That you can move the qubits around.
Faster quantum computers
Mark Fielding: I literally sit here and ask you questions all day so I won’t. I want to move on to some bigger questions and then I’ll carry over question aware of time. Before I do that though, somebody who’s listening to this who knows far more than I and they’re like I tuned in to hear about EeroQ what have I missed for those people who know much more about the industry? What would they want to hear from you?
Johannes Pollanen: It’s a good question. I think I’ve mostly touched on one of the more subtle points that we it’s further down in our roadmap and something that is a little bit more of a trade secret on our side, but I can allude to it because people might be interested, is this electron motion that we have allows us to make two qubit and with some other pieces, single qubit rotations that allows us to do all of quantum computing. There’s a questions of speed. How fast can you run the machines? And so our 10,000 qubit chips, which we’re working on building right now, are relatively slow. They’re similar to neutral atoms or ions, but we actually have some trade secrets that allow us to get to much faster gates ultimately that we’re actually starting to build now, the first prototypes of. And so we ultimately something that I swept under the rug and we didn’t really talk about is you need Quality, quantity, and ideally you want speed. You want to be able to make the computations happen quickly. I mean that’s sort of intuitive, right? It’s hard to imagine having a slow, small quantum computer. You want something that’s you know lots of qubits on a small footprint going really fast with high quality qubits. That’s what we tried, that’s what we want, and that’s what we’re building. At least that’s our vision on this.
Mark Fielding: What is the biggest unanswered question in quantum mechanics or quantum computer? I don’t even know if I should differentiate between those terms today.
Johannes Pollanen: I think yeah, I think we understand quantum mechanics pretty well. I mean, like I said, I there’s not a lot of I mean, so okay, one of the issues with quantum mechanics that I think people get hung up on is it makes no sense, right? Intuitively. Like there’s a counterintuition between our everyday lives and when you hear about superposition and entanglement and all these things. But we know those things are right. We know superposition works, we know quantum entanglement works, We just don’t have a link it’s just unfortunate that our human brains don’t have a language that allows us to describe that other than mathematics and experiment and like actually making technologies
Mark Fielding: Poetry.
Johannes Pollanen: Poetry, you know, that’s a great actually leaning on analogy goes a long way. It builds intuitions and But it does also bring about this mystique and this mystery. It’s not mysterious. It’s really, we understand quantum mechanics. I mean, we understand it so much that we can build technologies out of it. I mean that you if you do if there’s like big unknown things about something, it’s really hard to build. If we didn’t fundamentally understand something about how transistors work, we couldn’t make an iPhone, you know? And
Mark Fielding: So what’s the biggest unanswered question challenge in quantum computing today?
Johannes Pollanen: How to build a big system. I think this is gonna be the real issue is like how to build a really big system. How do you build a large scale? Because I you know, we started this conversation with like quantum computers exist already. You can play with small toy versions of quantum computers in a variety of modalities. But when do we get to the point where we actually build a machine that’s big enough to do something useful? That’s what I want. I wanna build that machine.
Can IBM make a useful quantum computer?
Mark Fielding: I don’t know what they’re up to now, but just briefly on IBM, over a hundred, I think they’ve spoken about a hundred, their Starling chip that we’re running out of time for that. We’ve got three years left, I think, until that comes to fruition. Do you think are you saying that they’re not gonna hit their goals? Do you think they’ll hit their goals? And if they do hit their goals, do you think that actually is a useful quantum computer?
Johannes Pollanen: I my suspicion is is that I’m, you know, the large scale quantum computers that are built out of I won’t single out IBM, but superconducting qubits are they’re really great because they’re fast. You can print them out on a circuit, you know, you can fabricate them, so you can make them wherever you want. You can change the connectivity of the circuit by making different designs. Each qubit is a little bit different, and so that’s a big problem for superconducting circuits. And the other problem is that they’re quite large. You know, they have they typically each you know, I showed you this millimeter scale chip that fits thousands of electrons or tens of thousands. And we double the size of this and we can fit millions or tens of millions. Each one of the superconducting qubits essentially has a millimeter size length scale built into it. Each qubit does. So that’s tricky. I think that that’s a hard challenge to battle. I mean, I think what their solution is is quite brute force to build big systems with lots of dilution refrigerators network together. I’ll be interested to see how that evolves. I think there’s a lot of really hard engineering challenges that make I don’t know how to hook up a bunch of dilution refrigerators together that, you know, my background is in really hardcore low temperature physics, and that sounds like a hard problem to me. I’d like to build everything as small as possible into a with as many qubits as densely packed using something that we understand in one dilution refrigerator. That’s my vision. And maybe we need two or three, but I don’t want to have to build thousands or tens of thousands of these. So I don’t know. I’m hopeful. Like it, but I do want to, I’m a everybody always gives me a little bit of a hard time because I’m an internal optimist. So I think all of these technologies are gonna continue to be successful for a while. Because we’ve still got a long, a long way to go. There’s all, you know, you read all this hype, but there’s
Mark Fielding: We’ve barely started by the sounds of it.
Johannes Pollanen: Yeah, I mean, there’s so much good work that still needs to be done for ions or neutral atoms or superconducting or us or anybody. You really have to have there’s a lot of good stuff that needs to still happen. But the potential makes it worth it, in my opinion.
The quantum lessons of nature
Mark Fielding: Right, I’m gonna bring Jeremy back on, who’s not here if you’ve just joined us. Jeremy has been stolen by big tech. And he would talk about Richard Feynman now, and he would talk about nature and technology and technology in nature, and he would ask you how do we combine these two things? What have you learned from nature? And what have you learnt about nature?
Johannes Pollanen: I’ve learned, I think, that we have to move in concert with nature. So we have to nature provides us things like electrons or atoms or ions that are identical, that are pre-made by the universe to be perfect. There is no other perfection, I think, in the world, but like It’s amazing that every atom in the universe is created the same. I mean, modulo isotopes and things. But if you have two of the same isotopes, you can’t distinguish these particles from one another. You can’t put a label on there, right? They’re as perfectly matched as possible. That’s, I think, that we can’t do that in our own technologies. We don’t know how to make things perfectly. Even the best transistors that we make, each one is a little, they can’t be identical. And so we have to figure out how to take the things that we know how to make to the best of our ability and interface them with the perfection of nature. That’s sort of, I think, a real powerful. That’s what I’ve that’s one of the things I’ve certainly learned from nature. The other thing I’ve learned is quantum entanglement is truly complex. It is really, it’s ultimately the many particle entanglement that gives quantum computing its power. The fact that these systems can be They require so much information to understand a many body or many particle entangled state. That’s a true statement. And the flip side is if you have the ability to control that in a quantum computer with many qubits that are all entangled with each other, then you have immense computational power. Those are two sides of the same coin.
Carry over question
Mark Fielding: Now there should be some cut to the music. We should have some cool funky music because we have our thinking on paper handover question where one guest leaves a question for the next guest and then you leave a question for the next guest. Last week’s guest was Jeff Chang, and Jeff Chang is a writer and musician. He wrote the book, Water mirror echo, like the history of Bruce Lee in Asian America. He wrote another book called Can’t Stop, Won’t Stop, which is the history of hip hop. And he, the producer of DJ Shadow, Incredible Man, we spoke about AI. His question he left for you was If not for humans, then what is technology for?
Johannes Pollanen: That is a great question. If yeah. I mean it sort of is for humans, but he wants something else. So Yeah
Mark Fielding: I’m not sure what he wants, like yeah, cause maybe if he was kinda it was he was riffing on it and he I mean I don’t know if his real question is like, what is technology for? Well you can take it wherever you want.
Johannes Pollanen: Yeah. I mean it yeah. It’s a great question. I think you know, when I think of what is technology for, I think it’s my feeling is it’s try to at least so one of the things that I find very satisfying about physics, but more generally about my interaction with the universe is that I like to figure stuff out. And in some sense, it’s I like to figure out what’s true, which statements, which things are true, which things are false, and how do we take the true things and learn more about the world? When we’ve done that, we’ve done amazing things. And technology helps us, it’s a grounding on how to solve problems and figure out what’s true. So I think that’s an important piece. And it connects with you know, human flourishing and well-being. I think, you know, I think that’s another really important thing. I mean, there aren’t very many things that I you know, there’s often you can argue about values and traditions and approaches, but you’d love technology to help you improve the lot of humanity. You know, I think everybody can agree that we want people to be well and to flourish and whatever that means, you know, there are good answers to those kinds of and where technology can help that, I think it’s really important. I don’t know if that’s what he was looking for, but that’s what comes to mind immediately for me.
Mark Fielding: I will share it with him, I’m sure that’s what he was looking for. And you are working with a technology that has perhaps the ultimate potential to make that a reality one day. So no pressure, Johannes. No pressure.
Johannes Pollanen: No, we the pressure is the pressure’s there, the pressure’s real, but the pressure is good. Because if you’re not pressured, what are we doing?
Mark Fielding: Do you have a question that you could ask for our next guest? It doesn’t have to be about technology. It can be about anything you want. Maybe something’s gnawing away at you at the moment and you’d like somebody else to have a go at, or maybe it’s something deep, profound. It can be something frivolous.
Johannes Pollanen: Yeah. I did have a question, and it’s associated with AI because we interface with AI a lot here, in our work, but it’s also surprised me I’m no expert in AI for the record. I don’t know how to I’m but I’m a user of AI. And one thing I’m always really curious about when I interact with anybody, the question I have for them is how has you know, there’s all these promises of how AI is going to be great for everything. But my question is always, how has AI changed your life already? What’s an example? Not something generic, something very specific. Like, I feel like people find these little solutions or little uses of AI. I found them myself. I have a few examples if you want to hear. But they’re really fantastic ways of it, it’s made my life better in so many ways already. And I’m a little bit surprised that it’s been able to do that so quickly.
Mark Fielding: Share your examples, please.
Johannes Pollanen: Yeah. So so one, for example, I at some point I’ve been using it as a I really love distance running. So I’ve been using it as a coach for my running. It’s an amazing coach. It’s incredible. It can build workout routines, it can do all this thing, and it’s like super easy. And another one is I I think I mentioned earlier I’m originally from Finland. So I’m a native when I was born in Finland, we moved to the States when I was a little kid. And when my parents were still alive, I essentially only spoke Finnish to them. We spoke Finnish at home, and that was the driver. And I don’t get a chance to speak Finnish anybody until I realize ChatGPT is an amazing conversation partner in Finnish. So if you want to learn how to speak a foreign language, ChatGPT or Claude can just talk to you. It’s incredible. It’s like talking to, it’s like one of the most amazing hacks I’ve ever discovered. And it’s brought back my language for me. It’s like I suddenly dream in Finnish again because I do this every morning, just chat with ChatGPT about in Finnish about like the news or some, you know, some something totally mundane.
Mark Fielding: Right, Johannes. Thinking on paper, our tagline is well, we start every show with Disruptors and Curious Minds. Welcome to Thinking on Paper. It’s the human story of technology. How do you could you say that in Finnish on to the camera?
Johannes Pollanen: boi. Yeah that’s right. Except a bunch of Finnish people are gonna listen. Yksi asia, minkä mä voin sanoa, että ChatGPT on oikein hyvä, jos sä haluut oppia puhua suomen kieltä. If you want to learn how to speak Finnish, you should chat with ChatGPT. Kiitos.
Mark Fielding: How do you say thank you in Finnish? Kitos.
Johannes Pollanen: CSO of EeroQ, thank you for a wonderful (Wait, Mark Fielding should be saying this, but it was split. Correcting for formatting consistency).
Mark Fielding: CSO of EeroQ, thank you for a wonderful conversation. Thank you for thinking on paper with us. If you’re listening to this, please comment with questions. We have lots more quantum computing episodes lined up. But we want to hear your questions because your questions will help our questions.xyz and please share this with one quantum curious friend. Because you must have one. Until next time. Thank you and stay disruptive. Be curious. Keep thinking on paper.
Johannes Pollanen: Thank you.
In GQI Portal
The players behind the news
The team that writes QCR tracks every company, deal and technology in the GQI Factory, GQI's verified database of the quantum industry. Next up: every story linked to its players, coming to QCR's paid plans.
- Players Companies and institutions across the quantum industry, by segment.
- Scorecards How the players compare on hardware, software, funding and more.
Newsletter
QCR Alerts in your inbox
The latest reporting and analysis from Quantum Computing Report. Free, unsubscribe any time.