Xanadu Interview With CEO Christian Weedbrook

Frequently Asked Questions

FAQ

• Where is the transcript for this webinar?

   

  Speaker 1:
  Quantum computing is rapidly transforming our understanding of computational possibilities. We’re fortunate to have Christian Weedbrook, CEO of Xanadu, who will share his unique insights on how photonics-based quantum technologies are pushing the boundaries of what’s computationally possible. We’ll explore Xanadu’s mission, their innovative approach, and get an insider’s perspective on the future of quantum computing directly from a leading innovator in the field.

  Speaker 2:
  Hi everyone. Hope you can hear me well. My name is Naren Ramaswamy. I’m part of the Alumni Ventures investment team. I work with Spike Ventures, which is the Stanford fund within Alumni Ventures, as well as the Deep Tech Fund. I’m a Senior Principal on the team. I am very glad to have a CEO of our portfolio company, Xanadu, here with us. I’ll just share a brief background on Christian, and the format of this event is just Q&A with Christian for the first 15–20 minutes, followed by questions from the audience. And then after that, you’ll have the chance to—Christian is generous enough to stick around for a little bit of time—so you’ll have the chance to ask him your own questions after that as well.

  So Christian Weedbrook is the founder and CEO of Xanadu, a leading quantum computing company focused on photonic quantum computing. Alumni Ventures was fortunate to participate in the last round of the company, which was the Series C, alongside investors like Georgian Partners, Bessemer, and others, which valued the company at over a billion dollars.

  A little bit on Christian’s background: He holds a PhD in Quantum Information Theory from the University of Queensland, Australia. He also did postdoctoral research at MIT and the University of Toronto. He founded Xanadu in 2016 to build quantum computers that are useful and available to people everywhere. Since then, the company has achieved significant milestones which excited us as investors. In 2022, Xanadu became the first independent startup to demonstrate quantum supremacy with their Borealis computer. Quantum supremacy just means solving a problem that would take a classical computer millions of years to solve.

  Earlier this year, Xanadu published another breakthrough in Nature, solving how to network quantum computers together to build a data center. Today, the company is headquartered in Toronto, has 200+ employees, and has raised over $275 million. So thanks again, Christian, for joining us today.

  Speaker 3:
  Thank you.

  Speaker 2:
  Would love to start with a little bit of your story. You had an amazing career in research. What prompted you to start Xanadu?

   

  Speaker 3:
  Well, it actually started when I was a teenager, before I went into academia. So I always used to love reading about people like Henry Ford, Rockefeller, Edison—people from, I guess, over a hundred years ago in the U.S.—and devour those types of books, and still do to this day. So I always wanted to start a company and be an entrepreneur first.

  I tried a few different startups in my teenage years—silly things that didn’t really work out—and then decided that wasn’t working out, so I paused that and tried to do a PhD. I went through step-by-step: go to undergrad, then master’s, and then PhD, and each time I was enjoying it and wanted to keep learning more. And that was in quantum computing. And then it was a postdoc, as you mentioned, after Australia at MIT, and then at University of Toronto. After, I wanted to stay in Toronto, so that was one key aspect.

  And they kick you out of the university—I did two postdocs back to back—and they want you to move on after a while unless there’s a position open as professor, but there wasn’t any. And then so I decided at that point—Rigetti was a company that got funded, and so I was keeping an eye on that and thought—

  Speaker 2:
  That’s an Alumni Ventures portfolio company too.

  Speaker 3:
  Oh, nice. And public now. And so I was thinking of how to combine this love of entrepreneurship and also being part of potentially a big industry like Henry Ford was, and others like that from the people I read about. So I really decided to give it a go and just merge both those things—the sort of expertise in quantum computing, and I was a part of, as you mentioned, contributing to the research in photonic quantum computing during my PhD and beyond for about 10 years—and then decided to see if I could start a company.

  Actually, my PhD was on quantum computing and security, so using photonics for both. So originally I started a company called CipherQ, and that was about commercializing quantum key distribution, or quantum security. The interesting thing there is that the technology is far simpler and already there compared to quantum computing, but the market’s not there. And the market is driven, ironically enough, by the advent of a large-scale quantum computer.

  So I was getting some contracts from Boeing and others to sort of just survive. But the investors were all interested in: “What about quantum computing? Do you know much about that? Because that’s something we would be interested in funding.” So at one point I had Xanadu and CipherQ, and then Xanadu was far more interesting to funding. And it’s kind of like—the investors at that point were customers in some sense. It’s like, you listen to your customers. So CipherQ slowly died a noble death, and then Xanadu just kind of started from there. So yeah, it was really a combination of love of entrepreneurship and the knowledge that I loved accumulating and contributing to in academia.

  Speaker 2:
  That’s awesome. Great background. Thank you for sharing that. So the title of today’s event is An Introduction to Quantum Computing. So I’d love if you can just provide a bird’s-eye view on what is quantum computing, how is it different from classical computing today—maybe define a qubit for the audience, and maybe talk about different approaches. That’s four questions in one, but we can start anywhere, and then I’ll follow up with others.

  Speaker 3:
  I always pretend I’m speaking to a family member when they ask about what I do. So I always try to answer it from that perspective. And a simple way is to go to the name: quantum computing. So it’s still computing—it’s still a computer that we’re looking to build. It’ll be something like a data center. So we’re all kind of familiar with computers and data centers for the most part, so that makes sense.

  And then you go to the “quantum” bit—and that’s kind of where it gets very unique. Traditional computers are based on classical information, classical physics—so things like Newtonian physics and so forth. So what you’d like to do is leverage properties of quantum physics—things like entanglement, interference, superposition—these weird properties that you only really observe or see once you’re investigating things on the atomic level and the individual photonic level.

  And then you want to combine these mysterious things like entanglement with computing. The reason why you want to do that is: for certain problems—important problems—you can actually see speedups. The overall thing is still a computer. It leverages laws of quantum physics. And the reason why you do that is because, for certain tasks—important tasks—you can now do things that would take millions of years and do it in a few minutes.

  And really, the things you can do—you mentioned we demonstrated quantum supremacy—that was for an esoteric math problem. The goal now is to really do it for, over the coming years, important problems. And these problems would fit in areas like logistics, pharmaceuticals (so drug discovery and design), quantum chemistry/material design—we focus on next-generation batteries, so how to use a quantum computer to build a battery that on a single charge lasts 10 times longer, for instance. Very revolutionary things.

  AI is another big one, and finance. So those are the fields that we’re all working towards.

  Now, you need a large—so the quantum supremacy demonstration we had had 216 qubits. You actually need millions of qubits to get to this point that I mentioned. But there’s ways around it where you can have a smaller quantum computer and copy and paste thousands of them and network them together, which was our big result we did with FORTH and Xanadu.

  So that’s kind of the “why.” And we’re used to using bits in computers normally—in our phones—encoding information in bits, also transmitting bits of information in Wi-Fi and internet and so forth. And that could be a zero or a one.

  Now with a quantum bit, which is a qubit, you also have zeros and ones—so that’s still the same—but now you have combinations of zeros and ones at the same time. And to give you an idea of why that’s important: it scales exponentially.

  So let’s say we have three bits. Let’s say it’s 0, 0, 0. There are eight combinations to have. It’s 2³. So classically, let’s say you have millions of bits, but let’s say you just had three bits going into a computer—you can only have one of these eight choices. So you either put in 0, 0, 0, and all the way down, you can put in 1, 1, 1, and any other combination—but that’s all that goes in.

  And then let’s say 0, 0, 0 goes in—that gets operated on by a bunch of gates. These bunch of gates implement ultimately an algorithm, or basically something that you want to find the answer to.

  But in quantum computing, what you can do is you can actually create a superposition—so you can have all eight states occurring, going into the computer at the same time. So now, instead of just only being forced to choose one, you can actually have eight going in at the same time. So you’ve got this 2³ where you’ve got three bits going in. And so now you can see that you can still have gates operating on all these things in a superposition, and you’re sort of scaling up—and it scales very, very well.

  Now, the catch is: that’s all true, but you need to really set up algorithms in a way that you can actually make them interfere in a certain way to give you the answer. Because the catch is, like in a classical computer—even though you’ve got 2³ states getting acted on—you can only measure, you’ll only get at the output three bits of information. So even though you can feed in, say, 0, 0, 0, you’ll only get 0, 0, 0 at the output. You’re not getting an infinite amount of information.

  So that’s the catch—you have to be able to store it, compute on it, but at the end you’re going to get an answer. So it’s very tricky. Quantum algorithms work in a very peculiar way where you try to basically interfere in a positive way and interfere in a negative way.

  Speaker 3:
  So where you’re pushing it to the solution that you want. So when you do measure at the end, you get the answer that you want. So that’s qubits. And as I mentioned, the other important feature is—you asked about the type of quantum computers. You can build a quantum computer out of anything as long as you can leverage these quantum properties like entanglement, superposition, or interference.

  So if you look at this wall, if you zoom in, you get to the molecules, and then you get to the atoms. If you can operate at the atomic level, then you can extract—you can create—entanglement properties in that.

  So there’s two ways to build a quantum computer. One is using photonics or one is using electronics. And electronics—one you can think of as using atoms or electrons. That would be most of the industry. So that would be IBM, Google, Rigetti, others like that. And then you have us using photons.

  And again, it comes back to this idea: as long as you can work on the individual particle or photonic level, you can then start generating entanglement, which is a precondition for all these crazy speedups.

  Speaker 2:
  Thanks for that. You answered four out of four questions in one go. Thanks for that overview. Could you touch a little bit more on the different approaches that are common today and maybe some of the approaches to build a quantum computer that are upcoming? So we’ve heard about ion traps, superconducting qubits, photonics—what are the pros and cons of each of those, and why did you choose photonics?

  Speaker 3:
  Well, the “why” is always an interesting one. If I’m being honest, you can rattle off the pros and cons, but honestly, it comes back to historical reasons—where did you do your PhD? And mine was in photonics. I mean, that’s the honest answer.

  Having said that though, there are pros and cons about all approaches. I would say superconducting qubits—that’s IBM, Google, Rigetti—they’ve been around for a while doing some good stuff.

  Let me start off, to be fair, with the challenge about photonics. The challenge about photonics is loss. If we fail, it’s because we haven’t been able to cope with or deal with loss. So if you’ve got—our computer is made up of lasers, photodiodes, fiber optics, and light traveling through a bunch of things—basically, whenever light travels through a waveguide or through fiber optics, not all information makes it there.

  The light gets scattered, it gets absorbed in the medium that it’s traversing through. So you’ve lost information—that’s loss. And that happens wherever the laser is—and the laser is everywhere.

  So we need to actually be able to counteract the loss. And there’s a few ways to do that: working with foundries to get better materials. Also, we’re a fabless company—we do all the chip design in Toronto and send them off to different foundries around the world. So working with them is important.

  And also, there’s another aspect where fiber-to-chip coupling—you’ve got to align that very, very well. Again, wherever there’s coupling, there’s going to be loss.

  Error correction and fault tolerance—everyone needs that. That’s our physical reason why we need it—loss is what’s causing it, what’s happening. And we need to have redundancy through error correction.

  So that’s our biggest challenge. But in the last 12–18 months, we’ve had about 30 times improvement in each of the optical elements. And we’ve also done fundamental loss studies. The worst thing would be if our computer’s architecture requires this amount of loss, but fundamentally you can only do here—which would suck, because if you pass fundamentals, which you can’t because it’s fundamental, we’re still in trouble.

  But thankfully, the fundamentals were below what we needed. So we’ll get there. But that’s definitely our biggest challenge.

  Now, on the photonic side of things, the good thing about it is you can actually do your chip designs and send them off to different foundries around the world. And when quantum computing takes off, you can actually work with TSMC, you can work with GlobalFoundries, which we have worked with in the past. So that’s a really great benefit.

  And the other one is the networkability side of things as well. We showed how to network four server racks together. So you can take out one and the rest still work. Or more importantly, you can have thousands together and network.

  So that’s the benefit. The other big thing is—there’s another good company called PsiQuantum doing good work there, photonics as well—they have those advantages too. But one of the good things about our type of photonics is far less cooling is needed for our approach, which is always good.

  Or—most approaches—I think every approach needs some form of cooling, whether it’s laser cooling or cryogenics. You need to cool these systems down. For us, 90% of our final computer will actually be on tabletop—no laser cooling, no nothing. So that’s kind of cool.

  The other thing is—so the other approach is like Google superconducting. They’ve done some great results recently. You would’ve heard of the Willow announcement and their chip. So they’ve done some really awesome stuff with error correction and fault tolerance.

  Neutral atoms are also doing some good stuff. I think they’re one of the ones that have really shot to the front of the pack in the last year with some of their great logical results.

  I think a dark horse would be spin qubits. They’re very interesting. In principle—I mentioned that we have to build a quantum data center—in principle, you could have all those qubits not in a data center, but on a single chip. That’s their promise.

  I would say it’s less proven out than other approaches. It’s also going to be hard to have yields—get highest-quality yields—but it’s a very interesting approach as well.

  The challenge with the atom or electronic or ion trap approaches is how do you network these things together—which is a real plus for the photonic systems.

  Speaker 2:
  Yeah, and I know that you and I have talked about the two major challenges being networkability but also scaling. You can prove something out on a small scale, but how do you actually solve real problems that are across industries?

  And that sort of touches on the next question, which is going down the applications of quantum, right?

  We expect that in the future, we have two kinds of computers that will solve different kinds of problems: classical computers that we have today, and quantum computers that solve other problems that classical computers can’t solve. BCG estimates that the value creation from quantum will be half a trillion dollars.

  And you touched on this in the beginning, but in terms of timeline—what industries can quantum really impact first? And maybe five, ten years after that, what’s sort of the rollout across industries?

   

  Speaker 3:
  The truthful answer is no one is 100% sure. We’re all sure it’s going to have massive impact, as you mentioned the BCG study. But a lot of us are coalescing around this idea: it’s probably going to be in things related to quantum chemistry. So that’s what we’re focused on.

  And quantum chemistry is a big field. You can think of examples like material design—so discovering new materials that can do whatever properties you care about. And also, as I mentioned, next-generation batteries. So how do you have it still safe, but a single charge—10 times faster charging, 10 times longer distance, and so forth?

  And the reason goes back to the original idea of Feynman in the early ’80s. There’s so much amazing stuff happening on the atomic level in nature that we’re trying to simulate using traditional computers that it will never be able to fully simulate with high fidelity.

  And so you’d like another quantum system that you have full control over that’s mimicking that other nature. And that other system that’s fully quantum, that you have control over, is a quantum computer.

  So I think that will be a good example—pharmaceuticals and material design. A lot of—it’s not like we don’t have drugs, it’s not like we don’t have new materials. So we do have them. But the effort that goes into getting them—like drug discovery, similar to material design—you spend 10 years to create a drug, going from computer simulations all through clinical trials, and 90% of them still fail after 10 years. And one to two billion dollars, for the most part, wasted.

  And so the same for material design. And so the reason is you’re trying to model systems that a classical computer is not the best at. We still have drugs, but again—a lot of time and money wasted.

  So the hope is a quantum computer can do it much more efficiently and say, “Look, you don’t need to worry about 100 different candidates for your drug. Here’s two—and with 90% probability or more, these are the right ones.”

   

  Speaker 2:
  I’d love for you to touch on the hardware and software approach. Xanadu has gone the full-stack approach. There are some players that are software-only. What prompted you to go that route—first of all, the full-stack approach? And maybe take a second to tell us about PennyLane, the quantum software platform that Xanadu has built, which is now among the world’s most used quantum software platforms. Yeah—just curious on that.

   

  Speaker 3:
  Yeah, for sure. So we are full-stack, like you mentioned. And for those perhaps who don’t know, full-stack just means you work on the hardware, software, and also in our case, the algorithms as well.

  Why we did—maybe it’s we’re control freaks. That’s kind of the answer. But I think you need to. The hardware and software are so interconnected. If you don’t have the ability to do a hardware company, then a software company is much easier—you can get a couple of million dollars and get started. You’re going to need a lot of money to build the quantum computer hardware. So there’s a lower barrier to entry.

  Having said that, we never wanted to do software—because then you’re relying on everyone else performing their hardware. So we really wanted to do both, because if you’re doing hardware, you really need to understand how the software gets integrated with it.

   

  Speaker 3:
  We’ve got software engineers rubbing shoulders, sitting next to the chip designers, for instance. So it’s very important to be tightly integrated from that point of view. So that’s kind of the reason why. And the PennyLane side of things—our software is called PennyLane—it kind of encompasses many things. If you think of the end goal of a large quantum computer, this is the programming language that you’ll use to program, to run things on it, run algorithms, and also the operating system. It’s open source, fully open source. It has Python as the underlying programming language. And what’s great about it from our point of view, obviously, is the fact that it runs on our hardware but also runs on everyone else’s hardware as well. So it’s hardware agnostic, and that was always the goal from day one because we think it’s a great product. It’s a lot of fun.

  People really love working on it and then having as many people building their ecosystem around rather than having something that’s closed-sourced or just available on our own hardware. So that was really key. Yeah, and PennyLane also—the stack, the full stack—is at the top. You have PennyLane. You may not know anything about quantum. You want to do some function, and you just type function and whatever it is. But then you can go down the stack where you’ve got compilation as well, and then you go down to the very lower part of the software stack where you’re controlling the hardware itself as well. And then obviously you have the hardware. And the other thing is, depending—there’s also a lot of, in the error correction side of things—you need a fair amount of classical computer as well. So you have a large enough quantum computer where you need to diagnose if and when errors are happening, and that has to get converted to bits. The bits have to be analyzed to tell you how to correct. That gets fed back on the quantum computer. All that needs to be done in a microsecond. So you need really powerful FPGAs or GPUs to do that as well. And so I bring that up because that also needs software that’s very much tailored to hardware and the software, the GPUs as well.

  Speaker 2:
  Absolutely. And tough technical problem to solve, but huge upside to whoever’s able to make it work. The final question from me before we hand it over to the audience to jump in: looking ahead 5, 10, 15 years from now, how do you see AI sitting alongside quantum? What does the world look like?

  Speaker 3:
  That’s always a tricky one to answer. Imagine if we didn’t have—I think Intel started, as one example, in 1968—imagine if we didn’t have digital computing today. We’re just obsessed with our phones and computers, and that it’s really proportional to human progress and innovation. No emails, no nothing. I mean, it’s crazy to do that. I mean, we obviously did it before, so we would be fine, but the progress would be severely diminished. We kind of look at quantum computing—we want to build this data center in 2029. So that’s when it has enough qubits to start solving important customer problems. But beyond that, they’ll get more and more powerful. And it’ll be kind of like imagining if you didn’t have digital computing, as I just mentioned. That’ll be like what it’s like to have quantum.

  In terms of the AI, we see some really cool things in terms of merging quantum and AI. First off, both of them have linear algebra at the heart of the mathematics, which is an important thing, because quantum computers aren’t going to do everything. It’s not going to load Facebook faster or Instagram faster or whatnot. But the small subset of problems are very, very, very important, as I mentioned. So I think that with that subset of problems, it is very hard to imagine how—with the AI, for instance—we look at models where the amount of parameters that need to be trained, and the amount of money that’s going into building these data centers, and the amount of GPUs—the trend isn’t right. It’s not getting more efficient; it’s not getting less energy, at least if it keeps going in this current one.

  So our idea, and others are working on this as well, is maybe these models that you’re training are still very challenging to train for a traditional classical computer, but maybe now you can actually have quantum mechanical models that you train, but you’re training with a quantum computer so it’s done in an efficient manner. And then in that case, maybe you can keep heading this progress that we’re seeing with the amount of training of data, but you’re not paying that huge overhead.

  And I think also, just in general, AI is going to probably need some sort of different paradigm of computing. It’s hard for me to believe—and I’m very ignorant—it’s hard for me to believe that AGI will be achieved with just bits, classical bits. First off, it’s missing analog. You’ve just got the discretization. And that’s where quantum mechanics comes in. It’s discrete and analog—wave and particle as well. So I think something is going to be needed, whether it’s quantum or biological computing, to really have these future breakthroughs as well.

  Speaker 2:
  Fascinating to think ahead. Thanks, Christian, for that. Now we’ll open up to the audience for questions. Feel free to raise your hand and please share your background before asking the question. So, I have one over here.

  Speaker 4:
  Hi, I’m Angela, Version One Ventures. You probably know my partner Boris.

  Speaker 2:
  Oh yes.

  Speaker 4:
  Yeah, I live here and I’m very excited that you’re here because we’re digging into quantum. You said something—and it might’ve been unintentionally—you said “if and when” or “when.” And so why did you say “if”? I’m just curious. I am thinking it’s inevitable, but I’d love to hear your thoughts on why you said “if” and what are the biggest, hardest obstacles still?

  Speaker 3:
  I definitely think personally it’s “when,” but “if” is more associated with—this is probably one of the most difficult technological feats that humanity has ever attempted. There’s a lot of amazing stuff with space exploration and that. But now the other end of the spectrum, where you’re actually working on an atomic level and you’re trying to encode information there and manipulate it—so it is just inherently hard. And so I guess it’s good for us to stay humble with that sort of thing as we’re kind of grinding it out every day. But also from day one, it was always a case of, “This is crazy. Are we going to actually try to build this quantum computer?” So I think it’ll happen.

  If I’m being honest, our biggest challenge is loss, as I mentioned. But historically, we’re on track to get to the loss levels that we need. So I think it’s more just trying to be as humble as possible in the face of what we’re looking at. But I would say it’s looking—in the early days it was like, “Yeah, let’s just try it. It’s going to be fun.” But now, in the particularly last six to twelve months, we’re seeing the light at the end of the tunnel. It doesn’t seem to be as crazy as what it was in the early days. It was kind of binary—at least the investors thought that, and we did too. It’s like you’re either going to do it or you can’t. And it’s still that, but it’s more weighted to “this is probably going to happen this decade.” And so that’s probably the way I said it.

  Speaker 2:
  Go ahead.

  Speaker 5:
  Carsten. I am a cloud computing part of the Berkeley Angel Network. Anyhow, I received my PhD from Stanford and Berkeley, but trusted. My question is, I think you were talking about data center-scale quantum computing by 2029. What sorts of challenges do you foresee as you scale those up to data center scale, and what are the unsolved problems that you’re still having to tackle before you get there?

  Speaker 3:
  I sound like a broken record. Sometimes the answer is always loss, and it’s kind of an easy answer, but it is definitely true. But it’s a hard problem. But the good thing is, there’s not like five or ten different answers there. It’s really just loss. There’s a lot of things. The two big challenges—getting back to the question—is like, well, let’s assume we want to build this data center today, just for argument’s sake, and you’ve got thousands of server racks, and they need to be networked. The networking is very hard. We achieved that in Nature a month or two ago, but it’s very hard if you use non-photonic-based approaches. So let’s say you have thousands—but let’s say you just have two—you need to convert from atoms or electrons to photons, because you’re going to use photons to distribute as a networking, defined that way, and then back to atoms or electrons in terms of the information sharing.

  Speaker 3:
  Very, very hard to do that. No one’s really done that before. It’s very highly probabilistic, meaning every time you try it, it doesn’t always work, and it’s very noisy and lossy as well. So that’s one big challenge we’ve been able to solve that others haven’t yet. And the other big one is—I mentioned loss—more broadly, you can think of that as the other big bucket everyone needs to solve: improving the capabilities of error correction and fault tolerance. So this basically means that you’ll have a bunch of qubits. If they’re not encoded, we call ’em physical qubits, and you’ll want to have some redundancy. So typically the average is like a thousand physical qubits will give you one very good qubit or logical qubit. And so that’s very difficult to scale up. And the noise model matters, the code distance matters, the codes that you use matter, and we’re seeing some good breakthroughs by us and others in the field from this.

  But that’s the other kind of the final thing. And it’s a big thing too, that can be broken down into many parts as well. But at a high level, that’s the other big one. And investors often ask us, what should we look for in the coming years? And these are the things—we’ll be seeing more advances in error correction and fault tolerance, the lowering of errors through other means. We work with foundries that we can work with on the ground to work on the processes there that allow them to have lower loss as well from a physical sort of substrate side of things.

  Speaker 2:
  In the back—

  Speaker 6:
  Three quick questions, technical line. First one is regarding—we’re talking about data centers, and data centers use a lot of energy. This is one—I’m building off that—I’m an urban planner and into energy efficiency. So is this thing more energy—will it make the data centers more energy efficient? And the second question was related to that about the car or storage battery—understanding how the computerizing of things would make the battery ten times more efficient. And then the third question is not very esoteric. I grew up—I was born the year before Intel, we were talking about—and I grew up, the big tech was television, and we had rotary phones, those kinds—record players. And I remember life was quite good, and I think when I was born, when Intel came out.

  And now how has our lives as human beings improved? I can’t count on one hand. I think, okay, the cell phone means, like, me and my wife can communicate to each other when we get lost in the mall. What else? How else is it making—as far as my family life, when we’re eating, it’s like I got the telephones out and we’re not having these conversations—deep conversations like we had back in the seventies. We’re still driving the same cars, they go the same speed.

  Aircraft is still flying at the same speed as when I was a baby. And I don’t know what’s—how it’s improved. I mean, yeah, I don’t have to type on a typewriter, and I can read and cut and paste—

  Speaker 7:
  Believe—

  Speaker 6:
  That—and my drawings—I’m an architect. So that’s where files help, I think. But there’s so much negative—like, news—because everybody can post things. I don’t know what’s real news and what’s not—fake news—and all the news sources that used to—and editors that used to be this—are not. So now our democracy is threatened because people are believing fake news, which is basically propaganda—the rise of Nazism in Germany and the United States—and rise of authoritarianism tied to fake news. So I’m just wondering how all this quantum computing is going to help us in the future. And it’s dangerous—I watched a sci-fi movie where the computer basically took over the humans and was completely autonomous.

  Speaker 3:
  Yeah, I think it’s discouraging sometimes—the negativity—particularly when it comes to the news and not knowing what’s real. So I always find that a challenge. I also can’t stop watching too, or reading. That’s always hard. But I think in terms of your question, I definitely see two areas that technology has really changed in the last 20 years. And going back to, say, ’68, the education side of things—availability of, say, Wikipedia and all the information at our fingertips—I think that’s transformed things. And also the other one would be medicine as well. So I think those two things—and hopefully quantum can have an impact in those two. So I’m hopeful in general. The space exploration and so forth has really come along as well, and that ties back to having, say, Starlink and internet and information being shared. So overall I’m optimistic. But yeah, it’s hard sometimes to stay that way. But—I forget the first parts of the questions. First two—

  Speaker 6:
  About energy—

  Speaker 3:
  Efficiency—

  Speaker 6:
  Centers. And the second one is about storage.

  Speaker 3:
  The energy efficiency inherently comes from the fact that you’re going to solve problems with—the easy answer is that, let’s say you have a thousand server racks—it’ll be whatever the energy efficiency is for a thousand server racks in a normal data center. So it’s kind of, for the most part, one-to-one. The cool thing though about this is that for certain big problems, it stays fixed for quantum computing. So the problems that quantum computing works on, it stays fixed, meaning that it would’ve taken hundreds of data centers—traditional data centers, classical compute—to solve the same problem. And so that’s where the energy savings come from. You wouldn’t build a hundred data centers to solve that problem. So you’re getting things that would’ve taken an exponential amount of time.

  The battery one that you mentioned—that’s still early days. We work with a lot of car companies, but basically it comes back to anything you work on in quantum chemistry. It’s like—there are some certain types of interactions that are described by a Hamiltonian, by a certain energy, and you need to understand how that evolves in time. And it’s very, very hard to actually do that in general. There’s DFT and other sort of ones that help you with the dynamics of systems and so forth, or finding a low energy state of systems. And that’s where quantum in general plays a role. So if you can describe a Hamiltonian that you care about in this, and you can sort of exponentiate it to get a unitary, that unitary can then be broken down into a bunch—millions—of gates, and then you run your quantum computer there to understand how the system evolves. And there’s a lot of quantum chemistry applications like that. We just can’t do that effectively enough, because if you did a one-to-one correspondence, it could take millions of years to do.

  Speaker 6:
  The bottom line is, using this quantum chemistry—the batteries—you can use the full efficiency of the…

  Speaker 3:
  Yeah, it’s much more deep than that. At a high level, you have electrolytes, you have the cathode, anode—you don’t—when we work with the car companies, they say, “Okay, with this part over here, we can classically solve quite simply.” So we have to find things by working with car companies, where a certain interaction—whether it’s in the cathode or anode or wherever—they know it takes them months or years to do. So that actually takes time itself. I always say that even if you had a large million-qubit quantum data center today, people wouldn’t know what to do with it. That’s why all the research is going hand-in-hand with the hardware developments as well. It’s not like—let’s say we could go into the future 10 years, get the latest NVIDIA chip and bring it back—let’s say it was compatible. You just put it in, you’re expecting it to go much faster, whatever you’re doing. It’s not like that. If I could go to the future and bring back a quantum data center, the math is completely different at the fundamental level. So there’s still work to do to match these things up.

  Speaker 2:
  In the back—

  Speaker 8:
  So two questions. I think for the quantum computing industry, a lot of the funding, at least on this side of the ocean—or every time Google or Microsoft announces something—it’s not necessarily even a US company. It’s more of a Chinese company comes out and says, “Oh, we have a chip equivalent,” etc. I was wondering if you—maybe not do an ode of Canada—but do you see that as a good thing, kind of that competition? A little bit of—it keeps you honest. Otherwise, why would Google innovate if they don’t have a competitor that’s knocking on the door? And then the second one is, how do you—in the LLM market—you see the concentration of a few people having the directive to say what gets put in the training data, let’s say. And there are some open source models, but you still need that compute. And I think—how do you make quantum more distributed? So it’s not just these frozen data centers that only a certain amount of people can tell who gets access and who doesn’t. And that’s the two-point question.

  Speaker 3:
  Yeah, that’s tough, that last one. Because within a quantum data center itself, it’s distributed. But then if you’re talking about how to have something in Toronto and something in Palo Alto—a data center here connected—you really have to just preserve the entanglement over hundreds or thousands of miles, or however far it is. That’s very difficult. You’re going to need a quantum internet for that. That’s a big challenge that no one’s solved yet. And it’s up there with building a quantum computer. So that’s the biggest challenge.

  In terms of the first part of your question—it does help. Because we’ve raised $250 million from venture capital and about $40 million USD from the Canadian and US governments. And it creates a paranoia—a healthy paranoia—that they want to then fund people like us and others in the industry. They definitely don’t want China to get ahead.

  And also the ultimate—the reason why quantum computers exist in the early days is the fact that it can break encryption. And so they need to be ahead of that, needing to understand how to get there first. They need to understand the preventative ones. There’s post-quantum software that can solve this as well. But yeah, we’ve had four DARPA contracts now. We can—yeah, there are certain ones that we—one of them, for instance, with Raytheon at the moment—we can’t be the prime, but we can be the subcontractor because we’re a foreign entity. But soon that’d be solved in the 51st state. I don’t anticipate any issues then.

  Speaker 2:
  I think we have time just for one more question. I want to be conscious. Go ahead.

  Speaker 9:
  So I tried to imagine some parallels between the classical computing evolution and also quantum computing—vacuum tubes, transistors, then the mainframe, then supercomputers, then the desktop, now with the mobile, now back to data centers. So where do we put in this spectrum? Where do you put the current stage of the quantum computing? That is one. The second one is, it is just more of a comment. You mentioned that you are focusing on the entire stack. In my mind, that is the right approach because you don’t want to just build a quantum computer, wait for somebody to build this stack. Right? Exactly. Because there are a lot of examples today—Nvidia is successful not because of the GPU, it’s just because of the ecosystem. Otherwise other companies would have done this with GPU a long time ago. So that is the first question, right? Yeah. So I’ll stop there. I think there are a lot more things.

  Speaker 3:
  Yeah, you’re right. That’s how we see the ecosystem play as well. And in terms of the first part of the question, it is hard to do a one-to-one correspondence—

  Speaker 9:
  Stage-wise, right? Because I have my personal experience—I played with UNIX way back then.

  Speaker 7:
  (Laughs)

  Speaker 9:
  I worked at Indian Space Research Organization. Then I worked at Sun right from the beginning, ’89. So I have seen all the platforms, how it all evolved the last 30 years. So using that data point, when I hear something, hey, should I spend time on this—chasing this quantum computing? Because I’ve chased a lot of flashy things.

  Speaker 3:
  Yeah, it is hard to do a complete one-to-one here, but there’s elements where we’re kind of post-vacuum tube, but other elements where we fast-forward decades—we have networking already, we have the fiber optics—so you can kind of see, if you look—like you mentioned—the different histories of computing, there’s some over there, some over here. And also the fact that we need a lot of traditional compute as well—we have that at our fingertips as well. So it’s kind of a combination of things. But we can say that the first one will be a quantum data center. It won’t be that you can sell these computers themselves without being in a quantum data center.

  Speaker 9:
  But that is how it evolved. I played with the DEC 2015—people who know it, VAX machines—they were all there in very specialized buildings, right?

  Speaker 3:
  That’s right. Yeah.

  Speaker 2:
  Thanks a lot, Christian. Thank you to the audience for attending and for your thoughtful questions. Christian is generous enough to spend a little bit of time after this event, so if you have any questions, you’re welcome to come find him. I have one more question if you don’t—

  Speaker 9:
  I was thinking about these platforms, right? Platform—Is there anybody else that has this kind of full stack vertical support?

  Speaker 3:
  There’s a couple. Yeah, most fit into one or the other, but there are a few to varying degrees that have it. IBM is a good example—they have Qiskit, their software, and they also have their own hardware as well.

  Speaker 2:
  No worries. That was a good question—thanks for sneaking that in. And yeah, thanks again, Christian. Appreciate your time, and thank you everyone for attending.