Sachin and Adam
00:15:46 · Pete Shadbolt
[00:15:46] S0: -Yeah. So the thing is, this has been the scientific consensus for twenty five years that you need about a million qubits to make money, to have impact, to solve valuable problems, to change people's lives with these downstream products of machine. Like, that was the consensus. That's what I learned when I took my undergraduate quantum information courses at Leeds University seventeen years ago. And so, you know, it's a great question. Why did PsiQuantum take this kind of provocatively bold approach of going straight for the big machine? The answer is that the big machine is the only one that's gonna make any money or have any impact. So you gotta build the big machine real quick. And that's an unpleasant reality. I don't like that. I'd much rather make money with a small machine, but there are no useful applications for the small machines. And so I think, you know, I'm very grateful actually to my co founders and for having had that experience that you mentioned of messing around with small machines in the university. So we put the first quantum processor on the Internet at University of Bristol fifteen years ago. And for a moment, we convinced ourselves that maybe that was worth doing. Like for a couple of years, we got real excited because we had these real quantum processors running. We could run code, people could get API access. And we thought, you know, if we're lucky, maybe there's actually real impact from these small machines. And basically, we burnt out that optimism in the university. And we went and did our arithmetic again, did our homework again and said, you know what? You gotta build a big machine. And so when we founded the company, it was with this sort of violent insistence that every hour that we work, every dollar that this company spends is only gonna be in pursuit of the machine that is actually having a real impact and making money and paying our investors back. Like, that's the promise that we made when we founded the company. And it's challenging, but that's necessary. I would point, and Alex is going to stop me from doing this, but I would point to Waymo as a very analogous example. You know, Waymo had to spend enormous upfront capital, and they were very extreme. I would say they were the most extreme in terms of the, you know, spectrum of self driving car companies in saying, we're not going to kill anyone.
00:18:22 · Pete Shadbolt, Sachin
We're going to be incredibly methodical. We're going to throw every tool at the book every tool in the book at this. LiDAR, you know, enormous teams, capital, the whole thing. And, you know, they were doubted for a long, long, long time. Even when they were doing demos, driving the thing around, people would laugh at the prospect of them pulling this off. Now if you've sat in a Waymo, you know, like, they're coming for every major city in the developed world. And and, you know, that that was the right strategy. They understood that a half self driving car isn't worth it. It's to real be thing. Yes, quantum computing just has the same dynamics. [00:19:06] S1: So you mentioned that going after the 1,000,000 qubit computer is the most scientifically rigorous thing to do. Why approach it with photons? Obviously, there's a lot of smart people in this field that don't think that's the best approach. What's the secret you guys know about this that others don't?
00:19:24 · Pete Shadbolt
[00:19:24] S0: Quantum computing is like a crazy field in that there's all these different teams. They're all really well funded, and they're literally using different physics to build the qubit. This is different to, like, the AI accelerator chip companies where there's differences in architecture, but at least they agree on, like, transistors and silicon. We're, like, so much further from that, where this person is using atoms floating around and this person's using light and then that person is using electrons floating on helium. And it creates this, like, incredibly combative competitive landscape where everyone is trying to argue their approach is the best, and everyone else's approach is, like, horrible, never gonna work. It's kind of silly. But with photons, the reason that we always liked photons was not because of the qubit. Photons are actually a pretty annoying qubit. They're hard to entangle. They don't interact. Light doesn't interact. If I just cross two laser beams, nothing happens. The reason that we always were compelled by photons is that they scale. And this was again with the understanding that we're only interested in getting to the scale necessary for real impactful applications. When you think about building a million qubit machine, you got to take into account manufacturability. How do we manufacture, test and yield millions to billions of devices? You can't do that in a university clean room. You can't do that with exotic materials or atomic scale fabrication or various of the science fiction characteristics that we normally associate with quantum computing. You've to think about networking. You can't fit a million qubits on a chip. With one or two notable exceptions, you're going to have to network chips together. And with photons, can network using standard telco optical fiber. We can just send the photon through fiber, no transduction. We do a two fifty meter qubit link in The UK with 99.7% fidelity. That was basically three or four people who put that together. Actually kind of easy to do. We don't need the extreme cooling of a superconducting qubit. So Google and IBM systems, which are amazingly impressive, they're running at 10 millikelvin, so oneone hundredth of the temperature of deep space. We can run at a few Kelvin, so 100 times warmer.
00:21:57 · Pete Shadbolt
That gives us about a million times more thermal budget in terms of the cooling power at base temperature. And so these are kind of scaling considerations that if you think about it, are closer to what we're familiar with in data centers manufacturing, cooling, networking, control. That was all the motivation for photonics as opposed to the qubit itself. And there were some major unlocks that happened with photonic quantum computing really early on, and a lot of that work was actually done in Australia. So there was a 2001 paper from a bunch of theorists. They actually set out to prove that photonic quantum computing wouldn't work, And to their own surprise, they found the first architecture that showed that you could do this. And then my co founder Jeremy actually did the first two qubit gate for photons in Brisbane about twenty five years ago. And we're derived from all of that work. But yeah, it was always really attractive in the context that you have to build a big machine.
00:23:04 · Adam Miller
[00:23:04] S3: You mentioned before that it feels like an unpleasant journey because you've got to build such a huge machine, and you've got to wait a long time until there's this sort of commercial progress, and starting any company is hard, but this is like starting a company on absolute steroids. What are the sort of parts of your personal journey that would get us to understand why you're sort of so excited to take on this unpleasant journey? Like, what is it within you that makes you want to strive to be on this journey for decades that milestones keep on getting pushed back and there's all this skepticism and stuff? What keeps you energized on this journey?
00:23:43 · Pete Shadbolt
[00:23:43] S0: Yeah, so firstly, it's not actually an unpleasant journey, and secondly, we're hell bent on not making it very long. Always the thesis with photonics was that we would leverage the big fabs, the exact same semiconductor supply chain that builds 100,000 GPUs for Elon and allows them to put that cluster together in one hundred and twelve days, that same machinery, we're putting that to work to build a photonic quantum computer. And we think that gives us a much faster path to machines that have real impact. And we see that, you know, the wafers that you guys have seen, those are built at the highest level of maturity of chip manufacturing that exists. They're built in a 700,000 square foot fab that normally builds laptops and cell phones. And we can make basically as many of those wafers as we could ever dream of. That's a level of leverage and scale that we've dreamt about for our entire professional lives and we're really at that point now. And then in terms of like the unpleasantness, I think what I mean by that is just there's a lazy, easy thing to do in quantum computing. We were all and I speak for the field we were all successful academics, right? Like all quantum computing companies come out of university research groups and we all got really good at publishing papers. And there's a pretty straightforward career path where you finish your postdoc and you go and raise a bit of money and you just keep publishing papers. And you do a venture backed research group and you keep kicking out press releases and you show a cool, interesting system. We could have done that. It would have been pretty easy. We had small processes running, but that's just not getting us anywhere. We had to go and spend hundreds of millions of dollars and do a 35 layer tape out on a 300 millimeter processor at Big Fab if we were ever gonna actually convince ourselves that we're building a quantum computer. And and I think that that was the key is that we just got fed up of not really putting our money where our mouth was. And so actually, it feels great to be every day knowing that we're solving the real challenges associated with building a really big, valuable machine. And it's the privilege of my bloody life to get involved in the scale and the magnitude and the madness of this company. This company has done absolutely insane shit.