Sachin and Adam
00:30:37 · Pete Shadbolt
[00:30:37] S0: Yeah, so the machine looks like a supercomputer. It's a big building. It's full of racks. The rack is full of chips, and then the chips are wired together with optical fiber. We can make the chips. We make them at the highest level of maturity of chip manufacturing that exists. There's no next step. The performance in some cases is good enough. Some of the other components not quite good enough yet. And so we're continuing to improve that component performance, but we're on a really good trajectory there. We feel real good about that. And we have all of the devices like it's feature complete. Right now, as we sit here, there are three racks sitting next door. And these are like three meter high liquid helium cooled racks. Each rack supports hundreds of chips, kind of like the racks you'd find in a data center. And we've been running chips inside that kind of system for a few years now. So that works, the cooling works, the networking works. And at the moment we're in the business of loading thousands of photon sources into those cryostats. That's our Alpha system. When we've got that thing up and running, that will be a huge milestone for the company and that will basically be something where you can look at it and say, okay, if we improve the few component specs remaining according to the plan that we've had for a long time and if we just basically add more racks, this thing is going to become a full scale system. And so, yeah, right now we're in the middle of integration hell of the mechanical failures and the electrical failures and everything that happens when you build a system two orders of magnitude bigger than anything that's ever been constructed. But none of that is really rocket science. None of that is new technology. It's just really complex and a big step up in scale. And so this system, I think, is going to be hugely significant as a prototype for then the next systems that we're building next year in Chicago and then of course the really big system that we're building in Australia. But yeah, I'm extremely encouraged by the device performance, the system performance, the networking, the cooling, the manufacturing, all of these parts of the puzzle are really coming together. And it remains the hardest technical challenge that humans have ever really undertaken.
00:33:22 · Pete Shadbolt, Sachin, Sachin
But we just keep knocking down these milestones, improving the numbers, and we're going to get it done. [00:33:31] S2: It inevitable that it does happen? [00:33:33] S0: Yeah. Unless, you know, it a full answer [00:33:39] S1: or a technical answer?
00:33:41 · Pete Shadbolt
[00:33:41] S0: Well, that's my technical assessment. Yeah. You know, that's my best technical assessment, is that you can definitely build a quantum computer. You can definitely build a photonic quantum computer. And there's thousands of hard things along the road. But again, I'm mostly encouraged by the experience that I've had in the last decade of this company. And I want to emphasize, right, like so many of the things that we've done, when we first put that to our engineers, okay, you got to get down to a 0.1 dB per meter of losses in the waveguide. You've got to take a superconducting single photon detector. You've got to put that in a chip, not in your university clean room, but in a commercial fab where the engineers are mean and they don't care about how cool your device is, and you've to be able to make millions of them, and you've got to hit, like, 99% detector... You know, some of this stuff, when we put that to our team members, they were, like, nearly mutiny ing. Right? Like, this is ridiculous. It's just not gonna happen. We're not gonna hit the schedule. Leadership are foolish. Whatever. And they've... Credit to them. Got on with it and then went and delivered on those things. And, you know, sometimes they're a little late. Sometimes they're actually a little early. Sometimes they're a little under spec, sometimes it takes them a little longer, but they did that stuff. And and so having that decade of witnessing the company overcome these, like, you know, almost unbelievable, difficult technical accomplishments. That... That's the basis for my optimism and belief that there's just no reason you you shouldn't be able to do this. And the right thing to do is to not wring your hands and spend your time worrying about how difficult it is and just get on and do it. And that's always how we have success in this company. From one point of view, quantum computing is alien technology. We're going to build huge entangled states that humans have never seen before on this planet. And there are aspects of the physics that remain baffling to our best theorists. But on the other hand, our chip, that's a 45 nanometer fab process. That's like my grandmother's fab process. You take that to TSMC, it's like an ancient, you know, nineties fab process.
00:36:05 · Pete Shadbolt, Adam Miller
And then, you know, the cabinets, they look really cool, but they're basically just chunks of steel welded together. It's a big metal box. It's rather insulting to our cryogenic engineers, but I think they would support that. You know, there's two points of view on the thing. On the one hand, it's alien technology. On the other hand, it's silicon, steel, fiber, you know, helium. Build it, put it together, make it work, fix the problems. Like, that's how we get this thing built, and that's generally the attitude that we try to have at the company. [00:36:41] S3: In AI at the moment, there's obviously a whole swath of national security and geopolitical concerns. [00:36:48] S0: Yeah. [00:36:48] S3: Whether it's Anthropic and the Department of War or people talking about supposed race between America and China and who's going to have the best models, what will they use them for? When quantum computers get developed, what will that race look like geopolitically, and what do you think the national security concerns will be?
00:37:06 · Pete Shadbolt
[00:37:06] S0: Yeah, it's a great question. And I think, you know, the I have to confess my novice level comprehension of geopolitics, But what does seem very apparent to me is that the world is increasingly moving into a regime where power is captured on the frontier. The frontier of advanced technology, the frontier of scale, like... Again, ten years ago, nobody, no politician really cared about a fab. Now they all know that it's extremely important that TSMC is located in Taiwan. And we see this across various different regimes, as you described. AI supercomputers have the same dynamics. Energy has, of course, profound geopolitical implication. And I think that quantum computing is going to have the same kind of dynamics. And that's a function of our earlier discussion. Were it the case that small quantum computers could be profoundly impactful, you wouldn't have to worry about geopolitics. And you guys are very familiar with the discourse around open source models and edge AI computing, blah, blah. It's a very sympathetic idea. Like, we're all going to have our own small models and it's all going to be democratized. And that would be a lot less geopolitically intense were that the case. Again, I'm not an expert, but at least from my point of view, it doesn't seem to be going that way. It seems as though the foundation models, they really have the power and you need billions and billions and billions of dollars of GPUs to capture that power. And I don't see that changing, at least at the moment. Quantum computing, unfortunately, has the same dynamics. Fabs have the same dynamics. Everyone wants to build a small fab. It's a nice thing to do. But the power, the geopolitics is dominated by the tier one fabs. And they're bloody expensive, and there's just nothing you can do about that. And and so I think... And and, again, it's amazing to think, like, that is coming from the most foundational pieces of microscopic physics. Like the power that TSMC has is rooted in the fact that they can make transistors that are a few nanometres in scale. It's really amazing, like nation states, geopolitics, you know, this scary grand stuff is being controlled by something that's the size of, you know, a few atoms. And that to me seems like a real trend. And quantum computing, of course, fits firmly into that trend.
00:39:53 · Pete Shadbolt, Sachin
We're going to gain mastery over the microscopic foundations of our world. You need a big old machine to do that, unfortunately, and we think that's going to be geopolitically significant. [00:40:08] S1: Once we get one of these machines up and running, which sounds like it'll be sooner than ever on things, who actually gets to decide what these things are used for? [00:40:17] S0: The market will decide. If someone's paying real money, like serious money for something, that's a pretty good signal that it probably is useful and if that's sustained, then I'll believe it. Of course, there's an entire academic apparatus that is extremely well equipped to differentiate between uses of quantum computers that are mere demo and applications that are really impactful relative to conventional supercomputing. And that benchmarking and evaluation exercise is ongoing. We work with DARPA on a lot of that work, but there are many other organisations that we engage with and that are doing beautiful work to really deeply understand the technical benchmarking. But yeah, ultimately, in my opinion, the true test is going to be whether people pay for it. [00:41:14] S1: Do you think these things are too powerful for capitalism to decide?