theCUBE + NYSE Wired: Physical AI & Robotics Leaders | Pete Shadbolt, PsiQuantum

Clips
More from theCUBE + NYSE Wired: Physical AI & Robotics Leaders

Speakers

Pete Shadbolt

Co-Founder & Chief Scientific Officer

PsiQuantum

play_circle_outline Quantum computing advancements and commercial interest

play_circle_outline Comparison to NVIDIA's evolution and success in the industry

play_circle_outline Focus on building high-qubit quantum computers for specific industries

play_circle_outline Importance of surrounding oneself with a strong team and ecosystem for innovation.

play_circle_outline Unlocking Quantum Computing: Harnessing Semiconductor Industry Innovations Through Collaboration and Government Support

play_circle_outline Push for commercialization in quantum technology, transitioning from academia to fab

Info
Transcript

Pete Shadbolt, PsiQuantum

Pete Shadbolt

Co-Founder & Chief Scientific Officer PsiQuantum

Pete Shadbolt Discusses Quantum Computing Evolution at NYSE Wired Robotics and AI Media Week

Pete Shadbolt, co-founder and chief scientific officer at PsiQuantum, takes the stage during NYSE Wired Robotics and AI Media Week to share insights into the evolution of quantum computing. Shadbolt's expertise offers a unique perspective on the industry's rapid advancements and the transition from academic research to commercial application.

In this insightful session hosted by theCUBE Research's analysts, Shadbolt discusses the current landscape of... Read more

explore Keep Exploring

What technology was once considered a fantasy or pipe dream, but has now become a reality in the same way that self-driving cars have? add

What are some of the implications of the end of Moore's Law on the semiconductor industry and the potential for growth in quantum computing? add

What is the approach to building the first systems for quantum chips and where are they being produced? add

What notable developments or partnerships surrounding NVIDIA are you excited about that could bring opportunities for discovery and innovation? add

What is the basis of our thesis and how are we leveraging the semiconductor industry for our project? add

What qualities are necessary for a successful AI startup and how does PsiQuantum differentiate itself in the industry? add

bolt Powered by CUBE AI

Pete Shadbolt, PsiQuantum

search

>> Welcome back everyone to theCUBE's live coverage here at the New York Stock Exchange floor. I'm John Furrier, host of theCUBE. We're here as part of theCUBE and the NYSE Wire as an open community of experts coming together in a trust network, sharing their ideas, knowledge on the top content, more importantly, engaging with the data that's crucial, critical for the community. We got focus on robotics, AI leaders, large-scale supercomputing's here, and of course quantum is a big part of it. Our next guest is Pete Shadbolt, co-founder and chief scientific officer of PsiQuantum from out of the lab and into the fab. I love that phrase. Pete, thanks for coming on theCUBE. We're about to close down the option trading here. Wild day on Wall Street. Wild days in tech for sure. Thanks for coming on.

Pete Shadbolt

>> John, I really appreciate you having us on and especially on a day like today. But yeah, hopefully we can talk about something positive and exciting. Yeah, happy to talk to you

>> The quantum computing, obviously, you see that and you go, okay, it's a game changer for sure, and everyone kind of loves it here in tech, we see it. However, there's been rapid significant advancements in the field. Commercial interest is high, and so there's a lot of engineering going on, and of course there's also fear breaking crypto and all these things and Bitcoin, and then you got the recent events this past year, Jensen Huang on CES, say, oh, it's like 20 years away. It's never going to happen. Then you got folks saying, hey quantum readiness is a big thing that's going on. Get ready for quantum. And then CEO of IBM is like, no, it's coming faster than you think. So where are we on all this? Break it down for us. What's your perspective?

Pete Shadbolt

>> Yeah, so I, first of all, really appreciate you having us on and for making it in time. I'm British. I moved here to Silicon Valley eight years ago to start this company. When I arrived here, very serious technical people told me that I would never see self-driving cars in my lifetime, self-driving car, fantasy, pipe dream, always five years away. And they'll get to 80%, they'll get to 90%, but you'll never actually see a real thing. And of course now if I go up the street to San Francisco, every other car that I see is a Waymo. I think quantum computing has the same kind of dynamics. It is a technology that has a reputation for being a science project that's always five years away. Jensen's comments, I thought were very well-placed in that he put his finger on the root of the problem, which is that you need about a million qubits relative to the hundred qubits that we see in the current system. So Google and IBM, for instance, are both in that regime of about a hundred qubits. What Jensen said was that, that's a hundred thousand x scale up, right? We need a hundred thousand x scale up. That sounds like a multi-decade exercise. And at GTC a couple of weeks ago, I was lucky to be able to put it to him directly that he himself, of course has the counterpoint, the xAI Colossus supercomputer, that's a hundred thousand GPU cluster, was famously built in 112 days. And the miracle that enabled them to do that is the semiconductor industry. It's the trillion dollars and 50 years that puts a billion transistors in your pocket. And our mission, the Lab to Fab mission has been to get leverage of that capability and deploy it to make commercially useful quantum computers much faster than anyone has expected them.

>> Pete, that's exactly a great point. I love that whole conventional wisdom kind of dis. It's totally true. I mean, you know what you don't know, but people think they know what they don't know. We don't know. If you look at Jensen's keynote too, he also spent a ton of time on scale out and scale up, and he spent, I think, 20 minutes of a very valuable keynote on these concepts. And then he then showed the progression of his chipset. So we're already seeing the silicon advantage. So I guess my question to you is, as you get the wafers, getting all that stuff on the Silicon side, you do all that materials management, how does quantum translate into commercial viability with Silicon? And is that the main focus or is it software around it and, et cetera? Because NVIDIA had that same path, right? They had to do the software, they buy and get the InfiniBand from Mellanox and boom, they got a system. So it could be that quick. What's your vision on the accelerated quantum keynote that you'll do when you're like the NVIDIA status of quantum? I mean, accelerated quantum is actually a great name, but is it accelerating and what happens? What happens next?

Pete Shadbolt

>> Yeah, so we of course... Jensen and various other leaders from Silicon Valley, from the semiconductor space, these people are heroes for me, for my co-founders. And it's just been astonishing to see the journey that that industry has been on, that those people have been on, and the way that the semiconductor industry has been able to drive growth and just move at an astonishingly fast pace from small toy systems to now these enormous, unbelievably powerful machines that people are building today. I think, obviously I would imagine that there's a lot of people in your building there thinking a lot about growth today. And one of the striking things about the semiconductor industry is that there is a limit to growth. We're coming to the end of Moore's Law. People are starting to resort to frighteningly extreme measures to continue to build more powerful computers turning on Three Mile Island, building nuclear power stations to run these systems. And what's exciting for me in quantum computing specifically, is that we're going to unlock a categorically new type of computer that is actually going to be at the beginning of that growth curve rather than the end of it. So the first computers that we build are going to look really primitive, you know that they're going to be powerful, they're going to solve extremely lucrative problems that we won't solve with any conventional supercomputer, but we're going to look back on those systems in 10 years or 20 years the same way we look back on IBM mainframe today as a very primitive and early stage technology. But our approach to building even those first systems is to leverage the highest maturity fabs. We build thousands of wafers of quantum chips right now in the production line of global foundries in upstate New York. It's pretty cool that that's a US-based fab where we're doing that. And that's essentially the highest level of maturity of chip manufacturing that exists on planet earth. We're very proud of that as a startup company that we've been able to take a technology that is normally thought of as a science experiment, research experiment, whatever, and put that into a fab shoulder to shoulder with chips for laptops, cell phones, cars, defense, etc. And we really think that that is the key to getting to millions of qubits and unlocking these profound applications.

>> What a great accomplishment. Congratulations to you and the team there . Just as a side note, the first Apple computer was a wooden box, right? So perspective, right? Don't judge a quantum book by its cover. I love that perspective. Now if you go to what you've got now, you got production going on. Okay, so talk about how you see the evolution. You got a little you're going to have commercial products. Are you guys looking at this as a stage kind of execution platform where you have, okay, let's get some chips out there. It may not do it all. It's got a limited qubits, not a million, but that's the roadmap. What is the use case design around the chip? Are there other components? Because the key to all this AI game we're seeing is it's about what's around the chips, right? So you're seeing systems is the new revolution, kind of coming back from the nineties when I was breaking into business in the eighties and nineties, everything's about systems. You mentioned mainframes. So there's got to be a theory and vision around how you see the progression of the chip, its relationship with other systems. Can you take us through that vision?

Pete Shadbolt

>> Yeah. So the unpopular reality, like the awkward reality is that there really is a bare minimum. There really is a bar that you have to exceed, and that's about a million qubits, below about a million qubits, I can actually simulate the quantum computer with a conventional supercomputer. And so those systems are commercially useless. Above about a million qubits, and these are approximate statements, but have remained consistent for about 20 years. Above about a million qubits, then I start to be able to solve problems that we will never solve using any conventional supercomputer that we could ever build. So there's really a very robust zero to one kind of nature. And again, I'd say this is like self-driving cars. Self-driving cars, there was no minimum viable product. They had to get above a certain level of safety and scale and so on before those things actually started to be compelling. And we've tried to be very realistic about that. We're absolutely building a system. It's a supercomputer-like machine. On that point, I'll say, again, when I arrived in Silicon Valley eight years ago, a lot of people told me, never say supercomputer, never say high-performance computer. It's very unsexy, very unfashionable. Everyone's doing apps, whatever. In that time, of course, NVIDIA has gone from a gaming company to an H100 company to now a position where their main product is supercomputers. Essentially, scientific computing is the main revenue source of the most valuable company on the planet. It's a pretty extraordinary journey in 10 years. But yeah, that's absolutely what we're doing. And we're very, very focused exclusively on the big machines. We're not interested in the small machines.

>> Well, it's just great anecdote there about good advice. Don't listen to other people who have not been in the trenches. I mean, high-performance computing was an old kind of way. It's a very niche market, but democratization happened. What Jensen took advantage of, like you said, was just the north star vision, software, hardware, gaming, then crypto, a lot of those crypto companies are now neo-clouds. Right? So why they just were in the arena iterating. So again, the advice is don't listen to anyone, just keep doing it. Right? If you believe it, keep doing it right.

Pete Shadbolt

>> I think, there are two tricks that I like. One of them is to forget all of the good advice that I didn't follow and only remember the bad advice that I ignored. And then the other thing is to, which I think is a huge piece of why we're in this giant AI wave today, where did this AI wave come from? It came from leveraging quite unexpectedly unrelated hardware, right? Gaming GPUs really were the key. Our approach to building a quantum computer is to leverage the silicon photonics technology that was developed for optical interconnects, for conventional data centers, cloud computing and so on. And it's only because that photonics technology is mature enough and that GF could make it that we can actually leverage that to build our machines. Without we would have to be inventing from scratch and we'd have no hope.

>> Yeah, and that's the great thing about community and ecosystem. I mean, photonics on silicon is huge. That changes the game on density, but also talk about the software stack involved because if you look at the old compiler days, you'd have to vertically integrate a proprietary tech stack from everything. Right? Now, I mean, who cares if it's in a hardened top, right? I mean, it's on a chip. I mean, if it goes fast and no one needs to really tinker with it, that's great. You can have a homogeneous system that's contained for certain speeds and feeds. Now, heterogeneous kicks in when you start thinking open source or software layers or whatever, programmability. So how do you guys view that? Because you're going to be building these systems, and I guess the question is, where's your zero to one moment? What's the one? How do you define that zero to one? Because if you're not hitting a million qubits?

Pete Shadbolt

>> Yeah, so it's a great question, John. It's very thoughtful. I think what I'd say is it's not a ChatGPT moment. That's a lazy thing to say. ChatGPT moment for quantum computing. You're not going to have a CHATGPT moment for quantum computing. And that's because the ChatGPT moment was about mass adoption. It was about millions and millions of people around the world directly interfacing with this thing and realizing that it was compelling. I think that the interesting observation for me is that if you think about so many of our legendary companies and the companies that I would expect to be resilient, especially in uncertain times, like those where we find ourselves, TSMC, ASML, SpaceX, NVIDIA, I would say Waymo, these companies... Sorry, TSMC, ASML, SpaceX and NVIDIA, the products of these companies are surprisingly not at all about mass market adoption. The number of people who actually use an ASML EUV machine, the number of people who actually do a tape out at TSMC, the number of people who design a payload for a SpaceX rocket that's a tiny, tiny, hundreds or thousands of people. It's the second or third order products of that that end up changing people's lives. It's the Starlink or the chips that we have in our cell phones or whatever. And quantum computing is the same. It's not about mass adoption. This is a very extreme unreproducible capability that we believe is going to be profound for car companies, materials companies, energy companies, pharmaceutical companies who want to innovate at a accelerated pace. But it's just not the kind of technology where you directly use it in your everyday life. One day I hope that people will use drugs, fuels, materials that wouldn't exist without a quantum computer, and I'll be perfectly happy if they don't even know that that thing was designed on a quantum computer.

>> Yeah, there's the element of a cloud computing kind of evolution. I mean, Amazon Web Services was misunderstood for a long, long time. In fact, Andy Jassy, who I've interviewed many times, now CEO of Amazon, told me many times privately and publicly in theCUBE, if you have a North Star and you have a vision and a mission, you don't need to explain yourself, but you have to be prepared to be misunderstood for a long time.

Pete Shadbolt

>> hundred percent.

>> So I think you're on that. But I do want to say there's a tipping point somewhere. Like the ChatGPT moment happened before the ChatGPT moment when people said, holy shit, this deep learning is amazing. And then the community changed. So there was a tipping point and then the evolution, like a slow roll.

Pete Shadbolt

>> I couldn't put it better. Like Jensen, of course, was selling GPU clusters quietly well before everyone got excited about AI. And I'm perfectly happy for PsiQuantum to be doing similar things without us being plastered all over the newspapers and being on the tip of everyone's tongue. I'm perfectly happy for us to be one of those kind of companies.

>> And after living in Silicon Valley for 25 years, and still there with the studio there and here but I came back and forth. I study all the CEOs, I meet all of them, most of them, if not all of them, all the best ones. But I've been studying Jensen and I have a theory. I'd love to get your reaction. I think he's playing with a made hand, right? He's counting cards because if you look at all those announcements and some of the hyperbole and the mojo, it's as if he's gaming everybody because he's got a supply chain, okay, $250 billion supply. This is just today. Last year I saw his keynote, I'm like, wow, he's basically making moves before he knows what the outcome is, so he's a little bit way ahead of everyone lagging in this video game of life, right? So I think it's clear that, I mean, he's confident, and so all his talks, there's not a lot of... I mean, there's some Jensen kind of fluff in there, but it's not fluff. It's real, right? So everything he says kind of happened. It's almost as if he's playing with a made hand. What's your reaction to that? Do you think he's got the made hand and he knows it and no one else does?

Pete Shadbolt

>> I think. Again, think about those companies, NVIDIA, TSMC, ASML, SpaceX, Waymo. These companies are legendary because they operate on a landscape of extreme technical difficulty. And they were the few organizations who managed to survive on that landscape. And if you're a CEO like Jensen or myself, I of course spend a lot of time communicating the company. It's easy to be confident, it's easy to speak about the company and to define plans for the company when you already have accomplished so many things that skeptics said you wouldn't do. We put superconductors into a commercial semiconductor foundry. We put new tools on the shop floor at GF, we develop high power cryogenics with. It makes my job really easy. And of course what happens is that these organizations compound on that advantage and you got to stay very active and beady-eyed and attentive, but if you do so, you can compound on that advantage and build these kind of organizations that can just bulldoze ahead. And that's what you see, I think, with NVIDIA. That they know what they've got and the onus on them is to continue to exceed expectations and it's going to be bloody difficult.

>> They work hard and they work hard, and they work hard, and they stay true to their mission. Yes, confidence and being a little bit cocky is not a bad thing. If you're arrogant, that's where you lose the game. I want to ask you a few questions. Look at NVIDIA and what you guys are doing. One of the keys to success is kind of who's around you, right? In the game, right? If you look at NVIDIA, a lot of things came together. Like I said, I saw Kevin Deierling there, he used to be the of Mellanox. They did a billion dollars in revenue, but they had InfiniBand, which by the way, they had all the customers. There was no more market. So it's either Intel or NVIDIA is going to buy them, right? So, okay, they were a customer of ours, I got to know them very well. But you think about what InfiniBand did for NVIDIA at that moment, it connects things in a way. Now they've got some ethernet with Spectrumx and whatnot. So again, evolution, but things happen. You talk about photonics. So the question, Pete, is what's going on around you guys that's notable that you're excited about that brings the propensity for discovery, the propensity for innovation where it's moving fast. And so you need to rely on fellow travelers, right? To invent cool shit.

Pete Shadbolt

>> Yeah. So I guess our entire thesis is based on leveraging the miracles that are worked on a daily basis by the semiconductor industry. And we are extraordinarily lucky to be able to use the global foundries fab and to be able to use the broader might of the packaging, OSAT, contract manufacturing, et cetera, ecosystem that has been built up around the semiconductor industry. We hire a ton of semiconductor people. We work with an enormously complex supply chain here in the US and elsewhere, and we wouldn't have a hope in hell of building a quantum computer without that. We've also been very gratified to have spent time with various of these legends of the semiconductor industry. Jensen, , Pat Gelsinger is now spending time with us, and I'm personally super grateful to have gotten the time with these guys. And then at a higher level, we recently announced a billion dollar deal, Australian deal with the Australian government where we're building our first system outside of Brisbane and a $500 million campus in Chicago. So we're getting an amazing level of support for what is an extreme project here. And the only other thing I'd say there is because our requirements are so demanding because we have to perform beyond the state of the art of the photonics industry, we have made photonic components that are beyond state of the art. And actually we're starting to get quite a lot of interest from people who do optical switching, who do transceivers, wanting to look at the things that we've been forced to do out of necessity. And that's pretty exciting position to be in.

>> Yeah, I love the whole point about technical challenge because you just stay on that and keep it going. That's where the rewards are. And there's also scale involved, right? Talk about differentiation, competitive strategy at scale is a whole nother level of new MBA discussions for the classroom. If there's still MBA classes are out there. But that is the business model transformations happening. We're seeing already here in robotics, like the physical and digital convergence, okay? I mean, that's a hardware software integration. Again, open source brings more goodness in, right? So as people work around the world, this is key. Now I want to ask you, well we have limited time left. There's a little bit of time left. You mentioned out of the lab and into the fab, I love that line. But you also said before you came on camera, out of the research academic to lab to Fab.

Pete Shadbolt

>> Yes.

>> So Quantum has been grounded in academics, rightfully so, because there's a lot of theory there, right? So it's been out of the academic and into the lab for what, maybe about a decade, so to speak. All right, now, fast-forward past three years. I mean, I remember Amazon wasn't even doing anything with Quantum until four years ago. So past four or five years has really been kicked up into how do we commercialize this? So how do you explain to the folks that are still in academia, get out of there or stay there and do something, be your role here, stay in your lane, contribute to the lab. What does the value chain look like? If you had to kind of break down the swim lanes between role of academic, role of lab, and now obviously you guys are pioneering this commercial push acceleration into the fab.

Pete Shadbolt

>> I'd say, again, very strong parallels. It's a great question. Very strong parallels with what happened in AI. With AI, of course, you can't have a good AI company, AI startup, without some of the most elite, mathematically oriented, academically rooted people that you can imagine, like PhDs who would normally be in a research group in a university. You must have those people in your company. We're no different. We've got lots of those kind of people here. I think that where the space has been slow to move is really on the hardware. It's experimentalists like me who come from a background of tinkering in a research lab. And there's a very unfortunate temptation to continue to tinker, to continue to do interesting and impressive demos. And PsiQuantum has always pursued a strategy that is best described as extreme in that we always were completely impatient to get away from that, to get into a commercial foundry, to figure out how we could manufacture scale and to do anything it took to get away from the research lab and we're occasionally accused of arrogance or of being deranged or a rebel.

>> rebel. You got to get to the fab. I mean, come on, that's the game starts there. It's the arena you want to be in.

Pete Shadbolt

>> And all I can say is that that has served us very well, right? Like taking those risks, avoiding too much incrementalism after a lot of blood, sweat and tears, eight years of work and hundreds of millions of dollars spent, we now can manufacture all of the components that we need beyond the state of the art. We've still got a lot of hard work to do, but to me, I'm so glad that we invested our resources, our time that way. And I mean, I think the only thing, to your question, that I would want to convey to my academic colleagues or to the industry more widely is that I would love if I could transplant some of my self-confidence into those teams, right? Self-confidence comes with risks, but insecurity comes with even more profound risks. And that's under-reported in my opinion.

>> Yeah, well, we'll certainly amplify the enthusiasm and confidence because you guys have made a major breakthrough getting into the fab. Again, that's just the arena. Welcome to the game. Now you got to kind get to the next leg of the journey, right? Which is, how do you bring along the ecosystem? Your smartest friends and other collaborators and-

Pete Shadbolt

>> Hundred percent....

>> There's still more breakthroughs to have and we really appreciate you taking the time. We'll do another one, certainly in Palo Alto. We have a studio there.

Pete Shadbolt

>> Palo, I'd love that.

>> You can come down, we'll do a full deep dive, but I really appreciate your time and I know you're busy.

Pete Shadbolt

>> This has been a joy, John, I really appreciate you guys having us on, especially on a crazy day like today, and I hope that you get some rest and a absolute pleasure speaking to you.

>> Cheers, thank you very much.

Pete Shadbolt

>> Anytime. Consider us a neighbor.

>> Okay. Okay, this is theCUBE here on the New York Stock Exchange. I'm John Furrier, your host. This is theCUBE and the NYC Wired. It's an open community of leaders. It's a trust network that's building through collaboration and contribution of knowledge and information and sharing, an open marketplace of ideas. This is where the future breakthroughs are coming from. Collaboration, people on the fellow traveler road together. This is where it's all happening, certainly on theCUBE and the NYC Wired. Thanks for watching.