From Google’s Lab to the Nobel Prize: John Martinis on the Quantum Computer Breakthrough

Google actually contacted me at the time. They had started to work on quantum computing. The big application that people are worried about for national security is using a quantum computer to factor large numbers. So I'm going to say if there was something that I thought was a huge breakthrough, say coming from China, I would talk to someone in the U.S. government to make sure they knew about that. Today, welcome back, Dr. John Martinis, recipient of the 2025 Nobel Prize in Physics and one of the architects of the quantum computing revolution. In this episode, we fast forward to the present day, to Google's Quantum AI Lab, where John led the team behind the landmark 2019 achievement known as Quantum Supremacy, demonstrating that a quantum processor could perform a task beyond the practical reach of the world's most powerful classical supercomputers. We'll discuss what the breakthrough really meant, separate science from fiction, explore the future of quantum computing, and hear how perseverance through setbacks ultimately helped shape one of the most transformative technologies of our time. This is Bay Area Innovators, and I'm your host, Steve Vispas. Dr. Martinis, welcome back to Bay Area Innovators. Thank you. Well, thanks for having you. We talked last week, and you're in Santa Barbara right now, so at least you're on the same coast as we are here. We're in Santa Clara, of course. Now, we talked about your Nobel Prize last time, your experiment, and what the foundational stages were, and this time we want to talk about the applications. And, of course, you were at Google, and in 2019 you had this amazing breakthrough with a quantum computer. But please walk us how you got towards being at Google and towards working pretty much on your foundational inventions. Well, we can just go back to my thesis in the mid-1980s. I think one of the important things that we learned during the thesis is not just demonstrating the experiment, but we learned about the technology of building these quantum devices, that we were able to relate quantum mechanics with microwave engineering. We learned about filtering, and what really happened over the subsequent few decades is that myself and many other people around the world continued to work on this, and I saw that it was a very fascinating subject, and developed the technology and the physics and the understanding to build more and more complex devices. The big event really happened in the 1990s with Peter Shore thinking about the factoring algorithm and understanding how to do error correction. Theoretically, this made all the difference. And because of that, the US government started seriously funding quantum computing in the mid-1990s. And when I saw that there was funding available and I was doing this kind of thing anyway, I really got interested because it's very natural for me to want to think about building a quantum computer, not just do basic research, but actually building the quantum computer. Because that's the kind of way I am. I want to have a long-term goal, and the way I can kind of figure out whether the important physics questions to ask is the thing about a long-term goal and application. So that's what I did. So I was at the time working at NIST in Boulder, Colorado, down the hall from David Weiland, who was working on these great qubit, ion trap qubits. So I got to learn about this from his group and develop that. And then in the early 2000s, I decided to move to UC Santa Barbara. And they had really good fabrication facilities there, so that was able to set up my lab there and continue just to make progress, as many other people around the world, in doing that. And then in 2010 or so, we started understanding error correction better and something called the surface code for doing error correction. Let's just clarify a little bit on that. So the error correction is needed because the results you get from the quantum computer have error. They're not as definite, right? They vary. Using my analogy with the cell phone, zero or one is naturally stable with a little bit of noise. However, if you have the cell phone kind of floating in space, you give it a little bit of noise, it rotates a little bit, and that gives you noise. So what you do is you encode what they call one logical qubit in many physical qubits, and then they look at the correlations between all the qubits and figure out what all the errors are. And this is the big problem with the quantum computer is you have to fundamentally make your errors low and then you have to be able to correct it in a way. And what happened around 2010, we started to understand how that happened. And in fact, there was something called the surface code that was out there. It was really very difficult to understand. For me, it was like reading string theory. I just couldn't understand it. But we worked with someone and we actually wrote a paper written by Andrew Cleland, my collaborator professor at UC Santa Barbara, which kind of explained how it worked. And I think this was kind of a nice paper because it was the first time that we went through and explained it in a language where, let's say, a graduate student could understand, certainly I could understand. And then from that, it was very clear what we had to build. And so I was getting quite interested at the time to figure out how to scale up and build this. And, you know, through a variety of events, Google actually contacted me at the time. They had started to work on quantum computing. They bought a D-Wave quantum computer and was using it. It was interesting but not quite good enough to solve real problems. And they thought about building, putting together, building an experimental effort. So I talked with them and decided, you know, that would be the next good way, step forward. So I basically moved my research program from the university to Google, which was good. First of all, you know, all my postdocs and students, instead of moving on and losing all that knowledge, you know, they could stay on and build this. And also with the resources at Google, we were able to accelerate how we would go ahead and build the quantum computer. So it was really a great thing to do. And then as we were doing this, we, with the theory team, we figured out how to do something called this quantum supremacy experiment, which would demonstrate that a quantum computer could be powerful, not do anything useful yet, but kind of solve a mathematical problem that would be much, much faster on a quantum computer than a classical computer. So that was, you know, really redirected everything into doing that experiment. And that's what we worked on for a couple years. I was going around talking about this at the time. People thought that it would take 10 years to do. I had seen things internally and thought, well, you know, it could be a lot faster if things work out well. And the team worked very hard and got that to work in 2019. And we published that at the end of 2019. And it was a big, a big advance for the field. And basically it showed that the promise of the quantum computer, being able to do a powerful computation with a quantum computer, much more faster than a classical computer, that that idea was really correct. And, you know, we showed with data, you know, and we showed with a classical algorithm how difficult that was, how long it would take, and the quantum computer was much better than that. So just to put that into numbers, I think you had a 53-qubit super quantum computer at the time, and it did a calculation that would have taken the fastest classical computer at that time, 10,000 years, and it was done in about three minutes, something like that. Yeah, those are the numbers, and that's how much faster it was. Now, the interesting thing about this is the Google group worked really hard to figure out the fastest classical algorithm they could think of. And then using kind of those techniques, theorists in a couple of years, it wasn't like the next month, it's a couple of years later, they figured out how to generalize that and make the classical computer computation better. And I think at the end they got the time down to, you know, minutes, a very short amount of time, but it took a lot of theoretical work. But of course, you know, we mentioned this in the paper that this would probably happen, okay? It took a few years. And then what we did is we actually published paper, in the paper, a data set where we couldn't check because we didn't have a classical computer that was good enough to check it. And then with the new algorithm, they checked it, and then our claims, how it would behave, were verified. So that was kind of nice. And then, of course, with the quantum computer, it gets exponentially more powerful as you add more qubits. So in the subsequent years, people have made it bigger and then made it so that you can no longer compute that anymore. So I think that there was a nice paper in that we did think about the future and, you know, opened up the possibility that someone could do it better. Can you give us a sense of what, and of course we'll show pictures, but what a quantum computer looks like and what the separate parts of it are? I'm assuming you still have the algorithms, you have the keyboard that you enter things, but what's the heart of it look like? Sure. So at the heart of it, for our superconducting qubits, are these integrated circuit chips which have aluminum metal on top of it that define the circuits. And you also have aluminum, aluminum oxide, aluminum tunnel junctions, which are kind of the active part. So that's the chip itself that does the computation. Now, a quantum computer is an analog computer in the end. So you have control lines that come down that change the qubits to do the computation. Then you do a measurement and then those lines come up out of the computer. So what we have is outside of, and then this is of course cooled to a very low temperature with what's something we call a dilution refrigerator cryostat. The cryostat is complicated in its own sense. And then we have a lot of wires that connect the electronics at room temperature connected through wires going down to the cryostat, going through the chip. So you have, you know, a bunch of wiring you have to get right. There are some microwave components that are gold that you have to get right, mostly to amplify the signals when you do the measurement. Lots of room temperature electronics that convert a digital waveform that's in your computer to an actual voltage signal that comes out of that. So then you have the electronics that run the digital analog converters and then a bunch of software that tells the digital signals what to do. And then software on top of that that takes your algorithm and compiles it down to that. So there's a huge stack of things you have to do to get everything to work properly. What would be the cost of a current supercomputer? What would it cost to make it or to sell? Well, I think the biggest ones they're selling for something like $50 to $100 million for, let's say, a hundred, a few hundred qubits. And given that they're unique instruments and a few people can buy that, then that's what it is. Well, our particular job in our company is to look at that and to simplify everything to try to bring the cost down and to make the quantum computer affordable when you're going to have thousands to millions of qubits. So cost is a big issue, yes. Now, what about national security? How are these protected from getting into the wrong hands that could potentially use them for doing bad things? This is a deep issue. First of all, the quantum computers that we make right now are only for scientific purposes. People run demonstration algorithms, they see how it works, they do research papers in order to do something more deep. The big application that people are worried about for national security is using a quantum computer to factor large numbers. So, for example, the Internet security that we use now is based on RSA protocol, which is based on the fact that if you take two huge primes, it's easy to multiply them together, but hard to get that resulting and factor it into the two primes. And, you know, that asymmetry, you can build a secure protocol. Now, eventually, a quantum computer can be made big enough, being able to do the factoring part of that, in which case you'll break the RSA. And this is what people are worried about for national security. There's also something called elliptic curve codes that have a similar way that it can be broken with a quantum computer. They're used often, they're used for Bitcoin, and that's going to be a little bit easier to break than RSA. So these are very widely used, important applications that, you know, people are worried about a quantum computer breaking, and obviously this has national security implications to it. What are the biggest challenges in building a quantum computer? We talk about the refrigeration part, and that's probably very expensive, right, to keep it that cold, and then the error correction. Are there any other challenges for building a quantum computer? So those are obviously obvious challenges, and what I would say is the biggest challenge is that there are many challenges, and that the many challenges tend to push against each other, so you could make one of these items really good, but then something else is not so good, okay, and it limits you. And, you know, you think about your cell phone, the number of technologies that are in here to get it to work well is absolutely astounding, and, you know, a quantum computer, it's no surprise that you have a similar large number of things to worry about. So it's not just one thing or two things, but it's many, many things you have to get right, and get it right simultaneously. So part of the confusion, I think, that are in the field is people talk about quantum computers, and they say, oh, we're making our quantum computer a certain way, it's really good, we can make a lot of progress with that. But what they aren't talking about is what's bad about their approach, okay, and what are the challenges and what are the things that are going to ultimately limit them. And for any large computer, large systems, it's usually the bad things, the problem things that limit you. And maybe now, you know, you can do a little bit better, and that's good, and you see that, but this is the problem. So, for example, with the superconducting case, going to cold temperatures is a technology you have to master, but it's something you kind of know how to do. You can buy these refrigerators, you can modify them, it's okay. Other kind of qubits, they have a problem of ultra-high vacuum. Okay, and one might argue that's easier to solve, but, okay, there are different problems that you have to have. I would say, what I do very differently, on most people, is I will explicitly talk about the problems with my qubit. And the problem with my qubit is scaling it up to thousands and millions of qubits. Right now, we can make a few hundred qubit systems, but if you look at that, there's a lot of wires and connectors and whatever, and what you really have to do is do what was done with CMOS, is to integrate it together, put it on a chip, make it more manufacturable. And that's what we're doing. Other people are working on that in their own ways, but, you know, it's what are you going to do to solve the hard problem for your qubits that, you know, I generally want to know about, because the reality is that engineering is solving the difficult problems, not just saying that it's good for one reason or the other. Now, I want to touch on AI. Of course, AI is growing very fast in so many ways, and also concerned about growing it too fast and being, you know, issues. Are there any parallels between quantum computing and AI or any collaboration or anything that would be similar as far as concerns? Yeah, I think it's similar in the sense that, you know, you see the trajectory of AI, and there's a trajectory of quantum, and we're just, you know, X years behind that trajectory. Or, you know, at least you can argue that. And I had the very fortunate experience to be at Google in the, you know, 2010s, where, you know, they were working at AI and using AI, and towards the end, they were really emphasizing that. So I saw the promise of AI internally from Google, who was the leader at the time. And then, when the Transformers paper came out, and people had an idea to build these really big architectures, that was mostly, it was discovered at Google, but then really taken advantage at OpenAI and elsewhere. You really saw that come to fruition. And then, of course, everyone is working very hard to do that. So, we're still at the stage that AI was, I don't know, 2000s, 2010, where it looks very promising, people are building systems, but, you know, it hasn't become powerful enough, exactly all the right ideas in order to make it a compelling technology. Now, we all believe that's going to happen soon, and, you know, there's a lot of projections, and people think they will do it. I think it's going to take a little bit longer than people think, from the system engineering point of view, but, yeah, that's what, you know, people are excited about it now. I know you can very much imagine it will take on a similar trajectory. Now, we talked a little bit about the applications of the quantum computer. Can you tell us more what else could be used for? And, from what I understood, some of the calculations, or, you know, the limits it has, it's outrageous, it's amazing. You know, it does calculations so fast, but what can that actually be used for in real life? So, first of all, I'm an experimentalist who's trying to build the quantum computer. I'm not the expert on the algorithms, but I can say a little bit about that. There are some people who are saying that computers, kind of at the size right now, can run kind of approximate algorithms, but well enough doing something useful, and I think there's still some debate in the community on whether people are doing that and whether they're comparing it to the best classical algorithms and what kind of results they're getting. So, I'm just going to leave that aside, because that's a big debate. I'm talking about making a big, general-purpose quantum computer that can be like the computers we have right now. It looks like the easiest algorithms to run is the RSA factoring and the error correction codes. I'm going to leave that aside. I want to do something useful. For me, as a scientist, one of the really fascinating algorithms are for quantum chemistry and quantum materials. The basic idea is that quantum chemistry materials are based on quantum mechanics, and that's what makes it so hard to solve those problems on a supercomputer, but there have been a mapping. People understand how to map the quantum mechanics of a molecule onto a quantum mechanics of a quantum computer. So, I'm really interested in that, and it's really interesting because you can combine the quantum computer with classical AI techniques to kind of combine them in the right way to take advantage of both technologies and to solve practical problems. So, there's a host of things that you can do with that, you know, making better materials, better chemical processes. People talk about protein folding, things like that. The one example I like to give is people are very excited about electrification of transportation, and to do that for the motors, you tend to need these rare earth materials, which there's a supply chain issue with that. And mining it, you have to worry about that. And, for example, if we can figure out how to use a not-so-rare earth to build these magnets, then, you know, that would be a big benefit to society, worth a lot of money, but importantly, help everyone, make things cheaper, more economical, more ecological. I really like that because, you know, we use computers all the time for doing design work, building electronics, circuit boards. My wife and I remodeled our kitchen and did that. If we can design, you know, quantum mechanics design using a quantum computer to make it more efficient, we have lots of applications. People also talk about optimization problems and how to optimize, you know, certain business tasks better. There's less certainty on the algorithms there, but, you know, it could be more that just people find the algorithms there and they find that it finds useful solutions and then it's good for that. So, there's many things we think you can do once you build this very powerful computer. Now, you have your own company, qolab.ai. Please tell us about what you're doing there and, obviously, how it's, you know, related to quantum computing. So, the company is pronounced qolab and it's q for quantum, of course, plus collaboration. And our basic business idea is to use the expertise in the semiconductor industry to fabricate our qubits better, more reliably, make it scale, make it cheaper, and to use those foundries. Most of the companies right now build their own clean room and make it internally, which is a really good model, but we think in the end to bring the cost down you want to use these outside companies. So, in the beginning of Series A, we've been working on improving qubit fab and especially to get rid of the way we make these Josephson tunnel junctions. It's used, it's made right now with a technology called Liftoff, which has been around for a long time, used by physicists all the time, been around, I don't know, probably since the 50s or 60s, but no one in the semiconductor manufacturing used that to make their transistors anymore. It's just, you know, it's a dirty process. It doesn't yield. So, we need to get the process more reliable and bring the yield up. The second thing we're working on is how to put it together in a way where you wire it all together just using integrated circuits. So, it's integrated circuit fabrication. And that's, you know, kind of what's done is why cell phones and computers can be made so cheaply as you do everything with an integrated circuit. So, we've redesigned the whole thing that everyone does right now to be able to integrate it on a couple of wafers that you have to bump on together. And that way you can make a very compact, much more inexpensive quantum computer that way. And, you know, we're really thinking about what does a quantum computer need, not next year, but, you know, five years from now, seven years from now, when we're really trying to scale it up and build a big system. And you would build these wafers in the US or outside of Taiwan? So, we are building the qubit wafers in the US and I'm going to say it's because we have the expertise but also we understand the US government wants us to do that. Some of the subsidiary wafers, which is just wiring, metal wiring, we can do that, we think, outside of the US, potentially with our, you know, good trading partners that we trust, although it's not quite clear right now, you know, where we're going to do that. But we can definitely open up the supply chain for a lot of the components to our trading partners. Next, I want to ask you your advice to young students, to kids who are interested in physics, and with that, I would like to ask you to please show us your medal, the medal you got from the Nobel Committee, which is really the pinnacle of what any student or any scientist would be aiming for. Sure, sure. I have this here in case people ask. So here is the Nobel Prize in the case. Okay, I'll turn it on its side so you can see that. So it's actually better than I thought, so it's a very nice award. What I want to say about the Nobel Prize is to be awarded the Nobel Prize is a big honor and it changes your life clearly. I enjoy the part of it where I get to be a science diplomat and talk about science to the general public, but especially students. And I find actually the students have some of the best questions and they're thinking about it and maybe how it relates to their career and the like. But given all of that, what I would say is you do science for the actual performing of the science, asking a question about nature, maybe not, you know, nature can go different ways, and finding out from nature the way it works, and then trying to understand that in a clear way. So just the joy of the discovery and then you get to write the paper and communicate this, and then you get to give talks to people and communicate, and all of that is just a real pleasure. And then you get to do that over and over in your career, and that's been you know, that's the real amazing part of this. And, you know, to be recognized for doing something transformative is fantastic, and I appreciate that. Of course, many people were involved in getting this field to where it's at now, so you know, I just happened to be fortunate to be at the right place at the right time asking the right questions and doing the right experiment. But, you know, you feel very fortunate about what happened. And then, just for students, I would say, what I try to tell them is, you know, just take the intrinsic joy of doing science and, you know, do experiments, ask good questions, do experiments, communicating it well, and then, you know, a lot of times what you're doing is kind of incremental and trying to understand something better and better, but always keep your eyes out for some new idea. You know, in our case, I listened to a lecture by Anthony Leggett and learned about the idea, thinking about new things, and thinking about something that, you know, maybe most people say won't work or doesn't make sense, but maybe you have a certain idea why it makes sense, and to keep an eye out for that and think about it, and then, you know, do some kind of crazy experiments or very, you know, hard experiments because if they work they could be very transformational and really change science, and, you know, just keep a lookout for that. Can you talk a bit about collaboration at the level, at your level or before when you were, you know, doing the experiments, collaboration and working with other scientists, how open are more scientists to this collaboration given the fact that someone else may steal an idea or use it for their own research or whatever? Yeah, collaboration is very important because that's how you, you know, get things done, and, you know, I had the distinct privilege to work with Michelle Devereux and John Clark and really learned how the collaboration worked, and there I really learned that working with other smart people is way more interesting than working by myself, which is what I had done prior to that, and, you know, working with the right group and right people is very important. But, you know, given that you're dealing with human interactions and emotions and whatever, you know, like you say, you can't, it's hard to collaborate, and you know, there's a certain amount of competition of different groups doing different things, and you're kind of talking to each other and informally collaborating that way, but you're also competing against each other, and, you know, that kind of dynamic is also very interesting to figure out how to do that, and the competition is a good part of the whole system, as well as the collaboration, because with the competition, different people have different ideas, some ideas that maybe you wouldn't have thought about or don't think are good, but then they do the experiment and you learn from that, and I would say what was really good about our field in the 1990s and 2000s, 2010s, is that people tended to do different things, different groups, and we weren't trying to beat each other by a month to publish a paper. We were all doing different things, and we learned from each other, and that's why, you know, we took the best ideas from each other, and that's why we were able to make a lot of progress. So, you know, that, to me, was really great, what happened here. But in the end, you know, egos can get involved, and, you know, things change, and, you know, that's not so good. Well, come on, we're all human beings. Of course, that's going to happen. Well, Dr. Martinez, we thank you so much for your time. One final question, however you are comfortable to answer. I wonder, the work you've done, and you're doing, it's so breakthrough and so important in so many ways, and also raises national security. Are you working with the U.S. government or the intelligence agencies for that matter to monitor that, or to say, you know, if you come up with some amazing things, you've got to let us know. How is that work? Can you talk about that? So, I'm working by myself and keeping track of things. I'm going to say, if there was something that I thought was a huge breakthrough, you know, say coming from China, you know, I would talk to someone in the U.S. government to make sure they knew about that, and then I've, you know, in conferences when I give my talk, I tend to be very open about what I think we need to do. I've been very fortunate with the Nobel Prize. I was put on the Presidential Commission of Science and Technology, the PCAST, which is an advisory committee to the President. We haven't met yet. Most of the people there are AI kind of businessmen and leaders, which I think is great. That's what it should be. I'm probably there because of the quantum, and also I'm an entrepreneur also. So, you know, I'm happy to serve in that way. We haven't done anything yet. I think President Trump has been worried about some other problems. Hopefully that will be over with soon, or is over with, and then we can work on these other things. But, yeah, I'm happy to help the government when appropriate. Well, it's so refreshing, you know, we've done a lot of programs here, and almost every single technology or company or startup has to do with AI, and it's great to see that that's not the only thing that is happening. You know, computing is so important for our lives. It's AI and other things, and it's just amazing that there's a whole new way to do computing that could be more powerful, and it's great to, you know, that companies, the governments around the world are investing in this. I would just say as a basic scientific question to understand that, and of course we're all trying to do something useful, but, yeah, I think it's an exciting time that we can work on something so technologically important. Dr. Martinez, thank you so much for your time. I really appreciate it. Thank you. Thank you. Thank you for watching another episode of Bay Area Innovators. If you enjoyed this episode and you know someone who should be on our show, please send us an email at sfbay.epochtimes.com We would love to hear from you. Thank you, and see you next week.