Transcript

Brian Keating:
Today we’re joined with one of my heroes. It’s not every day you get to talk to somebody who not only, you know, is such a great contributor to physics and has been inspiring not only the work that I do, but my whole collaboration is basically enabled by work that John and his collaborators did over the years. But he’s also, you know, once we say a mensch, he’s known for his teaching, for his group building. You’re just kind of the physicist physicist. So thank you for joining us.

John Martinis:
Well, that’s very kind. Not everyone feels that way, but I appreciate the kind words.

Brian Keating:
You’re now The 26th Nobel Prize winner I’ve had on the podcast. Every, every 9 multiples of 9, I write a book featuring wisdom, and I hope the, the 3rd version will come out with you in the coming years. Well, I’ll let you know about how that progresses, but I wanna take us to that October morning last year. I was teaching quantum mechanics that day, advanced quantum mechanics with perturbation theory, and I said, a guy I know just won the Nobel Prize. What was that like? What was the, what was the 2nd thing that went through your mind when you got the phone call at whatever time it was?

John Martinis:
Well, I actually didn’t get the phone call. The phone is way across the house, but my wife was up late reading and she heard the phone ringing, but she figured she’d get it in the next day. She looked at her email and there was a bunch of congratulations. So she knew about it, but she knows that I need my sleep, especially, you know, the next day when you, you know, you have to be on the whole day. So she waited till about 6 o’clock. There were reporters who showed up soon after that. So it was actually better because my wife woke me up, but she was very clever about it. She tapped me on the shoulder and said there were reporters coming over.

John Martinis:
And I realized, oh, it’s the beginning of October. So I opened my computer and, and, you know, saw Let me share something a tiny bit personal. There are things called Nobel symposiums where they look at a field and see if the field is noteworthy and look at people in the, you know, leaders in the field and the like. So you kind of understand a little bit that you might be on some kind of list. Okay. For me, that’s the biggest honor because the Nobel Prize is just so crazy unlikely, right? Just being invited to that is really very special. And And then for some years, you know, I’d wake up in beginning October. It’s like, oh, okay.

John Martinis:
And it’s just so wrong to be disappointed by this because, so after the years, you know, I just stopped kind of, I knew, well, it’s beginning October, but I stopped really thinking about it. That’s why my wife didn’t know. We didn’t talk about this at all. Right. Because it’s just, okay. You know, whatever happens and it’s actually better that way. Yeah.

Brian Keating:
There’s a joke I sometimes will, you know, pull out on October 1st at 2 in the morning. I’ll say, I’m working on my best, you know, Swedish accent to cause somebody I don’t like a heart attack. We are calling from the Swedish Royal Academy.

John Martinis:
Yeah, I’ve heard people get pranks like that.

Brian Keating:
Well, yours isn’t a prank. And the only thing that frustrated me is that I was talking about, you know, quantum tunneling, but I was talking about single electron tunneling and even nuclear tunneling. So you wanted for macroscopic tunneling. What was the impetus for— obviously you didn’t set out to win a Nobel Prize. Talk us through the history of why you thought macroscopic effects would manifest themselves instead of just the already mysterious microscopic tunneling effects?

John Martinis:
Yeah. So this experiment in line of research was very much motivated by Anthony Leggett. You can just go back to the Schrödinger cat paradox where you do a microscopic atom decay that’s connected up. So the atom decays, it kills the cat, and then, you know, okay, before you open the box, is the cat in some dead and alive state? And okay, I think there, for me, there are ways to answer this that are very sensible. But Leggett pointed out that there actually is no experimental evidence that macroscopic entities, especially a cat, can obey quantum mechanics. And he said, there’s— if there’s no evidence, we should be looking for evidence as a way to test quantum mechanics. And he came up with the idea of these superconducting circuits where you have a macroscopic number of electrons that are tunneling through the junction. For me, you know, as a young student, you look at quantum mechanics, it’s really wonderful.

John Martinis:
My personal hobby was electronics. I joined John Clarke’s group because he was doing things on quantum noise. Okay. And thinking about this. And for me, it was the most natural experiment to want to do. I’m surprised that there weren’t 30 other groups doing it. At the point now, everyone can understand that. But back in the mid-’80s, the idea of quantum information and doing these tests wasn’t as popularized as, as it is now.

John Martinis:
So for me, it was the most fascinating experiment. I thought it could be done cleanly. You know, obviously John Clarke, you know, was moving in this direction. Michel Devoreg came over. And for me, it was, you know, a perfect thesis experiment. And I’m going to say it was also very groundbreaking because Then John’s group, you know, we understood how to measure noise in these devices, but to, you know, build a system and engineer it and think about the physics combining microwave engineering and quantum mechanics and figuring out how to deal with the noise and the like to build a clean experiment was just really fundamental and groundbreaking. And of course, that’s, you know, I find that, you know, wonderfully exciting. Okay.

John Martinis:
To just try to figure all that out.

Brian Keating:
Why did you start off with a Josephson junction? Maybe first we’ll explain what it is. And keep in mind, my audience is highly technical, very competent, and I’ve tried to have Brian on the podcast as a fellow Brian. He’s gone in interesting directions, shall we say, in consciousness and other kind of fields, very different from what he did as a young, very young man when he won the Nobel Prize for his work. But tell us, why Josephson junction? Why not a quantum dot or trapped ion? What made you start with that?

John Martinis:
Well, first of all, the Josephson junction is a macroscopic system where you have a macroscopic number of, you know, Cooper pairs, electrons, paired electrons tunneling through the device. And to test the idea of whether macroscopic variables obey quantum mechanics, you need that. A quantum dot and an atom, they’re kind of single atomic systems, whereas this was very clearly macroscopic. And, you know, that’s why Leggett proposed it, and that being in the superconducting field with John Clarke’s group, that’s of course why we did it. Now at the time, we didn’t know if the Josephson junction was a clean system. In many physical systems, especially a macroscopic system, there can be dirt effects, other things that go wrong that, you know, may cause it not to work as well as you would like. And, you know, at the time we didn’t know any of that, so we just forged ahead. But I would say the reason I’m talking to you today about the Nobel Prize is it turned out that it was a clean system.

John Martinis:
And this was figured out by many, many people over decades of work, you know, testing it. I mean, we laid the foundation for understanding how to do that, but building these systems properly, you know, took a while for a whole lot of people to figure out.

Brian Keating:
What strikes me as so, you know, kind of magical and beautiful is that you guys ended up seeing you know, discretization and quantization in sort of spectral features. And it’s exactly reminiscent of what Balmer saw, you know, 60 years before the invention of quantum mechanics. Have you ever thought about what it was like to be him? I mean, grappling with something that wouldn’t be determined or even predictable for many decades. Did you ever have that kind of inkling of like, we’re in this weird, mysterious territory? Or was it, if we keep putting one foot in front of the other, we’re going to get to some goal that we’ve set out for ourselves?

John Martinis:
Well, that’s a really great question because The way we were thinking about it and we started it, John Clarke’s a serious experimentalist and I was a new student. But what he and we decided very early on is we had to measure the parameters of the system in order to do a careful test of the theory. And the obvious thing to do was to put on microwaves and have a resonant phenomenon and see that resonant phenomenon as a way to measure the oscillation frequency of the system, which is a very fundamental parameter. And what was interesting is classically, I set up an analog simulator in about, I think, a day or two and put in noise and then showed that indeed you can see some kind of resonant effect. So very early on, even though we knew about that, we had to do tests in doing it. Now, of course, you know, as we did this, we were quite interested in what would be the quantum effects. Now, what happens classically, both in an atom and in our system, is that this oscillates in this nonlinear well. It has a spectrum of frequencies, one frequency when it’s low and a lower frequency when it’s high.

John Martinis:
So just like when people talk about electrons circling the nucleus, classically, there’s a range of frequencies that it would emit light over. Of course, what you saw was quantized energy levels and quantized light, which, you know, it was clear we could see in the system. So I would say the fact that you see quantized oscillations is a key feature of quantum mechanics. And I think that what we were able to see is like the smoking gun thing that you’re doing that. Now, it could be that you have to realize, if you’re really being careful about this, it could be that there’s resonances in the circuit you’re connecting it to. So you have to be careful about, you know, claiming that. And that’s why we did a bunch of experiments detailed that are in the papers, but no one ever talks about to say that we had a well-defined experiment without these resonances. The resonances we saw made sense in terms of the physics of the, of the system and the like.

John Martinis:
But in the end, that was a key observation. Now, of course, what was also interesting is Leggett talked about what happened to tunneling when you had dissipation. That’s very new physics, which you normally don’t see in electrons and nucleus. And eventually we were able to do those experiments really well and show that that theory made sense too. So there was a whole series of things we were able to do because it was really well engineered, good connection of microwave engineering to quantum accounting.

Brian Keating:
One of the things I love to point out to my younger listeners, readers, and viewers in my books when I interview titanic physicists such as yourself is when you do the thing that you really didn’t think you would have to do. You just mentioned it. You can have this collective phenomenon from alternative effects rather than the thing you’re looking for, and those are called systematic effects. And I typically tell my students, anyone can get the right answer. We can get the Hubble constant, and it could be beautiful, but the real determinant of whether or not you’re a good scientist is how you account for the things that you could be wrong about, right?

John Martinis:
Yeah, exactly. You know, and the fundamental thing about nature is you can never prove anything about nature. You can disprove your theories, but you can never prove it because something else could be explaining it. But what you do as an experimentalist is measure enough parameters and do enough of the check experiments that if it’s another theory, it would have to be really kind of You know, a crazy, not simple theory. And, and that’s what you do. And that’s, for example, why we were measuring the parameters. Okay. The other thing to, to realize, and I talk about, I like to explain this to the students, is in the beginning we did some experiments at, you know, 1 Kelvin to 4 Kelvin to see what was going on.

John Martinis:
And the data didn’t make sense at all. And it was because— we were seeing noise. And very early on, we just did something. We compared it in a way that was really kind of tricky at the time and didn’t make sense. And then we started saying, well, okay, we’re going to have to filter it right and the like. And once we understood the microwave engineering, microwave filtering, we redesigned the experiment pretty quickly. And then all the data started making sense. Okay.

John Martinis:
So there were some internal checks that you, you know, you have to look at very carefully. There were a few prior experiments that hadn’t really done that properly. And I think the physics community appreciated all those checks and being able to do a beautiful experiment because of that. And, you know, all good experiments are like that. They’re well-designed, and then you think about things can go wrong, and you figure that out.

Brian Keating:
Yeah, it reminds me, there was a scientist named Ed Ohm who worked at Bell Labs on the exact same Holmdel antenna as Penzias and Wilson. You probably know this story. And he actually measured the BMB, and he attributed it to a systematic error. He basically said it’s a systematic error, or it’s excess noise, or all the atmospheric contributions add cumulatively and they don’t cancel out. And then Penzias and Wilson said, well, let’s do a calibration. Let’s measure that with a liquid nitrogen chopped Dickey switch load.

John Martinis:
Yeah, yeah, yeah.

Brian Keating:
That won them the Nobel Prize. But he had actually discovered it, you know, 3 years earlier in the same data. But that’s, that’s exactly right. And what you said is so important that we can’t prove things on the physical side. We’re not mathematicians. You know, mathematicians can prove 1 1 2. It takes 200 pages of piano algorithms and all these other things. But it kind of reminds me of what Eugene Wigner, another Nobel laureate, said once.

Brian Keating:
He said the mathematics is sort of unreasonably effective. And, and I think about that and I kind of have narrowed it down. I say that the square root is unreasonably effective because in classical mechanics you can make the Poisson bracket, you can make the commutator of momentum and position, right? And it’s zero, right? It doesn’t matter if you measure momentum first or position first, you get zero. But if you add the square root of negative 1, You get the Poisson bracket for the Heisenberg relationships, and those do not commute, right? What’s the weirdest thing about quantum mechanics to you?

John Martinis:
The weirdest thing. It’s really a complicated phenomenon, and it takes until at least your third year, typically, as an undergraduate, and then you take it more. And I’m going to say, after doing this for many decades, I kind of understand it fairly well at this point. But You know, it’s complicated, but at the same time, there’s this mathematical artifice where you can understand it well. And, you know, it’s the basis of many fundamental standards. So it’s extremely accurate too. It’s both complicated but understandable, unintuitive, but given enough time, to me, it’s intuitive right now. It’s kind of strange.

John Martinis:
And, you know, that it’s just a very deep theory and it’s kind of amazing. that nature works at this very, very deep level. And like I say, the other thing is, which is what our Nobel is about, it’s not just the physics of the small or fundamental particles. It’s actually a generic physics that everything can obey. It’s just really hard for ordinary objects to get into some parameter space where you can see it. So it’s actually a generic phenomenon that’s all, you know, that And potentially could be all around us.

Brian Keating:
And tunneling is that way too. And what really kind of surprised me about ordinary, you know, kind of electron tunneling is we have this kind of myth in both technology and in pure science that you look into the equations and then you invent the technology, right? So like, you know, Bardeen and the transistor, we couldn’t have invented it unless we understood quantum mechanics. When in reality, I think, I mean, you know this much better than me and I want to get your opinion. But, you know, kind of if you look at the first transistor, it looks like, you know, a chunk of rock, like the germanium, and a chewing gum and a coat hanger. And it’s all put together. I want to ask you, we’re going to talk a lot about quantum computers in a little bit.

John Martinis:
Oh, and by the way, our first experiment in Berkeley was carefully designed. Okay. And, you know, but if you look at what’s being made now, it’s, you know, it’s like, it’s like the Bell Labs transistor. But, you know, there’s some physics there. Okay. But that’s what you have to do when you’re first exploring something, is you do some experiment that’s kind of minimal and you can get it to work. And then once you understand the principles behind it, you can then engineer it and look deeper and deeper into it. And that’s what’s beautiful about physics, is there’s all these levels that you have to understand to get it to work.

Brian Keating:
What’s been the most, you know, kind of enabling technology on the STEP contributor to the work that you did? Was it the advances in superconductors? Was it the kind of fluxonium, the 3D cavities? What were kind of like stepping stones on the way to the revolution that you guys worked on and still do work on? What was some of the most important keystones?

John Martinis:
So what happened at the time is that we understood that this was a microwave experiment, and we went to the astronomy department and got their S-parameter meter and started understanding than reading microwave books. And in the end, what we did and the field did is combine the concepts of microwave engineering with the concepts of quantum mechanics. And it’s interesting because microwave engineering has wave phenomenon and resonance like quantum mechanics does. So they’re actually somewhat close. I also always think that you can understand about 80, 90% of our superconducting quantum devices with microwave engineering. And then you have to throw in quantum mechanics at the appropriate point to do that. It kind of reminds me, you have Maxwell’s equations, but in terms of understanding electrical circuit and the like, you use circuit diagrams. Okay.

John Martinis:
And what you’ve done is you’ve taken something very complete and almost abstract and then brought it down to a level where we can build, do complex engineering with it. And that’s kind of what we were able— what we started in that experiment. And of course, we explored that for many decades. And now, you know, we’re doing it and we’re still— I’m still exploring that in terms of materials and other concepts that we have here. Yeah.

Brian Keating:
And, you know, kind of makes me think about a statement I think you made once, you know, that people seem to hate decoherence until they need it. So Without decoherence, like friction, you know, if you’ve ever, you know, kissed a loved one, right? You need some friction, right? Life wouldn’t be fun without friction. But tell me, is decoherence necessary, you know, for these devices, or is it purely a nuisance that must be obliterated?

John Martinis:
I’m going to say decoherence is always here in the real world. And the problem is, if you take the Schrödinger equations, that’s just, you know, a pure, simple physics without decoherence. And of course, people know how to put in decoherence and do that. And it’s kind of like, you know, how do you understand thermodynamics without, you know, entropy? Okay. You know, you have the basic equations which are conservative, and then you introduce entropy, and then you could see the real world. And this is what happens with quantum mechanics. And also for quantum computing, It first, it’s, it’s a very practical, important thing because it limits your quantum computer. But also when you start doing things like measuring real circuits and let’s say doing error correction, in error correction, you’re removing the randomness or the entropy of that.

John Martinis:
And in some sense you need decoherence. And I would say decoherence, in my view, is kind of tied to how things get measured. Okay. And if you look at Exploring the Quantum by Ramon Den Haroche, it gives you a good description of that. That’s very integral to quantum mechanics. It sounds like the ugly side of it, but it’s actually quite an important part of it.

Brian Keating:
At some level, we have to always connect to the classical world, right? So there’s inevitability of dealing with classical effects. And so how do you guard against, you know, kind of these systematic biases? Like for us, let’s just take measuring a superconductor, right? So if you want to measure the superconductor, you could be very careful. You could do all the 4-point measurements you like. And you probably have been in a lab with my late great friend Paul Richards from UC Berkeley, and he was just the most careful person. And he wouldn’t let you do a measurement, you know, that wasn’t at least 4 points in this design. But at some level, you know, can you actually prove that these things have zero resistance in the junctions? Can you prove, you know, that the flux is purely being, you know, quantized in the way that the, you know, Leggett and other equations suggest that they are? Or do you always have to— Ah, we kind of have to— we know it’s not purely quantum mechanical because we have to these devices, or is it truly manifest that they behave as they should be purely quantum mechanically?

John Martinis:
It’s always a matter that there are certain limits where the flux will jump. Okay. And you could be, let’s say, near to the transition temperature. And then, and then you’ll see that flux is not quantized, or at least it jumps in its quantization. Physicists have been exploring this for a long time. In fact, In fact, the experiment I did in the ’80s was all about how, you know, when you put a current up to the critical current, at the critical current, it then looks like a normal metal. So it’s superconducting. And then when you hit the critical current, goes normal.

John Martinis:
Well, it happens a little bit before that, either due to thermal fluctuations or due to macroscopic quantum tunneling. And, you know, it’s an example of physicists understanding the limit. Now you can look at the limits of these various things and you can understand that it should be exponentially small. For example, for a superconductor, there are things, excitations called quasiparticles that limit the superconductivity, but there’s a gap and it’s e to the minus delta U over kT. And if you do the calculation, that’s tiny. But the problem is, is you have stray infrared light in a real experiment. and then generates the quasiparticle. So it’s not exponentially small.

John Martinis:
So I would say, you know, physicists are great at figuring out all these details and figuring out what’s wrong. And over the years, then, you know, this is why it took, you know, decades to figure all this out. Lots of experiments happened looking at all the details and not just taking the pure theory, but thinking about all possible ways that things can go wrong, and then you engineer around it. For the infrared case, you’d have to do very careful shielding, which we didn’t do at first, and then we realized we had to do that. And then there’s still a little bit of residuals, but we can deal with that.

Brian Keating:
So I wanna make a fairly heretical claim, and then I want you to demolish it and put me in my place. But my, my claim is that no one’s ever looked at an equation and out pops a technology from purely contemplating it, except perhaps quantum computing. We’ll get there in a second. But if If you look at the transistor, I just said, you know, they, they weren’t like looking at, you know, the, the Schrödinger equation saying, oh, we’re gonna get this technology if we put the chewing gum, the coat hanger, and the, you know, piece of germanium together. MRI came, you know, from Bloch’s equations being, being understood. Laser, maser came from population inversion, which was Townes’s kind of guess. Is the quantum computer perhaps the first technology in history that really came from the equation outward, or is it gonna be you know, sort of along the lines of, as I said, you know, the high-temperature superconductor. Really, we didn’t understand the theory until, you know, my late great professor Leon Cooper writes.

Brian Keating:
What do you make of this claim that I’m making that we don’t look into the equations and then the technology comes out? We experiment, guess, and then eventually technology comes and then we backfill in the explanation.

John Martinis:
Well, I haven’t studied this and it sounds like you’ve had, but I’ve been said, talked to, I’ve talked to theorists about this and they say it’s very rare that a theory kind of precedes an experimental observation. And the one example they give is the Josephson effect where Brian Josephson understood this. And basically you have to do the calculation to second order in order to understand what the superconductivity does. But the way this all came about, it was very murky at the time. And You know, if you look at it, John Bardeen gave Brian Josephson a very hard time with this, which is actually kind of amazing because superconductivity in BCS is a second-order calculation. Okay. People hadn’t put that all together at the time. So that was one of the few times, I’m sure it’s not the only one, but the few times where it preceded it and the theorist was given a hard time.

John Martinis:
But of course, the Nobel Prize. That meant that it was very strange. And I would say quantum computing, I hadn’t thought about that, but I’m— that’s right. This came from very theoretical concepts. And then, you know, people work through it experimentally once, you know, they understood it would be interesting to do that. Let me tell you what the problem with quantum computing is, is if you abstract it away to qubits, Okay. You abstract away to idealize qubits and the Schrödinger equation, and then it looks very simple and very nice. Okay.

John Martinis:
But the problem is real experimental systems are much more complicated. There’s all these dirt effects. And it’s kind of easy to think that, okay, you can just build that without having to go through and all the, you know, understand what’s going on. So So I like to say the best qubit out there is what I call the paper qubit, a theory qubit. And it’s only by doing the experiments do you know that everything is wrong. Everything’s wrong with it. And it usually takes decades to figure this out. Okay.

John Martinis:
And it’s not a magical thing. The other thing is a lot of the efforts are actually headed by theorists. If you, if you look at it, not all of them, but a lot of them are.

Brian Keating:
Yeah.

John Martinis:
And that’s because it’s very easy to abstract this away. I actually think, again, history will borne this out. I actually think that this is a little bit of a problem because in actually to build a thing requires you to, you know, really understand all the problems. Okay? So by abstracting all the problems away, you can be very optimistic and, you know, do things. But it’s only, you know, going into lab and realizing what all the problems are and then fixing them that you can actually build it because physical qubits are not perfect by any means. Some people claim that their technology is great. There are always problems. Okay.

John Martinis:
That’s just the way that nature wants to fight back. But I think in the end we can, we can fight back harder.

Brian Keating:
Yeah. Now I want to talk about quantum computing. And again, I tend to be a little more cheeky and provocative, so don’t be afraid to, you know, put me in my place. But in 1981, you know, Feynman didn’t say quantum computers are gonna replace, you know, your desktop, your MacBook, your laptop, your Chromebook, whatever. He said nature is quantum, so we should probably be using, you know, quantum systems to do computation. I always joke, and, and I’ve done work with a firm called Quantum Rings, which does a lot of software and simulations of quantum mechanical computers. But I, I kind of joke sometimes that, you know, quantum computers are the best system to model how quantum computers work, A, and then, you know, they’re good at—

John Martinis:
Quantum systems in general, yes. Quantum systems.

Brian Keating:
clear, you know, sort of like an answer, you know, to a question maybe. And again, I’m saying this with, with probably lack of humility, but what are quantum computers really gonna be good for? And you can’t use the words, you know, cryptography, and you can’t use Lagrangians or material science or quantum computers. So outside of that incredibly impressive domain of portfolio, I mean, it’s like if you said my computer can only be used for doing, you know, spreadsheets, word processing, and, and internet browsing, right? I mean, it could do a lot more, a general computer. So what can a quantum computer do? Besides those 3 things that are very important and very hyped up?

John Martinis:
Yeah, I’m really interested in, partly because I’m a physicist, okay, is quantum computers modeling, simulating other quantum systems. And, you know, there’s a huge amount that it can help with there because right now a lot of classical computers or supercomputers are used to do so. And, you know, you can only model something so big. Before you run out of memory because quantum computing is hard and run out of speed. Okay. And then you have to do approximation methods, which are fantastic, but, you know, they only work. And in fact, a lot of it is that, you know, certain approximations work for this problem and that problem, and you have to compare with physical systems to kind of choose that. It’s a little bit cheating, but, you know, okay, it’s very practical and that’s good.

John Martinis:
That’s what I’m, I’m really interested in. And, you know, just the example, I don’t know if this is a good example, but we all are interested in rare earths now, let’s say for electric motors and electrification of transportation system, et cetera, et cetera. But they’re rare and there’s a supply chain issue there. And I’m sure ecologically there could be issues with that. If you could use not so rare earths, let’s say by inventing a new chemical or process or maybe make it more ecological to mine, that’s a huge benefit to society. And you could say the same things with drug discovery and other things. I think this is actually a big application if you like academic industrial applications, but that’s more how regular computers got started. And then over time, I can imagine there could be other things, let’s say for optimization, it’s not so clear, There’s a killer application for that.

John Martinis:
Okay. A lot of people are looking at, a lot of people are claiming things. It’s not clear whether a clever classical optimization would be good. So it kind of can be a little bit like AI where people try various things for decades before coming on some, you know, the right way to do it. I also look at that as very important. So, you know, it’s a powerful computing engine. And it’s gonna take a while to figure it out. And, you know, the quantum computers we can build right now are too small.

John Martinis:
If we can make them bigger and then help with the theorists to inventing the clever algorithms, I feel, you know, very confident we can do something with this. But the big problem is we’re trying to compute— compete against these huge data centers, okay? Which are getting huger and huger every day. Eventually the exponential power of a quantum computer computer can overcome that. We just have to make it big enough and be clever enough for the algorithms.

Brian Keating:
Yeah, that’s right. And that’s what you and our mutual friend, Alan Ho, who introduced me to you and is co-founder of your company, Colab. We’ll get to that in a second. But now I want to take kind of the pushback on myself. You know, I’m kind of, you know, maybe bipolar this morning, but now I’m going to make the argument that these things are incredibly powerful and perhaps with great power. I just talked to the foremost AI safety researcher in the world, Roman Yampolsky, who coined the term AI safety. And he basically says super intelligence is either almost here or about to be here, and it’s uncontrollable. It is unaccountable.

Brian Keating:
It is unverifiable. We have no control over what we just created. So I want to make that argument for quantum computers, and then I want to take us back to, you know, like 1947. You know, the government didn’t let, you know, Oppenheimer set up Oppie’s Atomic Bomb Company, you know, just selling his own little portable nuclear device, right? He kept it classified. And, you know, should they be classifying, you know, is it okay that Google, IBM, and even Colab, you know, hopefully you’re going to be just as big as them, right? So tell me, make the argument. Why shouldn’t you be regulated right now before the genie escapes the bottle as it has for superintelligent AGI?

John Martinis:
First of all, we’re going to learn a lot from superintelligent AI. And that’s the immediate issue to deal with. And that’s here and it’s coming. And, you know, I agree people should be thinking about this. I think we’re going to learn from that. Okay. And we should take the lessons from that and then figure out what we’re going to do. The problem with quantum computing is it’s just not here yet.

John Martinis:
And yet there’s this big race. And to be honest, the race is the US versus China. You look at the papers from China, they know what they’re doing. It’s a serious race. developing kind of in the wild is actually an efficient way to get things done, just like with AI happened. Okay. Now, there was a secret program within the government for quantum computing, but that’s not where the biggest developments happened. I don’t think I have to explain that to readers.

Brian Keating:
Yeah.

John Martinis:
Okay. And it’s just that this competitive landscape I’ll just call it savage capitalism. Okay. Actually, it’s pretty efficient if you want to do that. In fact, I argue that the way that the projects in China is operating is maybe more savage than the capitalism in the US. I don’t know all the details, but it could be. These are very good questions and I’m concerned, but on the other hand, we’re trying to develop it in our own particular way. But we’re being very careful about who we do.

John Martinis:
We know that, for example, the US government is going to want us to build our quantum chip in the US. And that’s how we’re organizing the way that we do that. Some of the other more classical control that can be done worldwide with our good diplomatic partners. And we’re being a little bit careful about that. But I think it’s the Google and IBMs and where they’re really on the forefront, and I’m sure there’s a lot of discussion that goes on there.

Brian Keating:
Now, I can’t resist asking you this question. I mean, you’re the ideal person to ask. You’ve probably collapsed more wave functions than any human in history. What do you think is happening, John? Is it a collapse? Is it Copenhagen? Is it some non-unitary evolution? Is it a many-worlds branching? Tell me about your epistemology.

John Martinis:
What are you thinking when you do these I explicitly dislike the many-worlds interpretation because it sounds very Trumpian in the sense that you’re generating real estate. That’s, you know, that’s doing that.

Brian Keating:
So I’ve never thought of that. Now he’s gonna, now he’s gonna make a good point.

John Martinis:
In a humorous manner. But I’m very much thinking that, you know, the measurement and the dissipation and the decoherence is what’s giving you the state collapse. And again, if you look at Exploring the Quantum, they have a very nice, elegant way to talk about how these things called pointer states are exponentially sensitive to decoherence, and a small amount of decoherence can collapse you into these measurement states. And for me, that’s the clearest explanation around. I know some people don’t like that, and that’s fine, but that’s the way that I view it.

Brian Keating:
Jim Peebles once told me to shut up and measure when I asked him about some aspect of—

John Martinis:
If you shut up and measure, we wouldn’t have done that experiment, or people would have. So these are good questions. You need to do the experiments. And like in Exploring the Quantum, they did very nice experiments to flesh out what the theory was and to argue that this is what’s going on. So I think it’s important to study this and understand that. But for me, this is a question that has been answered via decoherence phenomenon. It’s just like not understanding entropy and thermodynamics. So to me, it’s the same kind of understanding.

Brian Keating:
Take us back to the, you know, the quantum supremacy and you had achieved this incredible result for the first time, but you soon after left Google. I’m curious, was that a blessing in disguise? I mean, it led you to co-found a company with, as I said, Alan Ho and others. It’s such a brilliant idea, this company. The point is the divorce from Google. Would you be willing to talk about that? We don’t have to, but—

John Martinis:
You know, after that experiment, Google decided to reorganize. And instead of being congratulated for leading this project, I was essentially demoted. Okay. And there were reasons for that that we don’t have to get into. And I tried that for about 9 months. And, you know, basically I went from the head of the hardware To, let’s say, 1 over N authority. It was very much a socialist thing, but I actually found I had negative authority after that. And if you want to understand negative authority, just think about when you had teenagers.

John Martinis:
Okay? That’s your negative authority. And frankly, I don’t think the people in Google thought that I was that technically competent. I was okay. But you know, you can tell when people feel that way and it was just time to leave. And what happened is that was definitely lemons. Okay. I still regret everything that happened, but it’s what happened. But what I would say is working with Alan and then Robert, we figured out, well, what is it we really wanted to do? Not, you know, next year or to meet the next milestone, but if we wanted to build a million-qubit quantum computer, what would we have to do? And we really focused on the qubit manufacturing and the wiring and scaling up and we came up with a series of ideas and we published a paper on that.

John Martinis:
We started a company and we’re feeling really good about this and we’re doing something that’s really different than everyone else. That’s exciting. And our view is that when we get this to work, it’s very foundational shift to the field, which is great. You know, that’s what you want to do is do something important. On the other hand, it’s risky because Because the general consensus out there is that you need to fabricate the qubits with this liftoff process because it’s much cleaner and the like. Whereas you do a complicated deposition and etch, you have problems. That’s the thing. And what we’ve figured out is that’s kind of right, but you have to fabricate it in the proper way and then you can get it to work.

John Martinis:
And we’ve kind of figured out what that proper way is and we’re working very hard to do the steps that you need to make it very clean. And, you know, in the end, semiconductors are— no one uses liftoff. I mean, this just doesn’t work. You use deposition and etch. But of course, I don’t know, there’s billions, trillions of dollars figuring out how to get that to work. We think we understand enough now to be able to do that on a modest startup economy.

Brian Keating:
I think your approach is so fascinating. It’s a sort of a 3D printing, but, you know, massive scale. My teenager This for me. That’s one of the few things that he does for me with my negative authority.

John Martinis:
Look, you know, young people want to do their own things. I get it, you know, and they want to break free from their family, which is what Google— what happened at Google. Normally the kids leave the house to break through and that they don’t kick the parents out of the house. But okay, you know, that was the easier thing for Google to do. And I understand that I had done things that the Google people— I’m too much like Elon Musk to work at Google. Okay, put it that way.

Brian Keating:
Well, I just note that it was exactly at that time that they went peak woke. And within a few months they had things like you ask it to create a picture of the founding fathers of America, and it was like Violet Davis, you know, Violet Davis, black and white hair and all sorts of interesting features.

John Martinis:
Well, you know, for example, all the co-writers of the Attention paper, which was the big breakthrough, they all left Google. I think They’re different reasons, but they’re similar reasons. And, you know, it’s not a surprise that certain people don’t fit into a corporate environment. They’re more entrepreneurs, and I’m very much an entrepreneur. And what I’ve been able to do is I’ve been able to kind of unleash my creativity in a private company. Now, we don’t have the money. I think I could be way more productive at Google, but if that’s not the way they want to run it, then, you know, it’s great. to be doing this in your own company.

John Martinis:
And in our company, we can set our culture and set what we do.

Brian Keating:
What’s the limiting factor just on a technical side? I mean, we have a dilution fridge, we don’t use it that often. We have deposition facilities here. What’s your limiting pacing item that is an obstacle, but you’re going to overcome it?

John Martinis:
If I gave you 10,000 dilution fridges, if I gave you unlimited time with 300-millimeter wafers, what do John, you need refrigerators to do a lot of testing, but you also need professional fabrication facility where you can do rapid turnaround. And then the third thing you need is a principled understanding of what’s going wrong. Right now it’s a little bit, you know, just people try things. However, I think we have a principled understanding now, so we have to work on the other two and, you know, obviously take more data and the like. It’s all of the above. In the end, I, I’m just gonna say in the end for us, it’s funding because with more funding we’d buy more dilution refrigerators and we could work with the companies and pay for having a bigger effort.

Brian Keating:
One final question is just related to the title of the podcast. The only way to know the limits of the possible, Arthur C. Clarke said, is to go beyond them into the impossible. John, what one piece of advice, you had 20 seconds with your 20-year-old self, What would you give the advice to him to go into the impossible with the courage that you’ve had over your career?

John Martinis:
Well, what happens is you’re a scientist, you’re always working on projects. There are projects that are kind of incremental and you know what to do and you’re going to advance your field and whatever, but always be on the lookout for the impossible, something new, something other people don’t think will work that if It does work. It’s very foundational and changed the field. Now, you’ll have to curate those ideas really well because most of your ideas aren’t going to work out. And I have ideas all the time and I curate them, and then you choose the best ones and try it. Our company, Collab, is what everyone thinks is not the right way to go. I’ve thought about it carefully. We understand why it could work and it’s looking good.

John Martinis:
But you have to think very carefully about it. But yeah, always be on the lookout for the impossible, right?

Brian Keating:
I love it. I’m going to make that the motto of the show. John Martinez, winner of the 2025 Nobel Prize in Physics, thank you so much for being an inspiration. You’re just a physicist’s physicist. John, thank you so much. Have a great weekend. We’ll talk again soon.