A Nobel laureate says he will build the world’s most powerful quantum computer

John Martinis is a hardware guy. He prefers doing physics in a lab to the idealized world of textbooks. But you couldn’t write books on the history of quantum computing without him: he was central to two of the most important moments in the field. And he’s hard at work chasing the next one.

It began in the 1980s, when Martinis and his colleagues conducted a series of experiments to probe the limits of what was known about quantum effects — work for which he won the Nobel Prize last year. When he was a graduate student at the University of California, Berkeley, we knew that subatomic particles were subject to quantum effects, but the question was whether the world of quantum mechanics could be extended to larger scales.

Martinis and his colleagues built and studied circuits made of a mixture of superconductors and insulators, where it turned out that the many charged particles in the circuit behaved as if it were a single quantum particle. This was macroscopic quantumness and laid the groundwork for building some of today’s most powerful quantum computers, including those currently being pushed by IBM and Google. In fact, Martinis’ work set in motion a trend of tech giants using quantum bits, or qubits, made from superconducting circuits—the most widely used qubits in the world today.

Shaking up the field a second time, Martinis led a team of Google researchers who built a quantum computer that achieved “quantum supremacy” for the first time. For nearly five years, it was the only computer in the world, quantum or otherwise, that could verify the output of a random quantum circuit. It was later surpassed by classic computers.

Now on the cusp of the 1970s, Martinis thinks he can achieve another historic victory with superconducting qubits. In 2024, he co-founded QoLab, a quantum computing company that he says will take a radically new approach to trying to create what everyone in the industry has been chasing: truly practical quantum computers.

Karmela Padavic-Callaghan: You made waves early in your career and did some really seminal work. When did you start to understand that your experiment could lead to a new technology?

John Martinis: There was the question of whether a macroscopic variable could escape quantum mechanics, and since I was young and just learning quantum mechanics, it seemed like something we needed to test. Maybe if you were older you just assumed quantum mechanics would work. But as a young student it sounded like a fantastic experiment to do a basic test of quantum mechanics.

The first thing we did was set up a very rough and tumble experiment using the technology of the day. When we took the data, the experiment failed. But we were able to fail quickly so it didn’t matter. In the end it was an experiment where you had to understand microwave engineering. You had to understand the noise, there’s a lot of technical stuff we had to do, but [success] it happened pretty quickly after that.

For the first 10 years after that, we were doing this experiment and building quantum devices. Then the theory of quantum computers advanced a lot, I would say Shor’s algorithm in particular [which factors large numbers for breaking cryptography]then error correction [algorithms] soon after. This gave the field a solid foundation. People could now imagine building something. Thanks to this, finances were freed up.

How has funding changed research and ultimately technology?

Things have really changed since the 1980s. Back then, people didn’t even try to see if a single quantum system could be manipulated and measured correctly. It’s interesting how things have moved in the last 40 years. Quantum computers have grown to a huge area! The proudest thing of all is how so many physicists are now employed to understand the quantum mechanics of these superconducting systems and build quantum computers.

You had a hand in the early days of quantum computing. How does this help you understand where the industry is going now?

Having been part of the field all along, I understand the basics of physics. I built the first microwave electronics for [quantum devices] in our group at UC Santa Barbara and then at Google, I built my own cryostats [devices that keep superconducting quantum computers chilled to the extremely cold temperatures they need to operate]. I was involved in the production of every element. I think a lot of people, if they haven’t been through all that, they’re just going to be optimistic that we’re going to move on. I know where all the problems are. If you want to build a very complicated computing system, it’s all systems engineering, and I think I have the advantage of understanding the underlying physics of everything quite well.

How do you think quantum computing hardware needs to change to make quantum computers useful and practical? What changes are you betting on as the beginning of the next breakthrough?

After leaving Google, I thought about the quantum computer as a whole system and rethought all the basics of what we actually need to build and improve. QoLab is based on that, with fairly dramatic changes to how we create qubits [in terms of manufacturing techniques] and how did you put it all together, especially the wiring.

We realized that you have to think about building quantum computers in a completely different way to make the technology reliable and to keep costs down. It’s hard and it’s hard for people to understand. We’ve had a surprising amount of pushback and skepticism, but from my experience in physics over many decades, it means we have a good idea.

We sometimes hear that a very large number of qubits, in the millions, will be needed to create a flawless quantum computer that is actually useful. how do you get there

In terms of where we want to cause the most disruption, it’s in manufacturing and especially quantum chip manufacturing, which is also the most difficult part. If you look at what they’re all doing—Google, IBM, Amazon, and many other companies—they’re using manufacturing techniques that date back to, I don’t know, the 1950s or 1960s. I don’t know [any other industry that] builds real circuits these days with these methods. So our view is that if you want to make a million qubits and make them reliable, you want to do something else.

We are very excited about how we can fundamentally change the way these devices are built. And we have an architecture for chips that can help get rid of all the wires. If you look at the picture [superconducting] quantum computers, it’s just a jungle of wires and microwave components. I want to put all that stuff on a chip and be able to scale it up. In superconducting qubits, the wiring problem is a big problem and we are working to solve it.

Do you think there will be a clear winner in the race for a practical quantum computer in, say, five years?

There are many different ways people are trying to build a quantum computer, and since the constraints of systems engineering are very difficult, I think it’s good to approach this problem in many different ways. I think it’s good that a lot of different ideas get funded, because then the chances of people breaking through are better. But when I think about those constraints, and there are a lot of them, I would generally say that a lot of the projects are a little, shall I say, naive in terms of what’s really needed to meet them, like managing costs or making devices at scale. On the other hand, I’m sure many research teams have ideas to overcome some of their design problems that they don’t talk about publicly.

And QoLab’s business plan is, I think, a little different, maybe even unique, in that we encourage collaboration because we feel we need all the expertise. We work with hardware companies that know how to scale and know how to do sophisticated manufacturing.

If someone gave you a very large and flawless quantum computer tomorrow, what would be the first thing you would try?

I am really interested in using a quantum computer to solve problems in quantum chemistry and quantum materials. Recently, there have been several articles about its use [extract more useful information from] nuclear magnetic resonance (NMR) experiments in chemistry and I really like that as a first application. This quantum problem is difficult to solve on a classical supercomputer because of the fundamental difficulties of quantum mechanics. But of course this is fundamentally solved with a quantum computer – you just map a quantum problem to a quantum computer. I can get excited about it, partly because I like defined ideas of how to build [a device] and humans developed defined algorithms [applications like enhancing NMR].

A lot of people might think about doing something with, say, optimization problems and quantum artificial intelligence. For me it’s more of a “try it and see if it works”. The theory of materials and chemical applications is much more defined. We know how big [quantum computer] must be. This machine is something I think we can build, both in terms of size and speed of execution.

Some of the possible uses of quantum computers were mathematically determined more than 30 years ago. Why haven’t they become a reality yet?

You can abstract away the behavior of a qubit and imagine how to build a quantum computer, and that’s great because then you can get computer scientists and mathematicians and theorists to think about it. But the real problem is that real qubits have sources of noise [such as heat from external wires, or impurities in the qubit’s own material]and problems that are physical things. A lot of big quantum computing efforts are run by theorists, which is fine, but a real system is just a lot more complicated, as is what you have to do to build hardware that can work properly.

IN [my graduate advisor] John Clark’s group, I trained myself to understand the noise. That kind of background was really beneficial for me and the people I worked with because we were thinking about qubits in just this physical way and trying to get rid of the physical noise mechanisms that make chips unreliable. This is what happened with the quantum supremacy experiment; [some of the noise comes from the fact that] you have these “two-level states” that are in your device and you control them to avoid them. You can get it to work, but it’s a real pain in the neck and just makes it harder to scale. I hope we do 1770142614 get rid of this effect or reduce it. To understand this, you need to go into the details of the qubit design.

The problem is that you have to have both the hardware and the ideas for the applications, and I think we need to improve the hardware a lot across the industry. So that’s what I’m focusing on.

topics: