Podcast with Steve Brierley — CEO, Riverlane

Classiq Technologies
12 min readJul 6, 2022


My guest today is Steve Brierley, CEO and Founder of Riverlane. Steve and I talk about quantum error correction — why it’s difficult, how soon we can expect it, and much more.

Listen to additional podcasts here


Yuval: Hello, Steve, and thanks for joining me today.

Steve: Hi, nice to meet you.

Yuval: So who are you and what do you do?

Steve: Hi. Yeah, I’m Steve Brierley. I am the CEO and founder of Riverlane, which is a company building an operating system for quantum computing.

Yuval: So what does it mean to build an operating system for quantum new computing?

Steve: Great question. I think when you think of operating systems, most people think of a laptop or a PC and really what it’s all about is managing the complexity for the user. So on your computer you drag and drop a file, that’s really simple, but a whole bunch of complex things happen underneath. And the same is true of quantum computing. Building a useful quantum computer is very complex. There’s a lot that needs to happen. And the most complex thing in quantum computing is quantum error correction. And so when I say we are building the operating system, what I mean is we are solving error correction on quantum computers. That’s super complex, and none of the users are ever going to see that. We will manage that complexity for the user.

Yuval: So your customers are those who are building computers?

Steve: So we partner with companies that build computers, because, of course, we need qubits as well to make this all work. We help them take the basic components, the individual qubits, and turn them into computers via error correction.

Yuval: So if you read the popular press, they say we live in NISQ era. And so one could believe that there is no error correction going on these days. So how does it work? Is there error correction? Is there just partial error correction? Where is this heading?

Steve: So the big challenge is to implement error correction. Right now, most systems don’t have much error correction in them, if any. And so they are limited in how much computation you can perform before the noise starts to take over from the system. So if you think about some of the best quantum computers in the world, say at Google or the one that is being developed in China at Hefei, these can perform around 100, perhaps maybe even up to 400 operations before errors or before failure. That’s amazing. That ability to control individual qubits is astonishing and it is a really huge breakthrough, but the number of operations we need in order to perform useful computation is way bigger. It’s on the order of billions of operations. So this gap between computers we have today, the hundreds of operations we can perform, and the billions of operations that we need, that’s bridged by error correction. And that’s what we are working on.

Yuval: So is your code running someplace in a customer computer or is it primarily under development at the moment?

Steve: We’re doing a lot of development and then we are really privileged to be working with lots of different quantum companies and research labs to test out parts of our system. We’re not yet at the stage where we have a full scale system up and running, but we can test bits of it, so we can test components. And I think this is really important because ultimately, a large scale quantum computer is this hugely complex system, it has literally billions of components and operations. And at the end of it, you get some results, you get some numbers out. And if it didn’t work, then it’s really hard to figure out what went wrong.

And this is particularly acute in quantum computing. This is a problem of any large scale system, but in quantum computing you can’t stop the computation and say, “Oh, hang on a minute. What’s happened here?” So you need ways to build a system up component by component in scale. And so that’s why we call ourselves a quantum engineering company, because we’re applying these systems engineering techniques in quantum computing, so we’re testing components right now because that’s the way to build a large scale system.

Yuval: And is error correction done by taking some qubits that could have been noisy and putting them to use for the error correction or is it done in a different way?

Steve: Yeah, that’s right. Thanks for the question. So quantum error correction has lots of the same properties as classical error correction, traditional error correction, so we spread out the information across many quantum bits in order to protect against noise in any one of them. So that’s just like classical error correction. So the simplest error correcting code is, instead of sending a zero or a one, you send 0, 0, 0 or 1, 1, 1, and then if any one bit flips, for example, you receive 0, 0, 1, you can guess, well, I suppose you meant to send 0, 0, 0. So that’s a simple code. It’s called the repetition code. Quantum codes borrow a lot from classical coding theory, but there’s some really interesting twists. And I think the most important one is that quantum error correction happens during the computation.

So just to spell this out a bit, cause I think it’s an important point, between every operation on the quantum computer, we’re going to run this error correction cycle. So we’re going to read out some information from the qubits that tells us something about where the errors occurred. We’re going to solve a decoding problem. This is a bit like a Sudoku puzzle or something like that. So we’re going to solve this decoding puzzle and then react to that. So we’re going to change what happens in the future of computation, computational steps as a result of this decoding problem. So that error correction cycle, that readout, decode, correct, that’s the clock rate of a quantum computer.

So that’s the internal clock of the quantum computer, because you have to do that between every quantum operation, between every time you add two bits together or every basic operation that’s happening in the quantum computer goes around this cycle. So a result of that is that this decoding problem has to happen really quickly and you’re going to have to solve a huge number of decoding problem. Ultimately, 10 to the 12, 10 to the 15 decoding problems for every computation that you run. So that’s a huge amount of data processing and that’s the big technical engineering challenge that we are addressing at Riverlane.

Yuval: Is error correction the same regardless of the modality of the qubits? Superconducting and trapped ions and so on, would they all use the same error correction code?

Steve: I think the answer is yes and no. So we absolutely need to adapt how error correction is implemented based on the underlying qubit type. And I would go even further than just superconducting. For example, there are different ways to build superconducting systems or different types of superconducting qubit and even different architect, so qubits connected together in different geometries. So you have to take that into account in your error crafting code, so it’s very specialized to the underlying architecture. However, a lot of the machinery and a lot of the problems are the same across the different qubit types. So, for example, the decoding problem, most decoders are essentially solving a particular graph type problem, a matching problem. So that’s the core of the commonality across the different qubit types.

Yuval: Do the characteristics of the qubit in terms of the errors, do they change during the day or every second, or is it something that you measure once a month and then it’s pretty much stays the same?

Steve: Yeah, also great question. Yes, they do change over time. So you get drift in the system and we need to take that into account as well. I think over time this does stabilize and there are multiple processes running at various different layers in the stack, but the temperature changes or the various physical effects are causing drifts in the system. And again, that will depend on the particular physical architecture that is being used.

Yuval: So before error correction is implemented, do you guys provide data that could be used to, for instance, measure the fidelity of qubits and perhaps something that an application could say, “Oh, qubit two is much noisier than qubit 12 and therefore I should assign the busier operations to qubit 12 instead of two,” or something of that nature?

Steve: We don’t. Being able to measure and characterize systems is certainly something that is really important. You want to do some demo of a toy problem, then trying to get the most out of the qubits by trying to take advantage of what’s happening is a great thing to do. We are really focused on getting to large scale quantum computing as soon as possible. And so our engineering effort is really on architecting and building the solutions that we need in order to implement large scale error correction. This decoding problem is never been solved at speed before. It’s an entirely classical problem. It’s a classical algorithm that’s running, but nobody’s been able to solve this fast enough. So the industry’s, for a long time, understood that this would be a problem. And it’s really fantastic to be in a position to be able to solve that.

Yuval: And forgive my ignorance, but what is it that you sell? Do you sell hardware designs? Do you sell software algorithm? Do you sell consulting services? What is it that you go to your customers with?

Steve: So we partner with hardware companies and allow them to be able to solve error correction. So it’s essentially the system that sits on top of the qubits or that’s very close to the qubits, but it connects with the control system and runs error correction on top.

Yuval: So you have a ringside seat to the development of many of the quantum computers, how close are we to an error corrected computer in your opinion?

Steve: We have very privileged position because we work with lots of different qubit types, different groups ranging from big academic labs to commercial companies. I think the timeline is really one of engineering rather than some abstract thing that we are just waiting to happen. So the key things that need to happen are building the classical control electronics and decoding capability that’s large enough and fast enough to keep up with the qubits, solving some of the material sciences problems in… Well, certain different qubit types have different kinds of physics and engineering type problems. So from my perspective it’s less about time and more about a list of things that need to get done. And I think what’s happening right now is that’s been accelerating very rapidly due to the increased investment. So I’m hugely optimistic that we can get the early error corrector systems in the next four to five years.

Yuval: Are you worried at all that there’s going to be a quantum winter, that the expectations are going to so much exceed reality, then lots of companies will become disillusions with quantum computers and just put the project off for several years?

Steve: I think there’s some activity in the space that maybe will die away, but that’s not necessarily a bad thing. Building a quantum computer is a really hard problem. It’s not a toy system that we’re building, it’s something like NASA’s control center, it’s that kind of scale. And so those companies that are addressing that central challenge, there is the appetite there to fund that. And the reason is that the value of what comes out of the end is so huge. Quantum computers are not just a bit faster than your CPU, they solve problems that will never otherwise be solved. So there’s an exponential, computational advantage.

And that sounds a bit mathy, but what I’m really saying is that there are important problems such as the simulation of quantum physics for which classical computers will never be able to solve. And quantum computers will, for the first time, enable us to simulate molecular systems, proteins, catalysis materials, and it’s going to completely transform a whole bunch of industries. All of those industries that got left behind by the first digital revolution, it’s the quantum revolution that takes those industries from this discovery phase that they’re all in to a design phase, where they actually can design products using simulation just like happens in the aerospace industry. So I think the reason that quantum is going to get funded is because of this huge potential. Not so worried about the quantum winter. I think certainly there’s a lot of excitement right now. That’s actually very helpful. Maybe it’s a bit hypey, but that’s attracting good people to come and work on hard problems, and that will result in something very real.

Yuval: So if you were a master of the quantum universe and you could control what other companies are doing, is there anything you’d like them to do other than just work faster?

Steve: I think focusing on the core challenges. We focus on getting to error corrected systems because it’s quantum error correction that makes quantum computers useful, so all of the things that need doing in order to achieve that, that’s the space to work on. That’s really important. So the more we can understand about, for example, noise in quantum systems or how to control qubits, how to scale them up, things like networked quantum computing as a route to scaling. I don’t think I need to be a master actually, because this is already happening. I think if you look at every qubit type right now, every qubit type has some big science challenge. There’s something really big that needs solving in order for that to become the transistor of quantum computing. It doesn’t matter if you’re superconducting, ion trap physics says there’s something within any each of those that is a big, hairy challenge. And I think the great thing is that for all of these challenges there are companies that have got brilliant ideas for how to solve these, and that’s happening right across the different physics approaches.

Yuval: So as we get closer to the end of our conversation, I think that you’ve been working on this for 15 years. So I gather that you like it, but why is it taking so long? Is it just a terribly difficult problem? Is it that funding was not available early enough? Not many problems take 15 years to solve.

Steve: I’ve been working on quantum computing for 15 years. I started working on quantum algorithms. So this is what would you do with a quantum computer once one was built. And is this very theoretical. I worked in academia at Bristol University and then here in Cambridge, and actually I founded Riverlane because of a moment at a conference. I was giving a talk about a recent algorithm that had developed and at this conference there this straw poll of who thought there would be a quantum computer in the next 5, 10, 15 years, and so on, and about a third of the audience, so this was all the great and the good of quantum algorithms and quite a few physicists, a third of the audience voted that there would never be a useful quantum computer.

I was just really shocked. And, I guess, partly the pessimism, it’s academic space, it’s people are pretty pessimistic, but maybe I was working on developing algorithms for a computer that would never be built. And so I went back and started to talk to the experimentalists, talked to people building qubits, this was about five years ago. And they had, at that point, 10, 15 years of data of how good are you building a qubit and photonics?

So Jeremy O’Brien at the time was Bristol, so he was asking his group. So now PsiQuantum was talking to the ion trap folks over at Oxford and the superconducting people. And what I saw was this trend that we were getting closer to solving the underlying physics of controlling individual quantum computers. And so what was needed was to make this useful. What was really missing was the engineering approach to scaling up quantum computers. You’ve got lots of great physics labs around the world getting a handle and control on individual qubits, and now is the time to bring people from other fields to solve the engineering challenge. So that was five years ago at Riverlane and we are now motoring ahead on solving that.

Yuval: So, Steve, fingers crossed for your rapid success. And how can people get in touch with you to learn more about your work?

Steve: So email me, steve.brierley@riverlane.com, or we have a website that’s a great jumping off point, for example, we have a newsletter, so that’s riverlane.com. Or reach me on LinkedIn. Happy to answer any questions.

Yuval: Excellent. Well, thank you so much for joining me today.

Steve: Great. Thank you. I enjoyed the chat.

About “The Qubit Guy’s Podcast”

Hosted by The Qubit Guy (Yuval Boger, our Chief Marketing Officer), the podcast hosts thought leaders in quantum computing to discuss business and technical questions that impact the quantum computing ecosystem. Our guests provide interesting insights about quantum computer software and algorithm, quantum computer hardware, key applications for quantum computing, market studies of the quantum industry and more.

If you would like to suggest a guest for the podcast, please contact us.

Originally published at https://www.classiq.io.



Classiq Technologies

Classiq Technologies: Quantum Computing Software. Join our platform: https://platform.classiq.io/, and community: https://classiq-community.slack.com/