Bert de Jong, Director of the Quantum Systems Accelerator (QSA) at Berkeley Lab, is interviewed by Yuval Boger. Bert describes how the center develops superconducting, trapped-ion, and neutral-atom technologies in parallel, and the importance of certification in verifying quantum computations. They discuss the roadmap for integrating quantum into HPC at NERSC, the cost and scaling challenges of large systems, and the Department of Energy’s focus on scientific—not cryptographic—applications. Bert also highlights creative approaches to error correction, opportunities for industry collaboration, the need for workforce training starting in high schools and community colleges, and why he believes quantum will deliver practical scientific impact within five years.
Transcript
Yuval Boger: Hello, Bert. And thank you for joining me today.
Bert de Jong: Thank you for having me. I’ll enjoy chatting with you.
Yuval: So who are you and what do you do?
Bert: I’m Bert de Jong. I’m a senior scientist at Berkeley Lab where I do a lot of different things. So by training, I actually am a computational chemist and I like to solve science problems. And to me, things like high performance computers, computing, AI and quantum are tools that need to be developed so I can actually solve real science problemss. So within the quantum world, I currently lead one of the five National Quantum Initiative Centers, the Quantum System Accelerator, that’s co-led between Berkeley Lab and Sandia Lab. And I have a couple of other programs that are also focused on quantum, more on developing software stacks and actually trying to figure out how we can use quantum. computers to do some real innovative science.
Yuval: I think a couple of weeks ago, LBNL announced that they’re acquiring the Doudna supercomputer.
Bert: Yes.
Yuval: How soon before there’s a quantum computer attached to that?
Bert: The roadmap that NERSC has, the Doudna machine is the next machine. But the one after that, they are penciling in a quantum, component. So NERSC actually has a team working on quantum right now, engaging with vendors and engaging with the research community to really understand when quantum will be important, when quantum could actually be in addition to a user facility like that, and really understand the when, how, what, of making quantum computing a reality for a HPC center that is funded by the Department of Energy.
Yuval: Let’s talk a little bit about the QSA. What does the QSA do?
Bert: The quantum system accelerator is, as I said, one of the five centers. It also has taken a slightly unique approach. And that is, we want to get to a point where we can actually deliver or develop the technologies that can make quantum computing successful. And we don’t want to make a choice up front. And we haven’t. So the team that we have, it’s about 400 to 450 researchers, so 15 institutions. They do research that is focused on developing neutral atoms, trapped ions, and superconducting qubits, all three technologies. Not just to build these technologies and build them at scale, working with industry, but also to understand what the potential quantum advantage could be for these systems. So we’re doing a lot of work also on the applications, on the benchmarking side, to really, while we’re developing technologies, these technologies to actually explore these technologies, see what science can be done. Where are the biggest bottlenecks? Where are the biggest challenges to actually make this a scientific discovery tool for the Department of Energy?
Yuval: I’m glad you mentioned neutral atoms. Obviously, I work for a neutral and company when I’m not doing the podcast. But why not photonics or silicon qubits? I mean, you mentioned, I think, superconducting neutral atoms and trapped ions. Why not the others?
Bert: I would say when we started five years ago, so I’ve taken this job over a little bit over a year ago. So when the center started five years ago, the decision was based on where are kind of the technologies at that point in time, where can they go? And what are the best candidates to make big inroads going forward? And superconducting, of course, had been on the road, had already a long-term roadmap laid out. Trapped Ions had been around, and there were ideas on how to scale that. And neutral atoms, and you’re well aware of it, in 2021, was not very far yet, but had the potential. So I would say all the other technologies, they have potential, but you have to make a choice and you have to put a team together that can be actually making significant inroads in these three technology areas. What’s nice of doing neutral atoms and trapped ions, for example, is if you look at the roadmaps or the technology roadmaps, one of the big challenges going forward is, for example, the need for integrated photonics. Well, the national labs have those small foundries where they can actually explore these kind of capabilities. Sandia built trapped ion systems. Berkeley Lab has a long history in superconducting qubits. So it’s also partly the team that would make sense to actually make inroads, see if you can move the needle on those technologies and see where they’re going. But in reality, spin qubits or NV centers, there are small amounts of research, also happening here at Berkeley in those areas. Photonics, yes. I would say to me, that still feels like a technology that is a lot further out, even though I know PsiQuantum does a lot of work in this. But we’re still waiting to see kind of a first real small quantum computer coming out of that technology.
Yuval: Do you worry that some of these technologies may end up costing a billion dollars per computer if it’s large enough? Or does it not matter? I mean, it’s enough for a national lab to have one or two of these, and that’ll be great.
Bert: No, I totally worry about what you hear with some of these technologies, right? Where it’s like, oh, we need a million qubits. A million qubits is not that big of a deal unless you have to have all the ancillary pieces to make these one million qubits work. for atoms and ions, it’s lasers. It’s a lot of the electronics that go around it. I think I talked to one of the researchers within the Sandia campus about the photonics, and the statement was you would pretty much have to buy every component that is used by the telecom community to even build one. But that is not going to be a realistic thing. that then becomes a supply challenge. So do I believe we need a million? Eventually, you will have to get there if you want to build a fully fault-tolerant quantum computer because you need to get to 10 to minus 16. But what’s interesting about the Department of Energy is that a lot of the science problems we need to tackle don’t need that accuracy. We’re not after breaking encryption. Maybe the national security side is interested in breaking encryption, but the Office of Science wants to solve real science problems, make scientific discoveries. 10 to the minus 7, 10 to the minus 8, that’s all we would need out of a quantum computer. And that changes the scale of the systems that you might need. So I don’t think what we need for the Department of Energy is going to be a billion-dollar computer, quantum computer at least.
Yuval: You mentioned some of the government institutions and government funding is sort of in the headlines these days. Do you see an impact on quantum funding or is quantum funding still doing well?
Bert: What’s interesting is that quantum actually started under the same administration five, six years ago, and has been continued across administrations, across different political spectrums. And the messaging that has come out of Department of Energy, but also out of the White House has been quantum and AI are going to be important directions, technological directions that need to be pursued. So this has also shown up in early versions of appropriations bills or even the White House kind of ask where quantum is really put still. as an important scope of research that needs to be funded going forward. So from that perspective, I think we’re in a good space right now.
Yuval: I think I saw on the QSA website that you’re looking to deliver certified advantage, certified quantum advantage. Tell me about the certified part. How do you consider, what do you consider certified?
Bert: I’ll give you the scope that there was a scope that was defined pre-me. So certified actually means, so right now I can run a quantum simulation. How do I know it was correct? And that I got the right answer, especially if I go and run a quantum simulation, that is of a scale that we cannot model classically anymore. So the idea behind certified means literally, can we say, this is a reliable quantum computation. This was the program that I wanted to run as exactly the program that I ran. And so the interesting part is that this has been a collaboration between some fundamental computer scientists at Berkeley, Umesh Vazirani, for example. He developed the protocol to actually certify quantumness of the system. Then he worked with researchers that Duke researchers actually demonstrated on an Ising system, on a trapped ion system. So be able to certify that this actually is what I’m running, and the answer that I’m getting is the right answer, I think is going to be a critical piece. Once we go to scales that we cannot simulate classically anymore, and that’s an interesting conundrum we’re in right now, because the systems are still small enough that we can take a supercomputer or even a smaller computer and simulate and see if we get the right answer. But if you really want to get to something that is a quantum advantage or a scientific advantage over classical resources, we need to run simulations we cannot verify anymore. So, verification, certification is going to be critical to make sure that we get the right answer with the right reason.
Yuval: Help me understand the interaction between industry and the QSA or LBNL as it relates to quantum, you know. So if a vendor were to deliver a quantum computer with 10 to the minus 8 error rate and enough qubits, does that mean that your work just stops or does it just begin on the algorithm side? How do you know what to work on and what to leave to industry?
Bert: We had long conversations with different industry partners too. So I like to go back to the whole eye in a neutral atom situation because that’s a clear example of how these things can develop. Again, our team, our academic team working on that is Harvard, together with MIT and some other partners. They have been building neutral atom systems, advanced in technology, to a point where commercial entities are actually taking advantage of what has been developed. So where I would like to be as a center, especially for another five years going forward, is to be technology forward. So companies tend to be two or three years on their roadmap, but they have things they want to pursue beyond that. That’s where these centers can do more high risk development, high risk capability development, technology development, that hopefully de-risk industry in the future. That should be, I think, a role of these kinds of centers. This is really helping them, helping industry forward, but it’s not our job to compete with industry, right? So we need to find where the holes are. Integrated photonics is another great story. We have many companies right now in the neutral atom and trapped ion community that are ready to go and work with us simply because it’s on their roadmap, but they haven’t been able to do a lot of research or technology development. in it. And now there is a platform or capabilities that they could use and try and explore that in a more efficient way.
Yuval: One of the big areas of research in quantum computing, I believe, is quantum error correction. Yes. What do you guys do in the quantum error correction area?
Bert: So the original proposal for QSA did have not that much. in quantum error correction. There were constraints on what we were able to do at that point in time. We focused a little bit on developing flag qubits approaches, but there has been some interesting work in error correction, where it is can we actually utilize error correction in different ways? And the story that I like to tell is actually some work that comes from Sandia by Andrew Landau, where they took in a very different approach to error correction. So what they were able to use is take a surface code and encode fermions into that surface code such that you did not have to add more qubits. Normally what you do right now is we take a bunch of qubits, translate it into fermions, and then do error correction. Their approach, skip that transformation directly encode fermion and qubits into into the logical qubits. It’s a big resource saver when you do those kind of things. That is the kind of creative stuff that really needs to happen a lot more. Now that we are getting more and more into the era of error correction. I have other programs that I lead. One of them is actually focused on a lot more on error correction right now. We actually have a compiler suite called, Biscuit, which is one of the leading compiler suites here in a community, I would say, we’re expanding that now to be able to do logical qubits, for example. So there’s a lot of error correction work that is being done there. I would say where we want to go when it comes to QSA, we do want to have a much larger footprint in error correction going forward, simply because we now, I would say, in 2021, we had small systems. Now we’re getting to systems, or we’re getting to the threshold of where we have systems that are large enough that actually quantum error correction can be explored, tested, developed at a scale that makes sense, where we can actually have true insights of, does this work, where does it not work, how can we make this better or more efficient?
Yuval: I think you mentioned you’re a computational chemist by training. Do you see chemistry as the key application or the one that quantum computer will show the soonest commercial utility, or are you banking on something else?
Bert: So that’s an interesting. So again, if you look at what the Department of Energy does, right, so the big scientific problems that need big HPC right now are in high-energy physics, nuclear physics, chemistry, and materials. What’s interesting if you carefully look at them is that they have very similar what they call Hamiltonians. So very similar interactions that they need to describe in their system. And I would say that class of problems actually has a very good chance of delivering scientific advantage in the next five years. if we can get to systems that are in the order of hundreds of logical qubits at 10 to minus 7. So yes. And I would say chemistry, there is a lot of problems in chemistry that are not what we would say, strongly correlated or have a lot of entanglement, which might not be always suitable for quantum computers. What we have seen is actually a lot of the problems that are interesting are dynamics, problems. And that is something that a quantum computer naturally is going to be good at. As long as we can prepare the state up front, put the system in a state that we want to have as a starting point and let it evolve, I think that’s where a quantum computer will shine. And again, given that we can tolerate a little bit of noise, we can still get pretty reliable results out of these systems. And we have plans for that in high energy physics to actually tackle problems that we do not want to even think about on a classical computer. The same in chemistry and materials. So, yeah, I would say that class, especially since we don’t need full error, fully error-corrected systems, and there is a lot of interesting and challenging science problems that the Department of Energy would have a high interest in seeing solved. I think those have a very big chance to be early adopters and early deliverables for scientific advantage and whatever, quantum advantage, however people want to call it.
Yuval: I believe you spoke about the need for deeper industry partnerships to advance some areas in quantum computing, whether on the material side or the algorithm side or quantum error correction. how do you envision the ideal industry partnership? Is that IP sharing? Is that a test bed that has wide access? What would your vision be?
Bert: I think so a test bed is not at the level up. I would say the NQIs, the research centers, that’s what they are. They are research centers to really move the technology. forward, be technology forward, and explore these technologies, build maybe prototypes of systems. Testbeds are the next level up. And the Department of Energy has actually a couple of test beds. They have a trapped iron test bed and a superconducting qubit test bed. Those, to me, are the ones where you can go and build these systems at scale, but not fully industry ready. It’s not like a commercial partner that actually wants to use the quantum computer would be able to as easily use a test pad system. But it allows us to figure out software stacks, really understand where the challenge is like to make this a technology that is ready for a consumer base effectively. So I see the NQIs as much more heavily on the research, which means I would like to see much closer direct research collaborations between industry and the NQI centers. And as I said, there is definitely, when it comes to a lot of these quality companies, they are either startups or early in their life cycle, they have deliverables to their VCs. They have a two, three year time window. And it’s harder to think beyond or work beyond those three to five, two to three year of time windows. That’s where that partnership can be very, very fruitful. It would accelerate what needs to be done for industry, and it serves our purpose of helping industry build the technology and the capabilities that we need to do eventually to do our scientific discovery with.
Yuval: There’s a common belief that workforce training is critical to the success of quantum, and I think that most federal programs as well as state programs have a substantial workforce training component. Is the QSA doing something about that, or do you feel it’s handled well enough by federal and state programs?
Bert: No, no, no, no. And if you talk to anybody in industry, the first thing they will tell you, we need more people. We do not have the right workforce. And what I find very interesting, and especially in the last couple of years, talking to industry, is that there has a very big shift in the type of people that we need. It used to be physicists with PhDs. Now, the latest reports and the latest conversation are talking about people that are maybe a bachelor’s, maybe not even a bachelor’s, an AA degree that are capable of plugging the wires, putting a system together, doing the cooling the fridge down, doing the mechanics of that. That’s a very different skill set that we see come more and more on board. But what’s the interesting challenge is that if you look at AA degrees or a lot of community colleges or smaller colleges, they don’t talk about quantum. So people don’t even know that that’s a career opportunity. So what we’ve focused on is actually focus on going as early as possible. And we’ve taken that to the extreme because right now, actually, we have a very large program that started just in California and New Mexico between the two labs where we are teaching students and teachers at high school levels about quantum. So we’ve had now a couple of hundred teachers come through in the last five years, a couple of hundred high school students. We call that program QCamp. The teachers get hands on experience and they take that back to the classroom. So that’s the first thing. We need to start at that level already, get people excited. Kids excited about going into any college to get ready to be part of the quantum industry. So our next I plan of expanding on that, by the way, that has expanded from two states to, I think we’re at 10 or 15 states now. So more and more state agencies are actually seeing the value and actually contributing, funding to that to make that happen. But what we want to do next is actually going after community colleges. So that’s where we can get the kind of the people that with AA degrees that can do more of the engineering side or simple, the putting systems together and cabling and so on. And we need to reach them because most of them are not aware that this is an industry. And we need to teach them the right skills because even though the skills might be transferable, I think there’s a lot to learn to realize what is needed for quantum computers to be built, or quantum networks, for that matter. It’s the same problem. So that’s our next thing, is community colleges.
Yuval: As we get close to the end of our conversation, I’m curious. You’ve obviously been following this field for quite some time. What are the developments that you were perhaps most excited about, say, in the last 12 months?
Bert: Well, I think the biggest excitement that we have seen so far, of course, is the advent of quantum error correction, right? As I said, in 2020-ish, yeah, people were doing error correction. But now in the last 12 months, that has exploded in a way that people are actually demonstrating at least break even or even start to see some advantage. of it, that is something that I think has really transformed the field. People are excited that quantum is actually much closer now than people were expecting. And that’s the other thing, is like five years. I’m actually optimistic that we can deliver something that would be suitable for the Department of Energy in five years. So that to me is very exciting. And that comes with, of course, all the technology areas growing, right? So seeing the neutral atoms going growing and building systems that are able to do error correction. You see the partnerships with Microsoft and atom computing and Microsoft and Quantinuum showing that they can break even at scale and we’ll see bigger systems coming out of that. So there is a lot of positive things happening in a pretty rapid succession, which is what you would like to see for a healthy ecosystem. them to develop rapidly.
Yuval: I had a chance to write an article not too long ago, and I said that quantum is moving from “one day” to “day one”. Do you agree with that?
Bert: Yes, totally. And as I said, I’m very, maybe in five years, people will say you were way too bullish, but I believe in five years that we will have at least one technology that is going to deliver us a quantum computer of a scale that will rival or exceed what classical computers can do and really deliver new scientific innovation.
Yuval: And last, the question that I ask almost all my guests, if you could have dinner with one of the quantum greats, dead or alive, who would that person be?
Bert: Oh, that’s easy. I wanted to, I would love to talk to Feynman. and then talk to him about what has happened in the last decades and the progress that has been made. And I want to hear how he feels now about his kind of original, well, he might not be the original, but his original idea of use a quantum computer to do quantum simulations, and that is coming to fruition. And actually, it could be a reality now. I would be curious to hear what he would run on these quantum systems at that point in time.
Yuval: Bert, thank you so much for joining me today.
Bert: You’re welcome. Nice to talk to you.