My guest today is Sebastian Weidt, CEO and co-founder of Universal Quantum. We discuss his company’s unique approach to building scalable trapped-ion quantum computers. Sebastian explains how Universal Quantum focuses on overcoming key scaling challenges—such as error-free modular connections, laser-free control, and integrated electronics—rather than rushing small systems to market. He contrasts their 70K cooling and qubit-shuttling architecture with other modalities, addresses trade-offs like gate speed, and shares why algorithmic execution speed can offset slower clock cycles. We also talk about the company’s €70M DLR contracts, its emphasis on software alongside hardware, and Sebastian’s view that quantum utility will likely emerge north of 10,000 qubits. Reflecting on entrepreneurship, he shares lessons learned about market education, the company’s engineering-heavy team, and much more.
Transcript
Yuval Boger: Hello, Sebastian, and thank you for joining me today.
Sebastian Weidt: Thanks so much for having me.
Yuval: So who are you and what do you do?
Sebastian: My name is Sebastian Weidt. I’m a bit of a quantum physicist and a bit of a business person, CEO and co-founder of Universal Quantum. And I’m also still a professor in quantum computing and entrepreneurship. So that’s a fun combination for you there.
Yuval: Absolutely. Universal Quantum. You guys do trapped ions, right?
Sebastian: Yeah, absolutely right. Best performing qubit. We love the trapped ion qubit. And we, in particular, focus on how to scale that amazing quality up to the 100,000–million qubit scale. That’s very much what we focus on.
Yuval: Would you give me a trapped ion primer and especially there are other approaches to trapped ions, right? There’s quantum and ion Q, I guess now Oxford Ionics, there’s Quantum Art, there are all these people that, like you guys, that from afar seem the same, but as you drill up closer, they’re vastly different. Could you make some sense out of this landscape for me, please? Yeah, there’s a lot going on in trapped ions, for sure.
Sebastian: I think, you know, you’ve got great companies, like Quantinuum, who, you know, they have machines in the market and, you know, very much shown in the commercial world how why this qubit works. And, you know, so have many of the other players that you’ve mentioned. I think, you know, Universal Quantum takes a little bit of a different approach in that we don’t have machines in the market right now. That’s not been our focus. We’ve really been focusing on, like I was saying at the beginning as well, how do we we, you know, how do we get to a point where we can really naturally scale this up to where it ultimately needs to go to, so to the quality and the qubit numbers? And I think when you look at trapped ions, you know, I think people often agree that it’s a phenomenal cubit. It really is top notch. But I think people, sometimes find a hard to see how that really scales up all the way to where we need to go to. And I think this is really where we come in and even predates versus quantum back in academia when we did a lot of the foundational work, you know, really finding solutions to some of the challenges that people often point to. So be that around, you know, how do you connect the modular architecture together with trapped ions? You know, people often point to batoning interconnects. And I think it’s been shown that that’s not easy at all and a big stumbling block to scaling up. So, you know, we developed an alternative and demonstrated that we have an error-free way of connecting modules together and it’s actually a key enabling block for us to scale up. The other part, I think, around control and I think this is where the field is starting to maybe move towards that as well, where often people use laser beams to do the quantum logic in trapped ions. And while traditionally I think that’s given us really nice results, again, how those lasers scale up to the million cubit level has been a bit of a challenge to to see how that works. So we’ve, you know, many, many years ago decided to take completely different routes, develop a laser-free technology, which nowadays is becoming a bit more mainstream with, you know, some of the acquisitions that happen in the market where, you know, INQ seems to have pivoted away from lasers to a laser-free approach. So I think we’re, you know, because of our focus on scalability from day one, these are the sort of things, the roadblocks that we’ve tried to remove, as quickly as possible. And I think the other bit is when you talk about modular architecture, you have to talk about the module itself. And across the board in Trapped Irons, people love using chips, as we find in quantum computing in general most of the time. But I think, again, when you think about scaling up, you do have to think about integration, integrating some of the control electronics into a chip. And that’s something that I would say in the trapped iron sector, people don’t really do very much of. We do that and happen doing that from day one. So we develop these fully integrated quantum computing modules that you can then effectively stitch together in an error-free way to scale up in a nutshell. That’s where EQ sits.
Yuval: So let’s break it down bit by bit. It’s okay if the answers are short or long. It’s entirely up to you. Do you require cryogenic cooling?
Sebastian: We don’t have to go to millikelvin temperatures. So we sit at roughly 70 Kelvin, but we would call my scalable cooling list. There’s plenty of cooling power available there. And operating that at the million qubit scale list, it’s not a big headache. So 70 Kelvin is where we’re sitting.
Yuval: So that’s probably much easier to achieve. And how’s that power consumption-wise relative to millikelvin cooling? Huge, I mean, hugely better, right?
Sebastian: I think, again, I don’t want to do the number. for, you know, for other systems, but we know that we, we do have, you know, orders of magnitude advantage over this sort of millikelvin temperatures. You know, cooling is often seen as one of the headaches for really getting to large scale in certain systems. We don’t really have that. The qubits. We don’t need to create these big cooling plants that people often talk about. We don’t have that required, unfortunately.
Yuval: The cooling, it’s the qubits themselves. Do you need to shuttle them or how? How do you get interaction between qubits?
Sebastian: Yeah, so we love the shuttling qubits. So in trapped ions, what you can do is you can move qubits around in space. And not all trapped iron companies go that way. Again, a continuum does. We do as well. And again, I point to them to, you know, as a proof point in how way that works. Now, one of the advantages, one of the reasons we do like to move these things around is that you get this really nice enhanced connectivity between qubits. So the ability for a cubit to talk to non-nearest neighbor qubits in a reasonably error-free way. When you then translate that to algorithmic capabilities, you get significant advantages and efficiencies by having that sort of connectivity.
Yuval: So we get that through the movement of qubits. Some technologies, and obviously I see that in neutral atoms, allow for parallel gates where you could have a single operation on multiple qubits. Does that also happen in your implementation? Yeah, absolutely.
Sebastian: So, and that is a great, you know, great point for neutral atom systems. In trapped ions, again, at least the way we’re doing it, is that you have a two-dimensional grid of qubits. You basically separate that out into different types of functions. So this could be the processing part, this could. could be a readout part, a memory part, and we move the qubits between those areas. But you can absolutely have multiple memory or processing or readout zones on your architecture, and therefore you can do these parallel operations.
Yuval: Every modality has a drawback. Sometimes it’s speed, sometimes it’s scalability, sometimes it’s fidelity. What would you consider the drawback, if any, of fewer implementation?
Sebastian: Yeah, this is hard, right? building quantum computers is hard and I don’t think anyone should say otherwise. I think, you know, if you look at trapped iron, often what people would point to, in many ways quite rightly so, is that the gate speed side. So if you compare that to some of the faster implementations like a superconducting qubit machine, trapped iron systems are slower on that bit. I would say that nowadays that comes with a bit of an asterisk and we’ve done work in that That’s why, where there’s often this focus on clock cycle or gate speed. But actually, I think what you need to do is you need to go a bit higher. And at the end of the day, what an end user will care about is how quickly can you solve a problem that they have? And when you actually get to the sort of algorithmic execution time and you take error correction into account, which is really important in quantum computing, and you then crunch the number on a trapped iron system, assuming for a second it has these advantages that I talked about around connectivity, maybe slightly higher fidelity than what some other systems may be able to get. What you find is that for a quantum computer of similar system size to a trapped iron system, we can get similar execution speeds than a 7.5 qubit machine. So while on the gauge speed side, there may be a couple of order of magnitude difference in speed when you get to the algorithmic execution speed, all of a sudden we can actually get to a very similar time. That’s really work that’s being pushed out at the moment.
Yuval: And, you know, I’m keen for people to know about that because there’s, I think that’s quite exciting. So conceptually, you’re saying the individual qubit fidelity is higher. So you need fewer qubits per logical cubit to achieve a certain logical cubit fidelity. And then because you have essentially all-to-all connectivity, you can have fewer operations, which in turn also reduce the chances. of having errors. Is that about right?
Sebastian: Yeah, and then you can, you can, you know, use those qubits that you’ve just freed up to effectively do things more in parallel and thereby actually speed up the computation. So when you then compare like for like, you know, same number of qubits between, you know, these two examples, you can all of a sudden make up some of those speed penalties you have on the clock cycle time.
Yuval: You mentioned that you’re a professor of quantum computing and entrepreneurship, which is a fantastic combination. How did you think about sort of the process of going to market? Because as you mentioned, some companies are already in the market in terms of computers that one could access. You’re describing a situation where you were focused on something else and the computer will show up in the market in some point in the future. This is sort of, isn’t that a scary decision sort of could be a, hero or zero decision for you?
Sebastian: I think so, absolutely. I wouldn’t characterize it as scary for us, but it is most certainly somewhat view it is a bit riskier. But I think the way we view it and given what we are really passionate about, it’s the only way. I think it is very hard to be keeping yourself very busy with pushing, you know, small-scale machines into the market, trying to squeeze the most out of this NISC era that we’re in at the moment. and at the same time maintain the focus on actually scaling up as quickly as possible. So the view that we took is that, you know, look, let’s just focus on us for a while and make sure that all of this gets on a foundation that does naturally scale. And when we’re at the point where, you know, we really get to these sort of more exciting qubit numbers and can really show that we continue to scale, that’s really the time to then engage the market in a far more aggressive way. I think that way we’ve been able to just focus a lot more. That’s not to say it’s great that we have machines in the market. It’s a great engagement tool. We have to educate the end users. We need to show what this technology is capable of. But I don’t think it was the right decision for us to try and join that very busy sector and try to beat someone on, I have one more qubit than the next one, right? That’s just not what we’re passionate about. And also, coupled with that we’ve always said that it is I believe that there’s very little in NISQ.
Yuval: And therefore, that sort of focus in trying to get to more fault-tolerant systems as quickly as possible was always our big goal. I think that when we spoke a couple of years ago, it was right after you announced your deal with DLR, the German organization. How did that come about? and what did you have to do to fulfill their requirements? Did you have to open an office in Germany, a big development center? Tell me about how the company sort of works these days.
Sebastian: Yeah, sure. So, yeah, so a couple of years ago, very fortunately, actually, we got two contracts from the DLR, I think combined value about 70 million euros. And that is for two quantum computing prototypes that they were keen to have access to. Now, how did that come about? We weren’t actively looking for this at the time. However, and I take it back to what I said earlier, that we had this focus on and still have this key focus on scalability. Now, when Germany then announced the call, they were very clear that they didn’t just want a system with X number of qubit. you had to really demonstrate and build something that naturally scales up. And it turned out that that was what they were really after was so aligned with our roadmap and our vision for quantum computing anyway that we did then decide to engage. Very fortunate that they gave us, you know, both the contracts that we went for. But I think that that was number one. There was real alignment with what we wanted to bid and what the customer wanted here. And I think then in terms of how we execute this now, I’m a strong believer that, you know, you ought not to build a quantum computer, someone then throw it over the border. I think it’s about ecosystem building. It’s about helping nation states to get quantum ready. So what we are doing is part of that deal. We’ve set up an entity in Germany. It’s a major base for us these days. A lot of great work going on there. And we’re building these machines into the facility that the deal allows us building for us there in Hamburg, actually, north of Germany. So it’s a major effort for us in Germany these days.
Yuval: When will a machine be delivered or when will the public be able to see a machine?
Sebastian: I think I can’t speak for the DLR and exactly what their plans are to make that available. We’re hitting our milestones. I’ve been agreed with the customer there and at some point. you know, when they feel it’s right, I’m sure they make that available. But I don’t know how broadly public they would go with that. If they’re using more for internal usage, I can’t comment on that necessarily. How are you thinking about software?
Yuval: Some modalities require that to take advantage of the special capabilities of the machine, you have to write software in a different way, whether it’s to take advantage of parallelism special gates or something like that. How is it for universal quantum?
Sebastian: Yeah, I think the, I think one of the big errors, and I often make that error, you know, we talk very passionately about the hardware, because there’s often clear differences there to highlight, but the software is really, really important. And at you queue, one of the biggest teams is actually software, because the, you know, without the software, this whole thing won’t work effectively. So, yeah, I mean, we take a full-stack approach, but the further away we get from the hardware, the more we do open up for collaboration and effectively build the software stack, as modular as possible, so that we can slot in black boxes of what great other people have worked on either open source or some of the great software companies out there. But I think if you go down to the machine code level equivalent, it’s so different between different hardware approaches, that you do really have to, in my view, at least, retain that in-house and use that to really squeeze the most out of these machines and make sure they run optimally.
Yuval: What is your estimate for quantum utility or quantum usefulness? You mentioned that NISC is not very useful. When will be that OMG moment, a quantum computer really does something that has tremendous business rally?
Sebastian: Yeah, a great question. I think the, I can’t reveal our roadmap here on your great podcast, unfortunately today. But in my view, I think if you look at the applications out there, I would draw the utility line, you know, somewhere north of 10,000 cupits. I think, you know, if you get to that, maybe a bit beyond that, there’s some really interesting things that will start to happen. always depends on where your fidelity is exactly sit. But I think there’s some really interesting early applications that were kick in. And as you know, as you then scale up more and more, what would happen in terms of what time scale associates, that’s always the big question. But like I said, for us, I can’t review that here today. But this is the sort of thing that we actively, you know, work towards and trying to get so as quickly as possible.
Yuval: Going back to the entrepreneurship side, two-part question. One is, what do you wish you would have done differently in building this company? And related to that, if there’s somewhere in the world, another professor of quantum and entrepreneurship, is it too late to enter the quantum hardware market? Or is there still opportunity for new approaches, new modalities, new companies?
Sebastian: I mean, I think I’m obviously very much brought into what we’re doing. And I think the exciting thing for us is that we do have this very clear path to utility, to fault-tolerant quantum computers. I would, however, not go as far as saying, look, no one should enter this game. If you have a great idea, if you’re passionate about what you’re doing and you feel you, you know, you’ve got a better way of doing things than other people. I mean, absolutely, go for it, right? This is a marathon and we have a, you know, a long way to go. I know lots of people are, you know, shouting very hard to be like, this is it. We’re all done, you know, we’re, you know, let’s, no comment on that messaging. But I think we need to be open. We need to be open to new ideas always. We are at Universal Quantum. If I compare to, you know, what we had plans, you know, when we started the, the company and how much has changed, you know, certain things have simplified because you get great engineers on board who, you know, can leverage their great experience and simplify things. It’s quite different now to what we were thinking five, six years ago. And I think that we’re continuing we ought to be open for for new ideas and approaches. So I would always encourage people to make that leap into the commercial world. I most certainly think it’s a, it’s a fantastic one. But yeah, I mean, it’s becoming obviously harder and harder, right? And if people are getting further down the journey, getting closer to utility, they’d have to make that choice in the end. But generally, I’d always encourage that step. I’ve been very, very happy with that.
Yuval: And on the first part of the question, what would you have done differently?
Sebastian: Hmm. It’s a good question. I think we’ve done We’ve done quite well. I think things have generally worked out so far as planned. I think, you know, we could have maybe been a bit more aggressive in the early days in really getting the message out in our view and how to do that. We really chose to, you know, heads down and focus on, you know, and how we thought it’s best to scale up quickly. and a lot of attention and money went into NISC. And I don’t think that works out very well for a lot of the funding sources, so to speak. And I think we could have maybe had a bigger role in that to educate the market and really point to what was really, really important. So yeah, we could have been a bit louder on that side. But I think, you know, through learning, I think people have realized this now anyway, which is working in our favor. But I think we could have done better on that side. By the way, how large is the company today? there’s just over 100 people of us at the moment.
Yuval: Mostly physicists or what’s the sort of distribution?
Sebastian: No, not at all. We’re very, very engineering focused. I think one of the things you realize quickly, if you want to move away from, hey, I want to play with 10, 20 qubits and do some great things with that and you really want to scale, the engineering needs to be rock solid. And I think so we very, very early on decided to make this quite, an engineering focus and heavy entity and the engineers do outnumber the physicists and the company for sure. Physicists are still really important. It would be clear on that. I’m still a physicist at heart as why, but you do really need the world’s best engineers to pull this off as why.
Yuval: In last, hypothetical, if you could have dinner with one of the quantum greats, dead or alive, who would that person be? who would that be?
Sebastian: Good question. It’d be nice to have dinner with something like Von Neumann you know i think that would be that would be incredible um Yeah, I stick with that. That’s interesting.
Yuval: Sebastian, thank you so much for joining me today.
Sebastian: Yuval, thank you so much. Great having being here.