Maud Vinet, CEO and co-founder of Quobly. We discuss the company’s mission to build large-scale quantum computers based on silicon qubits. Maud explains how silicon offers unique advantages in scalability, leveraging decades of semiconductor expertise, and why the biggest challenge lies in qubit control rather than fabrication yield. Maud shares insights on her transition from semiconductor R&D to quantum computing, the partnerships Quobly has formed with fabs like STMicroelectronics, the anticipated timeline for achieving significant qubit counts, and much more.
Transcript
Yuval Boger: Hello Maud, and thank you for joining me today.
Maud Vinet: Hi Yuval. Well, thank you for having me.
Yuval: So who are you and what do you do?
Maud: What do I do? Well, I’m CEO and co-founder of Quobly. And in Quobly we make large-scale quantum computers based on silicon.
Yuval: When do you think a large-scale quantum computer based on silicon will be available?
Maud: Oh, that’s the billion dollar question. Well, honestly speaking, today we already do have small qubit chips. And we are already processing hundred of thousands of qubits. We are speeding up with the arrival of the semiconductor industry. We plan to deliver 100 qubit chip by 2027 and 1000 logical qubits by 2033.
Yuval: The apparent advantage of silicon qubits is that silicon manufacturing is well understood. People make chips with billions and billions of transistors. What are the drawbacks? If silicon is so good, what’s taking so long to have a computer?
Maud: Well, yes, you pointed it out. The major drawback is that it’s long to set up. The reason for that is that it’s exactly what you said. It takes 10 years to set up a silicon technology, an advanced transistor technology because everything is so compact. Actually we control things at the angstrom size or we do layer by layer deposition. So the major drawback for silicon is actually the time it takes to instantiate the technology. To be precise and provide you with a concrete example, in silicon qubits, the quantum information is encoded in a spin, in the spin degree of freedom of a single electron. And that means that we manipulate a single electron within a silicon crystal.
So we need to pay attention to each and every charge that is in the environment. And this optimization is somehow slightly different than what we’ve done in the transistor technology. And that’s the reason why it takes some time.
Yuval: And you yourself come from a silicon background, right? Tell me about your journey to start this company.
Maud: Actually I earned a PhD in quantum physics. I’m looking at the date, so something like 25 years ago. And I was frustrated at that time because I felt quite lonely in my PhD being able to talk with myself about what I was doing. I wanted to be part of a team. So that was the time where the semiconductor industry was looking for physicists to find solutions to Moore’s Law since we were predicting the end of Moore’s law. So I joined the semiconductor industry in CEA-Leti, which is a technological research organization. And I worked for the semiconductor industry for 20 years. I developed new technologies and just made progress and transferred them to the industry.
But at the same time I kept the relationship with the lab where I earned my PhD and for 20 years what we’ve done is that we’ve cooled down samples that we were fabricating for the semiconductor industryt. First there were their charge properties and then we looked at their spin properties. And in 2016, we made this worldwide first where we demonstrated that we can make a qubit out of a transistor that we were fabricating for the semiconductor industry. And that time, I was leading the advanced computing department in CEA-Leti where we were developing the hardware to tackle challenges of computing.. We were scaling down the transistors, we were working on memories for in-memory computing for AI and we were working in quantum computing. And then quantum computing picked up and I realized that we’ve got a potential and we launched the company.
Yuval: Because there are so many silicon companies that are very large and established, with plenty of resources. Why would a small company succeed in having a quantum computer, say before in intel that, that I think is also working on silicon qubits?
Maud: Yeah, that’s a very good question. You understand that we’ve worked a lot on that before we decided to launch the company. Well, I think it demands technology pivots from transistor to qubits at cryogenic temperature. And with the background, I told you about, we’ve been cooling down transistors for more than 25 years now. And we’ve gathered us some, some pretty fair amount of background. And to go back to your question of large semiconductor companies, they’re currently busy dealing with AI.. And so those companies, they’ve got a, they’ve got a business to handle and quantum computing is still a distraction for them.
In semiconductors, the way it works is that you develop technologies where you see a return on investment within the next five years. And so that’s not so easy for a big semiconductor company to go and deal with quantum computing. Why invest where maybe you will be successful in many years, while you know for sure that in two years from now you will make money with AI.
Yuval: When you look at other modalities, for instance, superconducting, so we see superconducting quantum companies say that they can only have a few hundred qubits on a single chip or chiplet before they get into yield issues and therefore they have to resort to networking between those chips. Does that problem also exist with silicon qubits?
Maud: I would say that no, the problem does not exist. We really leverage the know-how and the expertise of the semiconductor industry to build our qubits. So far. Just to provide you with numbers, I prepared for tonight and I counted the number of qubits that we fabricate in one single wafer and currently we fabricate more than 100,000 qubits on a single wafer. Our problem is to measure them and control them. But most of the time they do work.
We don’t face serious yield issues. our challenge is really in the control of those qubits. This is what we need to handle now.
Yuval: When we think about error correction, we often think about classical compute capabilities that have to be alongside qubits. Would it be correct to assume that in silicon qubits the classical compute could be essentially on chip or would it be a separate chip?
Maud: No, your assumption is good and this is what we’re trying to do. And that’s actually the rationale that we pursue. Our strong motivation is to start from existing transistor technology to be able to leverage the cointegration of classical electronics with the qubits. So we started from existing mature FD-SOI technology, which is a technology that is currently fabricated by Samsung, Global Foundries, and ST. And we turned this technology into qubits. When we are doing that, we can co-integrate the classical electronics. Currently we demonstrated that we can generate the signals to control the qubits and the amplifiers to read out the qubits on the chip. With the qubits, next step is indeed to add some more functions and we are developing the modules and just building like a puzzle to have the quantum error correctionon chip.
Yuval: If I remember correctly, a couple of months ago, it was around the time of Q2B Silicon Valley, you announced a partnership with a fab or a chip maker. Could you tell me a little bit about that and why is it important for your business?
Maud: Yeah, sure. Well, you kind of raised the point. Silicon qubits are promising. Why aren’t there more silicon spin qubits being demonstrated by now? The point is that the best quality in silicon is in fabs. And we spoke about the fabs, the fact that the fabs are focused on classical electronics. So far most of the demonstration in silicon qubits have been either made in academic labs or in R&D 300 millimeter clean rooms. And this is what we’ve done for the last, let’s say five years. We’ve been laser-focused on demonstrating qubits in R&D clean rooms, taking into account all the constraints of commercial fabs.
The next step now is to transfer the technology to actual commercial fabs where we will exploit the quality, yield, and reproducibility of the semiconductor industry. That is what partnership with the chip maker STMicroelectronics is about. really a tipping point for the company because now we’ve got access to the best quality.
Yuval: Going back to superconducting qubits, we know that superconducting qubits have limited connectivity. Maybe two or three neighbors for every qubit, sometimes five or six in newer architectures. Does that also hold for silicon qubits?
Maud: So far, yes. This is solid state and we don’t have global excitation modes. So this is nearest-neighbor interaction. So the architecture we’re building is with four nearest neighbors. And there is some research work to couple silicon qubits to superconducting resonators to increase the number of neighbors. But at this stage this is still experimental research.
Yuval: When you think about producing a quantum computer from silicon qubits, what is the cost driver? Is it a cooling system? I mean, I know in some other modalities maybe the lasers are expensive. What do you think will be the cost drivers for your type of computer?
Maud: Well, that’s a good question actually currently for sure, there is no doubt about it. The cost driver is the cryostat.But cryogenics is not yet a mature industry. So as the number of cryostats increases, we can foresee that cost will go down and we will try to drive the cost reduction for this industry. So I believe then we will have the control electronics that will be competing with the cryostats for cost.
Yuval: You’ve been doing this for a number of years. What have you learned in the last year that you didn’t know before?
Maud: Well, I’ve learned so many things. Every day when I go back home, I tell my kids, you can’t believe what I’ve learned today. So that’s a general question and as a general answer I would say thatI learn every day the power of the team and the fact that collective intelligence really is able to tackle challenges that you didn’t think you could solve.
Yuval: Tell me a bit about the company. How large are you, how are you funded? Anything you could share along those lines.
Maud: We incorporated at the end of 2022. We’re currently 70 people working on the project and we raised our seed round in June 2023. We raised 19 million. It was the biggest European seed round for a hardware quantum computing company. And currently with the agreement that you mentioned with ST is the trigger to speed up and we’re pretty close to launching our Series A.
Yuval: Where would the fab physically be? In what country?
Maud: Oh, that’s a good question. So currently we don’t build fabs, we benefit from existing fabs. We’re getting started with STMicroelectronics fabs, which is located in France, close to Grenoble. what we appreciated in ST is that they’ve got the know-how and expertise in technology development as well. That’s what we were looking for. So, so far this is in France and we are working with technologies that are around 20nm so it can be made in many places in the world.
Yuval: Forgive my ignorance, but I know the EuroHPC program funded several quantum computers of different modalities. Photonics, neutral atoms, trapped ions, superconductors. Is there also a silicon computer that they’re funding?
Maud: Not yet. Actually, to tell you about the Euro HPC program, they started with the most advanced technologies and they’re trying to span all the quantum computing technologies. So the next call, which is sometime this year, will be focused on silicon qubits.
Yuval: Other than silicon qubits. What’s your second favorite quantum modality?
Maud: Oh, that’s a very difficult question. Well, I must admit that I’m a condensed matter person. I just know electrons and so I’m fascinated by everything that deals with things that I don’t know. And I love neutral atoms for their physics, for the fact that they are just, oh, perfect objects.
Yuval: And as we get closer to the end of our conversation, I’m curious if you think about your journey, where do you expect to be? Where would you like to be 18 months from now?
Maud: 18 months from now? we will have demonstrated qubits on ST technology and the chip will co-integrate the electronics. We’re building a puzzle and I’ve set this piece of the puzzle as a big milestone.
Yuval: And I do get a sense that you are prepared for this question but you can answer it even before I ask. Go ahead.
Maud: Marie Curie.
Yuval: I think I could guess why you would want to have dinner with her but I’m curious to hear your answer.
Maud: I was prepared for your question, obviously. But what I like about Marie Curie is that she’s knowledgeable in many, many fields and she did not have any barriers that she had to go through. When you look at her biography, she didn’t stop herself because of something. She did what she wanted to do independently of what people were thinking of her. That’s amazing when you look at it from the perspective of the time that she was living in.
Yuval: Maud, thank you very much for spending some time with me today.
Maud: Well, thank you very much, Yuval, for having me.