Podcast | Quantum Dots: From Nobel Prize to Silicon-Based QPUs— with SemiQon

The 2023 Nobel Prize in Chemistry was awarded for the discovery and synthesis of quantum dots. Physics is also interested in these little marvels, and they may have a big impact in quantum computing. We need many more qubits of high-quality to tackle complex business problems and use cases. How will silicon-based quantum processing units (QPUs) boost scalability for increased performance into the million-qubit realm? Join Host Konstantinos Karagiannis for a chat with Himadri Majumdar from SemiQon.

Guest: Himadri Majumdar from SemiQon

The Post-Quantum World on Apple Podcasts

Quantum computing capabilities are exploding, causing disruption and opportunities, but many technology and business leaders don’t understand the impact quantum will have on their business. Protiviti is helping organisations get post-quantum ready. In our bi-weekly podcast series, The Post-Quantum World, Protiviti Associate Director and host Konstantinos Karagiannis is joined by quantum computing experts to discuss hot topics in quantum computing, including the business impact, benefits and threats of this exciting new capability.

Read transcript +

Konstantinos Karagiannis: The Nobel Prize in Chemistry was awarded for the discovery and synthesis of quantum dots. Physics is also interested in these little marvels, and they may have a big impact in quantum computing. Find out how quantum dots may take us to the million-qubit scale in this episode of The Post-Quantum World. I’m your host, Konstantinos Karagiannis. I lead Quantum Computing Services at Protiviti, where we’re helping companies prepare for the benefits and threats of this exploding field. I hope you’ll join each episode as we explore the technology and business impacts of this post-quantum era.


Our guest today is the CEO of SemiQon, Himadri Majumdar. Welcome to the show.


Himadri Majumdar: Thanks for having me.


Konstantinos Karagiannis: I want to have you on to talk about a few things, but before we get to the big point of this discussion, give our listeners a quick overview of how you found your way to quantum.


Himadri Majumdar: I’m now leading a spin-off from the VTT Technical Research Centre in Finland named SemiQon. We are the cofounders — we started off as VTters. I have a long history at VTT — had been there for 10 years in various roles. I have a background in applied physics. I did my Ph.D. a couple of decades ago, and then, after doing the due diligence in postdocs and all the other academic activities, I ended up at VTT as a senior scientist, first leading multiple European and local projects, and then moved from science to sales.


I ended up leading the sales activities in the microelectronics and quantum-technologies domain around 2015. Having led that activity, I jumped to the opportunity of leading the quantum activities when we focused on that part at VTT, basically around 2020. Among other things, I led the project with IQM on building the first quantum computer in Finland, and that was a lot of fun. While doing that, we realised that apart from doing superconductors and stuff like that, semiconductors were very interesting, and that ended up with me being here today.


Konstantinos Karagiannis: And whenever I hear about decades, because I’ve been around a while too, I always wonder, were you interested in quantum way back, and when you finished, you’re, like, “There’s nothing to do in quantum, so let me do this other stuff”?


Himadri Majumdar: No, that’s interesting because I started with nanotechnology, and at that time, we were doing research on quantum dots, but for different applications. Quantum, as we know now, was not that prevalent at that time. In that sense, I’m a newcomer in quantum, so to speak.


Konstantinos Karagiannis: Tell us about SemiQon.


Himadri Majumdar: We thought that that name would be the perfect merger of semiconductors and quantum. SemiQon is a spin-off from VTT, and the idea behind the company is to build quantum processors, or QPUs, from semiconductor quantum dots. Silicon is our platform of choice, and the reason for that is, we want to make sure we can build quantum processors for the million-qubit era. Whatever the other modalities are doing now is fantastic and great because they are proving how important and useful quantum computing can be. But we will, at some point, reach the scalability plateau, and that’s when we want to be ready with our platform that allows us to do it in a scalable and affordable way. That’s the long-term goal for SemiQon.


Konstantinos Karagiannis: We just had another company on with that same idea. I’m starting to see this more and more. Everyone’s worried about, how do we get more to the million-qubit level? It’s definitely something that’s on everyone’s minds right now. It’s the quantum zeitgeist, if you will. Normally, on this show, we would expect to talk about the Nobel Prize in Physics. That’s the thing that’s come up. Last year, we had entanglement. That was a big deal, and, of course, the work had been going on for a while — Zeilinger, etc.


But this year, it seems like the one of interest to my audience and my field is the Nobel Prize in Chemistry, which is a little strange. The 2023 Nobel Prize in Chemistry was awarded for the discovery and development of quantum dots. It’s definitely something that’s now buzzing about, and some people might not have even heard of it yet. Can you give a mini-tutorial on what quantum dots are and explain them to our listeners?


Himadri Majumdar: This whole domain of quantum dots was fascinating a couple of decades ago, and that’s when the original works were done. And the reason the Nobel Prize is in chemistry is because it was given for finding quantum dots. Quantum dots, first, are very small particles with diameter of 2 to 10 nanometers. They were discovered in colloidal suspension, in liquid form. And that was the starting point, when we could see fascinating things happening in liquid. We could have colors out of liquid when you shine them because they could excite them to a certain energy level, and that emitted the light. It was fantastic things at that time.


Then, slowly, we saw the emergence of quantum dots in more, let’s say, solid-state form. That’s when we started seeing televisions having quantum dots in them and things like that. It’s fantastic how you can find new ways of doing physics, even though it emerged out of chemistry. But you can do physics — fascinating physics — in those dimensions. For example, we are using single-electron transistors. That’s a quantum-dot territory where we want to confine electrons in a dot and then manipulate it using electrical and magnetic pulses to use them as qubits. The journey started long ago, where you could confine electrons in a dotlike space, in that few-nanometers space. Now we are utilising it to build technologies around it.


Konstantinos Karagiannis: It makes sense — with the Nobel Committee, they were just, like, “We found it in liquid. Give it to the chemistry guys. That makes sense.” But like you said, it has big implications for us. What are some of the benefits to using this approach — that single electron, the quantum dot — to manufacture a QPU? What makes it better?


Himadri Majumdar: Why we wanted to explore this option was, there are two aspects: First, the idea of confining electrons or holes in a dot and then manipulating it is a very elegant way of doing computing. Obviously, there are challenges with multiple aspects there, but the idea of it is very elegant. The second aspect, which is more on the commercial side, is that the semiconductor technology, by itself, is now reaching a very interesting phase. We are talking about a 2-nanometer dimension of semiconductor transistors, and they are going to be commercialised very soon as well.


The semiconductor industry is coming to the point, through Moore’s law and developments as such, where it’s reaching that quantum boundary, so to speak. The whole process that the semiconductor industry went through in manufacturing these devices is already known. And we all know how, according to Moore’s law and how it worked, we could scale up the volume in a small space.


Our approach is, why don’t we take it from the quantum angle and do the device-structure design — everything suitable for building qubits — but use the manufacturing understanding at the same time so when it comes to scaling, we can not spend too much time reinventing the wheel. We can just go the way it went with the semiconductor industry. Those are the two factors that drove us in this direction.


Konstantinos Karagiannis: “We’ve already done this,” you’re saying. “We’ve already scaled this general approach.” But of course, when you did that, you didn’t have to worry so much about the universe making observations of your qubit and things like that. One of the steps toward this is this approach of integrated cryogenic control circuitry. Can you talk about that — how you make this whole thing work together?


Himadri Majumdar: You are right that one aspect of it is the qubit when we talk about quantum dots, but there are a lot of control electronics and a lot of manipulation needed to get the information out of those quantum dots or qubits as well. I cannot tell too much of details due to obvious reasons, but if we look at a quantum computer that exists now, the first thing you see is, obviously, the cryostat. That’s the shiny, expensive fridge there. But you also see lots of cables going in, and they are the ones used to read and control the qubits.


The more we go into higher numbers of qubits, the more we are expected to see those kinds of cables and electronics there. And what we are saying is, obviously, we have the quantum dots and the qubits on the silicon chips. But silicon also offers us the opportunity to create circuits, to create switches or any other components needed to replace some of these cables.


That’s our approach. What about instead of bringing in cables, we do the electronics on the same chip where we have the quantum dots and do the control using those? There, the novelty is that you have to make the chips operate at very low temperatures. That’s the cryo CMOS where it comes from. The initial idea would be to replace some of the cables. We still would need some cables, but eventually, we would try to go to what we call a quantum IC domain — a qubit and the control electronics all on the same chip.


Konstantinos Karagiannis: We’ll talk about error correction later, but was there any goal given to how many, ideally, you’d want in one module going forward?


Himadri Majumdar: We are still looking at it. We have to see how densely we can pack them together. It’s a design challenge as well, not just fabrication and measurement. What we can definitely say is, we can densely pack a lot more than we are seeing now with other modalities in the same footprint.


Konstantinos Karagiannis: What temperature range do these operate within these qubits?


Himadri Majumdar: Therein lies another potential advantage. We are operating in hundreds of millikelvins. As a reference, that’s about 100 times more than where, for example, the superconducting regime is. What we would like to do is take it to the Kelvin range — the space-temperature range of 1 to 5 Kelvin. That allows us not just to be a thousandfold more efficient in a way, but from a technical perspective, this gives us a bit more energy freedom to use them not for cooling but for manipulating and controlling the qubit. That’s another advantage of having, hopefully, hundreds of millikelvins right now, but then going into Kelvin range.


Konstantinos Karagiannis: If you get it to 1 to 5 Kelvin, then you can just float it in space. You wouldn’t have to pay for cooling at all.


Himadri Majumdar: Exactly. Wouldn’t that be fun? But that brings down the cost of cryogenics significantly as well.


Konstantinos Karagiannis: People don’t realise how crazy it gets when you get to the millikelvin range. It’s literally the coldest place in the universe. It’s colder than deep space — colder than anywhere else. It’s wild that we can point to the coldest place in the universe. It’s right here on Earth.


How does this general architecture affect scalability then? You hinted at this, but are there any plans to have chips connect to each other?


Himadri Majumdar: There are many options to look into. First, we want to see how much we get on a single chip, and then the design rules and how good we get entangled — not just the near qubits but the far qubits as well, whether it’s a 1D array or a 2D array, and those aspects that will define the first set of architecture. Then the second set is whether we go 2D with the architecture or we go 3D — do we tandemly stack up chips, or do we expand sideways? There are challenges in everything in that respect. The overall goal is to reduce the footprint. We’d like to compact it as much as possible without losing the efficiency of the qubits. But other than that, it’s still quite wide open how it will be done.


Konstantinos Karagiannis: Do you anticipate needing the qubits to be very close together at all?


Himadri Majumdar: They have to be. From a general perspective, you want them to be close enough to talk to each other, but you don’t want what scientists call cross talk. It’s a double-edged sword, in a way.


Konstantinos Karagiannis: And if you think cross talk is bad in regular electronics, here, you might be having cross talk from other dimensions, from other parallel worlds. That’s even worse — an infinite amount of cross talk. I stick Everettian physics into this whenever I can.


What other barriers to fault-tolerant quantum computing do you see, and how do we overcome them?


Himadri Majumdar: That’s a very challenging question, because fault-tolerant quantum computing is something we all are hoping for, but it’s a very challenging aspect. For example, when Google came up with the Sycamore and they hypothesised about how much you need, the need was one error in 106 or 109. That’s crazy, because from a manufacturing and fabrication perspective, that’s a very stringent requirement, and it’s hard to get there.


But we will see a lot of quantum error correction. We will see an era where to get one logical qubit, we would need to have multiple physical qubits that will also have the error correction, and that will help us get to that point. But there is also the other aspect of repetition codes, or how we do the coding, because that also defines how we can build better processors from that perspective. There are obviously issues like noise coming from manufacturing, noise coming from defects and other places as well. We are talking about dimensions of a few nanometers, so that’s impossible to operate if you have, for example, a speck of dust — that ruins everything. We cannot even think of those things.


All these things are taken into account, even though we achieve, or thrive to get to, the fault-tolerant quantum computing or general-purpose quantum computing, we are still quite far away from that because of the challenges I mentioned. And the other aspect is how you do the codes, how you do the readouts, how you do the electronics — they all add up. It’s a community task to get the problems solved, rather than a problem for a company like us. We cannot ourselves handle that question.


Konstantinos Karagiannis: Do you see this technology being used only in a QPU, or do you think that if it’s not that cold, etc., this could be something for something like an ASIC — like, let’s say, a repeater for a communications network, or something like that?


Himadri Majumdar: Those are potential applications. Obviously, as a company, our goal is to stay focused on the QPU, but there are what we call side streams as well. There will be potential opportunities, like we discussed about cryo CMOs. That can be of interest for other modalities as well — and not just for us. I won’t go into that, because there are business opportunities in those cases I don’t want to divulge. But in principle, there can be multiple short-term benefits of doing this.


Konstantinos Karagiannis: And is there anything, thinking along those lines, about quantum dots that makes them useful for, like, a sensing application?


Himadri Majumdar: Yes. That’s a topic already under heavy investigation.


Konstantinos Karagiannis: I won’t probe too much there. I could sense there are nervous folks listening, going, “He can’t talk about our deal that we have underway!”


How soon do you think we can expect to see a QPU from SemiQon?


Himadri Majumdar: Hopefully, soon. What we are doing is, obviously, working on our first generation of QPUs. We will be competitive from a global perspective, and that’s what we want to do. But since we are a QPU maker, we rely on certain measurables, or KPIs, for the QPU itself.


We don’t build a full-stack quantum computer. If anybody wants to know, how will our QPU work in a quantum computer, there has to be a bit of a wait, because we know this is a challenge. For that reason, we paired up with a company in the Netherlands, and we have a joint project ongoing where we will build the QPU and they will do the system integration in a full-stack demonstrator of a quantum computer with that. By 2026, we should have the first demonstrator of a full-stack quantum computer with our QPU. That’s, at least, the plan for now, but who knows? Things can go even more quickly. We’ll see.


Konstantinos Karagiannis: Do you refer to it as the first Finland quantum computer?


Himadri Majumdar: That accolade has already been taken by IQM and VTT when they built it with the superconducting platform. That has already been demonstrated. But from a spin-qubit perspective with semiconductors, that’s still ongoing, let’s say.


Konstantinos Karagiannis: Is there any government partnership or anything happening. Are they interested in using it?


Himadri Majumdar: We hope there will be, because with any new technologies or any new breakthroughs, there has to be some sort of government programmes or government procurement. We hope there will be opportunities like that as well. The interesting bit is that the way the quantum computing is evolving now is through integration with high-performance computing or supercomputing centers. That would be one way of adding value to those HPC centers by quantum. Hopefully, spin-qubit will also have an opportunity to get to that direction as well.


Konstantinos Karagiannis: Around 2026, maybe this QPU is integrated into something. Did you give any thought to a roadmap from there? How does that go down to 2030, for example? What numbers?


Himadri Majumdar: Like I said in the beginning, all the modalities, including us, we are already, let’s say, not that advanced, but with all the other modalities, we are talking about hundreds, even perhaps a thousand at most. But that is not where we are. It gets interesting when you get to 100,000 and plus with the physical qubits so that we can get error-corrected qubits as well from them.


But the goal for us is, as a company, to reach thousands of qubits, or a thousand qubits, by the end of this decade. And then, what we hope to achieve is something equivalent to Moore’s law, where we can double, hopefully every year, the qubit count, and that takes us to a million by 2040. That’s our goal at this point. And we have to keep in mind that other modalities might get close to that as well before us. But the issue there will be, will those be scalable solutions that are deployable at an affordable price? And that question is still very open. Our goal is to make it scalable and affordable based on everything we are doing.


Konstantinos Karagiannis: It’s interesting you said Moore’s law, because while you might achieve something like doubling a qubit count, you’re doing way better than that in performance. Every time you add a qubit, it’s literally doubling the addressable information states. You’re doing way better than Moore’s law, technically, if you achieve Moore’s law numbers. And on that note of performance, do you have any idea of fidelity — what we’re looking at for these qubits?


Himadri Majumdar: Well, the fidelity expectation are very high. What we are trying to figure out is how we can strike a balance between scalability and the efficiency — not just fidelity, but all the other measures as well. This is a constant discussion within the community: How do we approach that? Do we have error-correction mechanisms, or do we do it the NISQ way, where we have a lot of physical qubits, which allows us to get logical qubits? Our approach at this moment is more on that side — that we want to increase the qubit numbers so that we get a sufficient number of logical qubits out of that system. Then comes, obviously, the question of, what are the fidelities? What are the computation capabilities? That remains to be seen. We still don’t know. Obviously, we target the best we can, but there is a balance between performance and scalability — or volume, in that sense.


Konstantinos Karagiannis: What no one ever seems to be able to share is, how do you go from understanding your fidelity to understanding what your estimate of a logical qubit is — that ratio? IBM said 1,000, 1,001. Where’d they get that from? Did they do it based on some calculation they didn’t share about the fidelity times some factor? Do you have any thoughts on that — what it would take to figure out that ratio?


Himadri Majumdar: For that part, we rely on the quantum information scientists. They have to tell us what the potential codes are that they can write on the qubits that are given and how they can get the information out. We don’t know exact numbers from that perspective. The errors from the architecture. It’s also coming from the hardware, the process and everything. All the errors add up. IBM’s statement of a thousand qubits is as good as anything. We can tell better when we can use our physical qubits to get to a logical qubit. Then we are wiser about how many we need to do that. Now, a thousand sounds like a reasonable number.


Konstantinos Karagiannis: One last point: This is quantum dot, and there are other companies, obviously, looking at this, and Intel is a good example. They have their quantum development SDK. When it comes time to make this full-stack, do you envision leveraging something like an Intel SDK and making this another back-end target? Or do you plan on pushing for a completely new stack all the way up?


Himadri Majumdar: We want to leverage the existing know-how and existing technology and semiconductors. From that perspective, it would make sense to rely on a bigger player to define those aspects. But having said that, there are certain challenges when it comes to designing quantum dots or QPUs, and they come from the stack itself.


Without going into details, I can say that the idea that we can take the current transistors and hope they will work as quantum dots is challenging. We are taking a more bottom-up approach in that sense. We define the stack which is best suitable for building QPUs. From that perspective, we might not be able to use the PDKs or SDKs available already. We might have to tweak some of those to some extent.


Then the question comes, Are we doing it in an industrially scalable way? Yes, we are, but that needs to be adapted to a larger foundry when the volume gets to that point. This is another aspect which not just we but also our competitors are looking at: What are the opportunities, and how can we take advantage of the existing SDKs and PDKs? But it’s not clear yet how it will go.


Konstantinos Karagiannis: When something like this Nobel Prize happens, does it instantly open more doors? Does it make conversations go more smoothly? I just wonder if, behind the scenes, when a company is involved with their modality and something that gets that kind of attention, does it become a “Yes!” moment — like, you’re ready to take advantage of that?


Himadri Majumdar: Obviously, it does. You have to ask that question to a marketing manager, and they will be like, “Yes, of course it is.” But being in that field long enough, you’d guess that that is obvious. That is inevitable. It’s bound to happen. It was the same thing — last year’s Nobel Prize as well. We were all happy that it happened, but a Nobel Prize makes it visible to the community and helps us get a wider audience for what we are saying, but the challenge remains the same. In a way, it’s good thing that everybody’s understanding and learning about this thing, but it’s also a challenge because the expectations go up as well. It’s good and bad, but we want to look at it from the good perspective.


Konstantinos Karagiannis: Well said. Thank you, sir. With that, I’ll wish you best of luck in reaching those timeline goals.


Himadri Majumdar: Thank you for having me.


Konstantinos Karagiannis: Now it’s time for Coherence, the quantum executive summary, where I take a moment to highlight some of the business impacts we discussed today in case things got too nerdy at times. Let’s recap. The 2023 Nobel Prize in Chemistry was awarded for discovering and synthesising quantum dots. These are 2- to 10-nanometer particles with properties that change based on size and quantum phenomena. For example, their color changes based on size, leading to their use in QLED screens.


Quantum dots are also used in single-electron transistors that can be manipulated with magnetic fields or voltage pulses to work as qubits, where the electron spin states act as a 0 or 1. Quantum dots can be manufactured in large quantities with precise control over their size and properties. This elegant way of manufacturing makes them ideal for building large-scale quantum computers, and it may be possible to stack qubits two or three layers deep.


The hope is to scale based on preexisting fabrication knowledge and reduce the overall footprint. SemiQon is also working on controlling these qubits without the wires and cables used in other modalities like transmon. The hope is to have control circuitry that operates at low temperatures right on the chip called cryo CMOs, and the temperature is warmer than those other modalities.


Quantum dots operate in hundreds of millikelvins and may stay coherent in single-digit Kelvin one day. That’s a thousand times warmer than transmon. If the approach works out, lowering cooling costs and potentially improving performance, SemiQon might be able to use cryo CMOS for other devices and is considering implementing quantum dots in other devices, perhaps in quantum networking or sensing.


The first generation of SemiQon QPUs in a complete system should appear around 2026. They hope to achieve a Moore’s law–like doubling after that yearly to scale into thousands and millions of qubits.


That does it for this episode. Thanks to Himadri Majumdar for joining to discuss SemiQon, and thank you for listening. If you enjoyed the show, please subscribe to Protiviti’s The Post-Quantum World and leave a review to help others find us. Be sure to follow me on all socials @KonstantHacker. You’ll find links there to what we’re doing in Quantum Computing Services at Protiviti. You can also DM me questions or suggestions for what you’d like to hear on the show. For more information on our quantum services, check out Protiviti.com or follow Protiviti Tech on Twitter and LinkedIn. Until next time, be kind, and stay quantum-curious.