Pasqal Holding SA (PSQL) August 2026 Conference Transcript

Hi, everyone. Thanks for sticking around for the, I believe, the final session of the day. Kingsley Crane, a technology analyst here at Canaccord. We have really pleased to have Pascal with us and Wasik Fokhari, who's CEO. Thanks for joining. Wasik, so Pascal has a decade of Neutral Atom research heritage. You have a Nobel laureate, co-founder, and Elaine Aspect. You came in as a CEO with significant operating experience at some tech leaders and deep tech at Amazon and Google. So maybe you could just start off by telling us how you came to Pascal and how you've seen the business transform in your time here since joining us, Chairman.

Great. Thank you. First of all, thank you for inviting me and great to be here. So, you know, I actually went to college here in Boston, went to MIT for my undergrad and PhD. So this feels like I'm back home. And I remember when I was an undergrad, I made a list of things that I would work on in my life. And one of them was quantum computing. So I still have that list. I have that piece of paper. And many of those things have been checked off. Some are still there, thankfully. So, but for the next 20 plus years, this was entirely an R&D field. And I kept up with it. And it's all good. I moved from MIT, went to Penn as a faculty member, went to Silicon Valley, did a bunch of startups, became a VC, left that, joined Google, and in all of that, all I'm doing is learning how to build a business, how to build businesses, how to scale businesses. I had some exits, so I understood how important it is to not think about technology, but to think about customers and the problems you're solving. After Google, I went to Amazon, used a report into Andy Jassy, and Amazon is a ferociously, operationally excellent company. And customer obsession is a big, big obsession of Amazon. So I was part of the team at Amazon where we decided to build Amazon's quantum computing effort. First we built a bracket, and then we started the hardware effort. Now that was a eureka moment for me because Amazon does not do fundamental R&D. The reason we were able to do that and convince Andy and Jeff to fund that was because we believed at that point that quantum computing can be reduced to an engineering problem and not just scientific pure R&D. So after I left Amazon, I knew this is what I was going to do. But having seen a lot of these traditional modalities, which are superconducting, which are iron-based, which are photonics, and there are very well-known brick walls for these modalities. What we needed to do was look at a fresh approach, and that's where neutral atoms came in. And eventually I joined Pascal as chairman, and then of course now as CEO. But during this time, there are a couple of things that have happened with Pascal that are actually very material, I think not just for Pascal, but I feel these are good signals for the industry as well. Number one, we are now second only to IBM in terms of deployed, operational, high-complexity quantum computers in the world. Number two, we landed some of key customers as reference customers for the verticals that we focus on. And number three, we have continued to deliver on our technology and our product roadmap, which I'll talk more about. So there has been this evolution, steady evolution and transformation of the company from primarily thinking about R&D to becoming an operating business which now has revenue, growing revenue, significant backlog, and a competitive commercial product line which is delivering business value today so in some sense I am perhaps contradicting the narrative of the previous session which perhaps was hypothesizing that quantum is not real until late of this decade or early to early of next decade

that is not true for Pascal so that's the transformation we'll see so as you pointed out there's other modalities we've had a number of superconducting companies here today in this room as well as other photonic companies. Can you help the audience understand the case for neutral atom versus superconducting and what optical tweezers and room temperature operations can enable for cost and scale?

Okay, so at the heart of it, you know, think about it this way. First of all, you know, when all of us, when we should think about quantum computing, we should forget the word quantum for a second and think only about the fact that it's computing. In the end, this computing has to solve a problem at scale with reliability, with performance, and with the right unit economics. So the problem with a lot of the traditional approaches is that you have to manufacture a chip. In superconducting, ion-based, or whatever, you have to manufacture a device, and every device is a yield issue. Every device is slightly different than any other, and quantum states are very fragile. So you cannot expect the same performance, and there's a lot of issues that you run into. Now the neutral atom approach was a dark horse for the longest time. It almost sounds like science fiction today, but part of it what we do is we don't manufacture a chip. We use individual atoms of an element called rubidium, which is the heavier cousin of sodium, so you find rubidium wherever you find sodium or salt, and we use that as our qubit. Now what's good about it? Number one, every rubidium atom is the same as any other rubidium atom, so you don't have any manufacturing getters. Number two, you don't have to connect these atoms to each other. When you put them in close proximity to each other, they will spontaneously become connected to each other through electromagnetic interactions, and we help that along. And number three, this makes a system highly scalable. In fact, neutral atom systems are the most scalable systems. We have no visible issues in meeting 10,000 to 50,000 physical qubits per single QPU. And our systems work at room temperature. We do not require deep cryogenics, which Which means they can go into standard data centers. And for the physics nerds out there, the coherence times of these systems are very long. Which means they keep their quantum state, in the case of neutral atom systems, that's about 10 seconds. For a superconducting system it's about 1 microsecond, so we're 10,000 times longer in terms of coherence times. So all of these things conspire to create structurally a more robust system and structurally a different system. The analogy I would like to give here is that a lot of people they were working on creating machine learning systems that could do natural language processing and they had all kinds of different machine learning models until coincidentally in 2017 Google Research created the transformer model and Gen AI was born. So the heart of the matter is when you change your architecture fundamentally and you don't follow the same thing and you try to get diminishing returns from the same approach, then you can get nonlinear advantages from that. So neutral atom systems were dark horse systems for the longest time until all the impediments that people perceived they would have to make scalable quantum computers were robustly demonstrated and now they are considered a leading candidate for making scalable quantum computers.

Somewhat you called out there in terms of almost arbitrarily adding physical qubits, room temperature operations. I mean, one of the things that superconducting advocates consistently fall back on is, well, great, you're building a larger system, but the gate speeds remain slow. Is there just a simple way to answer that question for generalists?

There's actually a very naive mistake, in my view, that people make. The problem is this. Let's go back to this. From a custom point of view, what matters is how many how many logical operations you can do per second that's the thing we need to solve for that is a different problem than the inherent gate speed so the number of logical operations you can do per second it is dependent on your overall quantum error correction architecture so if I could just simplify it and just create a simple formula for it right your Your logical computation rate is equal to, in the numerator, you will have the number of logical qubits times the parallelism of those logical qubits, and the denominator, you would have the number of cycles, operations you have to do to do a particular error correction cycle times the time it takes to do that. So those are the four terms that go into it. So in the superconducting systems, you have an advantage on the time for each of the error correction cycles. But in all three other terms, neutral atom systems have an advantage. So let me talk about that for a second. So number one, because neutral atom systems are not linearly constrained like superconducting systems or other systems, it means that we can have a completely different, far more efficient quantum error correction. we don't need to use surface quantum error correction codes, use parity checking long distance or parity correction codes which are inherently much more efficient. So our number of logical qubits is higher. Our parallelism is much higher because we can do all these operations in parallel without doing them one at a time. So the numerator is much larger. And on the denominator side we have the time and we also have the number of steps you need to do, to do a logical error correction cycle. And one of the beautiful things about neutral atom systems is that what takes multiple two gate operations for superconducting systems, we can do it in one go. Because it's inherently a much more entangled, simpler, more elegant system. So once we think about it in this way, which is that if we are not constrained by using the same error correction that superconducting systems are forced to use because we are fundamentally different system, then this argument goes away. And people know this.

So maybe to round out the the competitive set, so within neutral atom number of players, some that have near-term commercial revenue, it could be selling components like RF sensors or atomic clocks, inflection, atom computing.

For you, you do have the analog mode which is a critical component that will be useful for for many many years but also useful today in a commercial setting but maybe just help us think about uh how your approach is different than other neutral atom providers okay it's a great question so uh let me give you a bit of the background for the company um you know we were we're one of the very very few companies that has an actual nobel laureate as a co-founder so professor lns pay he's uh he's the nobel laureate in quantum physics because he's the one who demonstrated quantum entanglement so he's the father of quantum information sciences Our other co-founder is the co-inventor of the Neutral Atom approach itself, Professor Antoine Beauvais. But there are two decisions that we made early on when we started the company. One was that we will not ship lab equipment, which meant that we will properly engineer our systems to be computers versus lab equipment. And the second was that we will focus on problems that the customers have instead of just building a hammer looking for a nail. So those two things became the hallmark for Pascal. And the third, of course, is we realized that quantum computing is actually the most important problem in quantum technologies in general. So we'll focus purely on quantum computing because it requires that hyperfocus. Those are the ways in which we are different than other companies, not only just in neutral atoms, but with other quantum computing companies in general. So the number of systems that we have out there deployed, they are deployed in standard data centers. They work at room temperature. They meet the SLAs of compute for those data centers. So they're well-engineered machines. The second is that these systems then, they not only meet those SLAs, but they are running workloads today which are delivering value for customers. And those two things are because of our decision not to make lab equipment but to make engineered systems. And that's our primary reason why we are different than a lot of quantum computing companies. And this, by the way, was not an easy road because this takes intentional effort to design a system from scratch, which for us meant that building our system in terms of different engineered system Lego blocks, in a sense, where these Lego blocks are replaceable and upgradable, which means that for our customers, we don't have the issue of obsolescence of the system. and we can do in situ upgrades so for us to add more qubits means to add more neutral atoms rubidium atoms into our vacuum chamber and you have more qubits voila and your control system gets upgraded and so on so forth so i want to i want to point out that i think analog is still

underappreciated so just curious what customers are using that for today how much evangelism you think is needed for that type of offering today and then the value, the complementary value of that with your gate mobile system and the software layer over a longer period of time.

It's a great question. First of all, for the audience, I mean maybe the only exposure people have to analog quantum computing is through quantum annealing which is a small subset of overall analog quantum computing. Analog quantum computing is how nature does quantum computing. We're all living quantum computers. Every cell, every atom and a set of molecules in our body is a quantum computer in a sense. And it's not gate-based. So that way of doing quantum computing is exactly at par with doing gate-based. There are pros and cons to each. Customers, they actually don't care whether we use analog or fault-tolerant. Customers just ask the question, can you solve this problem? And can you do it with the right economics? Can you do it with the right ROI for us? And that's what we focus on. Now, the advantage of doing full analog quantum computing, we believe that we are the most advanced company in the world when it comes to analog quantum computing because we have explored it the furthest, is that we can do things today that people feel they could only do during FTQC regime. One of those examples is material simulation. We recently published a result with Los Alamos National Lab in which we were able to show scientifically rigorous quantum advantage in the modeling of TMGO which is a well-known material we wrote this paper in Los Alamos and the way we did that was we were able to create a digital twin of TMGO in our quantum computers and we were able to explain the magnetic properties of that material which people could not explain before and we were able to predict new properties which are not in the classical regime they're purely in the quantum regime, and they were experimentally verified. So that's an example of a problem that we solved today using full analog quantum computing that people otherwise think can only be done in FTQC regime. Similarly, our big customer, our big other vertical is oil and gas with Saudi Aramco is our big key customer over there, and they use it for upstream business primarily. In the upstream business, the major problems are reservoir modeling. It's various kinds of optimization problems for well placement and demurrage and other important problems. And in that case as well, we are able to use our full analog quantum computing capabilities to go solve those problems. The other vertical for us is financial services, where Credit Agricole is our reference customer. And two months ago, Credit Agricole did a press release. They're based on the two years of work that we've done with them. They shared that they were going to take that work, which is in the area of risk portfolio and risk-weighted assets optimization, and they were going to put it in production. I believe that's the first major bank that has made that commitment. So again, it's going into beyond simple optimization into full-scale modeling and being able to solve those problems today and being able to show that differentiation between the current performance using non-quantum or classical systems versus what we can offer today.

And when you sell a system into a customer like that could you help the audience understand price point, like revenue delivery, installation effort, maybe ongoing support that's associated with that and then to your point you know you could upgrade those systems by filling more atoms into the vacuum chamber over time so it could create an upgrade cycle.

Exactly so we haven't publicly disclosed our ASP so if you off of it. Yeah, we can. But there's two ways in which we sell our quantum computers. I mean, I think most of the room is probably aware of what the price ranges are for QPUs on the low end and the high end. So we are very competitive, both on price and on performance. But our customers typically use our systems in two ways. First of all, our systems are not just the computer. It's the software stack on top of it as well. So you get the full stack. So typically, sovereign customers, they buy the computer and the stack. So there's what they pay for the acquisition of the computer, plus there's an O&M contract, which is a typically five-year length, and that is a recurring revenue stream, plus any upgrades, which is extra revenue streams from that. So that's one cycle. And the other one is typically for enterprise customers, they utilize our quantum compute through cloud access, and that is paid on a per-hourly basis. So per-hour access for the QPUs, again, they get the full software stack as well. When you buy a computer, for us, the installation cycle and the commissioning cycle is usually quite quick. And I use the word quick in comparison to some of the other data points that I have. But typically, we have engineered our systems in such a way that we have these crates. And there's multiple actually photographs on our website that are available for the multiple deployments that we've done. The crates will show up, you assemble it, it's commissioned, and it's up and running. And for the cloud side, it's actually pretty simple. Anybody in the room, you can set up your cloud account within 15 minutes, go to Pascal web page, and you can start playing with our quantum computers right away and start running problems. So we have removed the barrier to entry for that quite a bit, and we have worked very hard on that as well to remove any sort of hard coding requirement or even not only just being a coder but of course not even being a physicist so we have removed all those barriers to entry so this becomes literally a computer resource that you can run workloads on.

You mentioned placing these these computers and data centers and two sessions ago we had another company working on optical interconnect and data centers but yeah the question being in NVIDIA Jensen originally was quite critical of quantum. He's changed his tone. Eventually, we're going to be routing workloads between a CPU, a GPU, and a QPU. Where do you think Pascal's place is in that architecture? And then maybe from a partnership perspective with NVIDIA as well, where that fits in.

Great question. So I think one important thing perhaps to point out is that it's on a public roadmap that we have committed that we will deliver more than 10,000 physical qubits per single QPU by 2028. For us, that is a rubinium atom array of 100 by 100 atoms. And that is something, if you look at that kind of scalability, and of course it goes from 10,000 to 50,000, that's all on our public roadmap. When we look at that kind of scalability, for us, the need to connect various quantum computers is not so overwhelming as it might be for others because they don't have that kind of scalability. So our vertical scalability is completely different than other systems which have to rely on horizontal scalability. Having said that, one of the great things about our approach is that we can use optical interconnects basically on well-known wavelengths to connect qpus as well to share entanglement across different qpus so horizontal scalability after that is also completely open to us we are one of the founding members of nvq link initiative with nvidia so that's one of the things that we do with them but there's a lot of other things that we do so our relationship with nvidia revolves a lot around the intersection between qpus and GPUs. And the intersection goes in both directions. In one direction, QPUs can generate unique quantum data that can be coupled with GPU-driven AI models to solve specific problems. And that is something that only our quantum processes can do. And the other side is where GPUs can work with us, and they can help accelerate the quantum error correction side. So it goes both ways. So that's part of the work that we do with that.

I have a final, but I want to make sure the audience has a chance to ask a question if they'd like. So I do want to touch on... Oh, yeah, go ahead.

That's such a great question. I think because a lot of people who represent quantum computing companies, they emphasize quantum. They come from the world of science. They don't come from the world of commercial, of compute at scale. So they never talk in that language. But in the end, we all know compute is compute. For us, because we are commercially in a different stage than a lot of other companies, we are post-revenue of significant backlog. When I go and talk to customers, they never ask, what is your modality? What is analog at FTQC? They couldn't care less. Here's my problem. This is the limits of what I can solve. Can you solve it? And the second is, if you can solve it, what do the unit economics look like? And the third is, can it go into my workflow? What do I need to do? What do I need to give you so that you can solve the problem for me? And this is something that we hope that we bring as an awareness to the market, that there is a different way of thinking about quantum compute, which is it is a compute, which is the third pillar of compute. In the end, it's compute. I don't know if I answered your question, but that's my take on it.

We have more, but we don't have too much time left. So is there anything that you'd like to leave the audience with today?

I think I would actually build on that question because I hope that the audience and the investors, they actually challenge quantum computing companies to be computing companies. And I think there's a few just thoughts that I have. I think this whole debate about what modality you have or whether it's analog or FTQC, this is a temporary debate. In the end, it's all about compute and can you solve a problem? For us, the way we approach this is our approach to solving business problems is that we don't boil the ocean. What we did many years ago was we said, what are the specific mathematical problems we are extraordinarily good at solving? And we map those mathematical problems to use cases and key verticals because that's what we need to solve. And I think that's probably the way the industry needs to think about this as well. Instead of just thinking I have a cool new hammer and I'm looking for a nail that I just need to go figure out, you know, what to solve this for. And I think that's what that's the next stage for for for the sector.

I think it's been inspiring. Thank you so much. Thank you so much for joining us today.