Kike Miralles, investment director at Intel Capital, is interviewd by Yuval Boger. They speak about corporate venture investing in quantum technologies. Kike explains Intel Capital’s focus on quantum hardware and middleware, compares leading modalities, and outlines emerging themes like QPU scale-out networking, hybrid classical–quantum error correction, and the growing role of logical qubits as a core metric. He also discusses check sizes and stages, government-heavy revenue, PR and hype in quantum, expected consolidation among hardware players, and what differentiates strong quantum investors and startups in today’s market.
Transcript
Yuval Boger: Hello, Kike, and thank you for joining me today.
Kike Miralles: Hello, Yuval. Great to be here.
Yuval: So who are you and what do you do?
Kike: Well, so I’m Kike Miralles. I’m an investment director at Intel Capital, the corporate venture fund of Intel. And I’m a generalist deep tech investor within our team. And so I run most of our quantum activity, but I also invest in other areas like semiconductors, photonics, power electronics, robotics and the like.
Yuval: So let’s start with quantum. What quantum investments have you made until now?
Kike: So within Intel Capital, we’ve made one investment into an Israeli company called Quantum Machines. That was a Series C investment we made earlier this year. And then at my prior firm, I invested in a company called IFOS that makes photonic chips for quantum and classical use cases.
Yuval: Every VC firm has an investment thesis. So what can you share with us about the investment thesis of Intel Capital as it relates to quantum and adjacent technologies.
Kike: Absolutely. So I would say, look, it’s pretty clear that quantum technologies are going to, like, it’s one of the biggest things to happen in computing technology, and I think it’s going to be one of the most promising areas of opportunity of the next decade or two. Of course, that progress is going to take time, as, you know, as we’ve seen, it’s not so easy to, you know, control these systems and get them to behave the way you want to. And so today I would say we are more focused on the middle layers of the stack as well as the quantum hardware itself and a little bit less focused on what you would call the application layer of the stack. So a lot of hardware and some of the middleware that we do think is going to be essential to kind of unlock the hardware capabilities.
Yuval: There are many types of quantum modalities. They have different requirements. Some, for instance, require cryogenic cooling. Others require single photon sources. Others require fast readout. Do you have a favorite or a set of favorite modalities that you believe in?
Kike: I wouldn’t call favorites. I feel like we’re at a point in the, let’s call it the race, where it’s still difficult to call potential winners. I think the way I like to think about it is, I don’t know if this is going to be controversial, but I personally consider photonics a little bit of a wild card in the sense that I could believe that they figure it out faster than everyone else, but I could also believe that they’re going to be stuck in thinking about losses and sources and resources for a very long time. So I generally think a lot about the atomic versus solid state question. And I would say the atomic platforms, like both neutral atoms and maybe ions as well, have an early scaling advantage that in theory or on paper could be superseded by the solid state platforms because they have a potentially much faster clock speed. But I would say the solid-state platforms have a manufacturing problem that clearly the atomic platforms don’t have. And so it’s also a big when question for me on the solid-state side. Like how long is it going to take for these platforms to figure out, you know, not just, you know, in theory, we can fit a million of these things in a wafer, but actually fit a million of these things in a wafer.
Yuval: I think Intel has their own quantum program. Are the investments from Intel Capital in any way related to the way Intel is approaching quantum?
Kike: So I would say more generally, like, ultimately as a CVC, we aim to help Intel across many regards. Part of it is providing exposure to maybe things that we’re not doing directly, providing exposure to, for example, companies with which we are collaborating. And this is not just in quantum, but more generally across all of our venture activity, as well as just put simply investing in things that we believe are essential for the future of compute and that we want to be an active actor in bringing that future forward. So what I would say is it is unlikely that I, as a corporate venture capitalist from Intel, I would do a spin-qubit company. That is very much the territory of our business unit. But we are not opposed to, for example, diversifying the hardware exposure of the parent company with other modalities over time.
Yuval: What does it take to be an investor in quantum? Do you need a PhD in physics? Is it more important to have an MBA education? I mean, I’m curious about your background and in general how you think about the right profile for an investor in quantum.
Kike: Well, to a certain extent, I would say it depends a little bit on the stage. I think the earlier you go, the more important I think is to have a somewhat technical profile. I don’t think you necessarily need to be a quantum physicist yourself. I am a quantum physicist by training, so I’m not trying to defend myself here. But at the same time, I do think that you at least need to be able to understand what the right questions to ask are. I think you need to be able to have a good rapport with a technical founding team and with an advisor or a set of advisors that would help you, you know, ask the difficult questions from that team. But I would definitely say that quantum is an area where I think it’s difficult to make good decisions if you approach it with the kind of classical venture sort of only-the-team-matters sort of approach. I mean, and for sure the team matters a lot. Don’t get me wrong. But there’s a lot that has to go right from a technical perspective, and so the tech stack matters as well.
Yuval: What are the trends that you’re seeing in the market? I mean, what are things perhaps that you would not have invested two years ago or not considered investing two years ago and now are more interesting to you?
Kike: So I think one of the more recent developments that I am very curious about is the scale-up versus scale-out question. I think two or three years ago, the question of are we going to need to network several quantum processing units to keep making progress was, I think, a bit of a silly question. Hopefully the founders working on networking are not going to get mad at me, but we’re still pretty far from a QPU perspective to really think, yeah, we’re going to be also stitching together these things. But as progress accelerates and we see more and more companies make multi-hundred qubit processing units, I wouldn’t be surprised if on the algorithm side we start to figure out ways to kind of break up algorithms and allow for better stitching and for multi-QPU setups to open, you know, doors for progress while we keep scaling up the size of a single QPU. So I would say scale out is definitely one of the areas that for me has appeared a little bit over the past two to three years. I would also say the transition from error mitigation towards like real error correction, different kinds of encoding, and so on has been a pretty strong one. I feel like a few years ago we were still very much in the noisy intermediate era, and we’re seeing much more activity now on… There’s a lot that you need to do well to implement error correction and achieve fault tolerance in real time. It’s not just… I mean, and I say just as if it was easy, but it’s not just having very high quality gates. But like you really need real-time feedback mechanisms that allow you to act on the things that are happening. You need very fast, very strong, I think, hybrid compute paradigm, such that a classical processor can really kind of act in real time and allow you to fix whatever it needs to be fixed, and a control system that can also react to that fast enough. There’s a lot that goes into making error correction a reality. And I think we’re firmly entering the era where error suppression and mitigation will remain useful, but the next 10 years are going to be defined by doing error correction better than others, I think.
Yuval: I should have asked this earlier. I’m sure there are a few quantum entrepreneurs who are listening to this episode and they’re wondering, how much money do you guys invest in? What stage is your typical investment?
Kike: So I would say our sweet spot is typically the Series A and Series B. We’re an early stage fund, technically multi-stage and we can pretty much go up until pre-IPO, but I would say, you know, our bread and butter is the Series A and Series B. We invest, I would say, check sizes vary, but I would say anywhere in the 5 to 15 range would be pretty normal for us. And then when it comes to, I think quantum specifically, I think we’re a little bit more flexible when it comes to stage because the technology is still pretty nascent. And so we do think that it’s not, for example, just because you’re doing a growth, like a Series C or a Series D, it is not necessarily growth equity, but it can still have a venture profile. And so we’re a little bit more flexible than usual, I would say, when looking at quantum.
Yuval: Quantum is often mentioned together with AI and also together with HPCs. Do you see investment opportunities in the intersection between quantum and AI and then quantum and classical?
Kike: I think in the quantum and classical for sure, ultimately for me it’s a little bit difficult to imagine a world where even when we achieve this real commercial relevance and commercial value, I think quantum computers are very much going to be part of the high performance compute stack. You’re always going to have a hybrid approach to whatever problem it is that you’re solving. Like even when you think about something like Shor’s algorithm, like the vast majority of the computation is done in a classical processor. You just need the quantum one for the very difficult step in the middle. And so I do think that there’s going to be investment opportunities in that kind of hybrid space. It also goes back to what I’ve mentioned around all of the classical processing that’s needed to support error correction. And then in the quantum and AI space, I would say it depends. I think AI and quantum are a little bit like frenemies, you could say. I think some AI capabilities could make it a little bit more difficult for quantum to create value in certain applications by kind of extending the range of where our classical compute gets to go, especially in certain materials and chemistry-related applications. But there’s many other areas where, you know, if we had access to quantum computers and, you know, one of the things that they do really well is finding structure in very large data sets. I would be very surprised if, you know, we cannot keep pushing AI forward with better quantum capabilities. That said, one thing that I repeat quite often is when you get asked the question of where will quantum create value? I feel like there’s a little bit of unknown unknowns in that answer because the areas where we know it is going to create value, it’s because we have a very good sense mathematically that the computers are going to be good at that. Either you know for a fact or you have a very good suspicion that these computers can help there. But at the same time, I’m also like once we have machines available, I’m pretty sure that the research and algorithm community is going to figure out a million more ways to have these computers be useful and create value in areas that today, I don’t think anyone would say, for sure, quantum is going to help there.
Yuval: So you say the inventors of the internet didn’t think it would be used for cat pictures. That’s sort of the analogy. You mentioned earlier that one of the topics that you find interesting and that you focus investments on is one of middleware. And in my experience, when I say middleware people hear different things. Are you referring to compilers? Are you referring to bridges between classical and quantum? What’s the definition of the middleware that you care about?
Kike: So I would include pretty much all of that in that definition. So the way we typically break the stack is kind of the infrastructure of vacuum, cryo, et cetera, the quantum processing layer, and I would put the networking and memories, et cetera, there as well, a control and interface, readout, et cetera, layer, middleware, and then applications. So I think about middleware and I think it’s sort of the most nebulous layer because if you think about companies, I don’t know, like companies like Classiq also kind of have clearly moved a little bit up towards the application layer and offer algorithms within their own platform. It’s not just an IDE necessarily or not just an SDK. Similarly, you’re also seeing middleware players where, for example, where do you place error correction? Is it in the interface layer because you need a decoder? Is it more in the middleware layer because there’s a pretty heavy software load to get this done? So for me, middleware is pretty much anything in software that does not necessarily go to an end user directly. Perhaps that’s a very broad definition, but that’s how I think about it.
Yuval: If you were looking at a company outside quantum, and they told you that their largest customer is going to be the government for the next couple of years, that might scare you. Doesn’t it scare you in quantum?
Kike: So I think when you think about quantum, sometimes it’s useful to draw parallels to the space industry where I’m not going to talk about defense because it’s a much more obvious case, but I think even within space, like the US government remains one of the biggest sources of revenue for pretty much everyone and I don’t necessarily think that that’s a bad thing. I think the question to me when it comes to government revenue is also related to how repeatable is it? Like how many of these things is the government going to buy over time? Will they scale up their purchases? Will they always want to have the latest thing you built? Or are they going to skip generations? Are they going to falter if someone else takes the lead? And so I think it’s more related. It’s less that it’s a government customer and more that the commercial use case is still difficult. And so your buyer is likely a bit more fickle than a typical commercial buyer would be.
Yuval: Where do you think we are on the path from research to commercialization? How much of the work, in your opinion, is science versus engineering?
Kike: I would say for the leading modalities, so like trapped ions, neutral atoms, superconducting, etc. I think we are relatively firmly in the engineering territory. There are some platforms where I think there are still physics questions of what things look like when you try to do things like the Majorana topological qubits and things like that. But I think that makes it sound easier than it is because ultimately these quantum systems really don’t want to do what we want them to do. So the engineering challenge is pretty steep. I think the biggest question in my mind is we’ve been making pretty good progress when it comes to qubit counts, particularly over the past year or two, I would say, and I think, and I expect this trend to continue, I don’t think we’re going to get stuck at this level immediately. But one thing that still worries me a little in terms of that progress is circuit depth. How fast are we going to be able to go from hundreds of gates to millions of gates once we achieve a QPU size that starts to be really relevant for real problems? And there, I think I don’t have a really good sense of, you know, is that tomorrow or is it going to take a little bit longer?
Yuval: How do you feel about PR in quantum? Some companies are out there saying, give us your problem and we’ll solve it today, and others are saying, no, quantum computers are a toy, you know, come back in seven years and we’ll have something fantastic for you. If you were on a board of a quantum company, how aggressive would you advise them to be in the way they present their maturity level?
Kike: I would be a terrible, I think I would be a terrible founder. The challenge is, you know, especially as a technical VC, sometimes when I hear a pitch and the founders don’t adjust it, knowing that I understand what they’re talking about, it can be a little bit difficult to take them seriously because I’m like, if I ask you two or three questions, the answers are going to immediately tell me that what you just said makes no sense. But at the same time, it’s a difficult balance because the typical investor is not necessarily technical. And if you were to tell them, well, you know, we have 100 qubits, but you’re going to need at least three or four thousand logical qubits, and then a billion gates and whatever it is. Like I think you can also lose track of the message and the narrative that you’re trying to explain to an investor. So it’s a pretty tricky balance. And that’s why I say I don’t think I would be a good founder at it. But I would definitely say, I think it’s fine to sell your story, but be careful about what you’re trying to stretch. I think it’s one thing to say we are near or we’re starting to be near that commercial relevance. It’s another thing to say, I can solve today a problem that is relevant to someone outside of a physics lab, and that cannot be done with a classical computer.
Yuval: Do you think there are too many quantum companies? I mean, superconducting, for instance, I’m sure there are 20 different vendors. How’s that going to play out?
Kike: So I don’t know if too many is the right way to characterize it. I do wonder a little bit about, given how much money it takes to make real progress and how much money it takes to, you know, keep these efforts alive for the time they need to, like, really make meaningful milestones. I do sometimes wonder whether we would be better off with a more consolidated ecosystem, particularly in areas where sometimes a technical differentiation is not super clear at first glance. But of course everyone has to defend their own approach. And so I do think that we will see some of that consolidation happen over the next few years because ultimately there are approaches and there are architectures that are better than others. And I think, you know, you can get away with a certain level of funding, but I think ultimately the market, you know, this is the weighing machine versus voting machine; like ultimately better approaches and approaches that make more progress with less capital will eventually get the recognition they deserve.
Yuval: As we get close to the end of our conversation, I’m curious how your perception changed. In other words, what have you learned in the last 12 or 18 months about this market that surprised you or that you were not expecting?
Kike: I think the biggest thing that has been like a real surprise for me has been how quickly the industry has shifted to the logical qubit concept. I don’t think in 2023 pretty much anyone—I don’t want to say it as a fact—had demonstrated logical qubit operation. We’re in late ’25, pretty much everyone is now talking about the logical qubit count in their systems. And we are seeing a lot of progress in this arena. And so for me, it’s been the first time that I’ve actually studied in-depth encodings for error correction and really tried to understand, you know, not just the basics of, oh yeah, this is surface versus something else, but like really trying to understand what are the trade-offs depending on the encoding that you’re trying to choose, and what are the trade-offs that certain modalities have to make because they don’t have access to every encoding possible.
Yuval: And last, a hypothetical, if you could have dinner with one of the quantum greats, dead or alive, who would that be?
Kike: That’s a difficult one. Has to be alive?
Yuval: Dead or alive, but not both, not both at the same time.
Kike: I think I would, it would probably be, I mean, this is going to sound very cliche for any physicists listening to this, but it would probably have to be Feynman. It’s just such a character and such a personality and someone that contributed so much to, you know, our understanding of quantum physics. Also, pretty much the first, if not one of the first, that ever proposed quantum computing as a concept. So it would, you know, I feel like that’s an easy one, if a little bit of a cliche, but an easy one.
Yuval: Kike, thank you so much for spending some time with me today.
Kike: Thank you so much. It’s been an absolute pleasure.