Next in Tech | Ep. 276: Quantum CEO Series: Andy Ory

View Full Transcript

[00:00:00] Eric Hanselman: Welcome to Next in Tech, an S&P Global podcast where the world of emerging tech lives. I'm your host, Eric Hanselman, chief analyst for industry research at S&P Global. And today we have the next in our Quantum CEO series with QuEra. Our lead analyst, Ellie Brown, is back to lead the discussion. Ellie, welcome back.

[00:00:20] Ellie Brown: Thanks for having me, Eric.

[00:00:22] Eric Hanselman: And as our guest today, we have Andy Ory, the CEO of QuEra, with us to talk about more types of quantum computing and what they're pursuing. Andy, welcome to the podcast.

[00:00:32] Andy Ory: Well, thank you. Hi, Eric. Hi, Ellie. I'm delighted to be here.

[00:00:36] Eric Hanselman: It's great to have you on. We actually had a chance to have a quick conversation at the AWS Summit in New York a few weeks ago. But I wanted to dig into some of the specifics about really next stages in quantum technology, all the challenges we see in, in realizing some of that potential, talking about stability, error correction, and the challenges that are there.

[00:00:58] Eric Hanselman: I guess to begin with, Ellie, can you dig us into some of the, the thoughts around technologies that we're looking at here?

[00:01:04] Ellie Brown: Yeah, certainly. So, I mean, obviously this is, I think we're on the fourth, you had mentioned, in the Quantum series, right? So, we've covered quantum computers as a whole from a lot of different perspectives, right? Usually, when someone is new to quantum computing, they think all quantum is the same, right?

[00:01:21] Ellie Brown: Every quantum computer is this big chandelier hanging from the ceiling. It's very steampunk, it's very cool. Cool literally because it has to be super cooled, right? But obviously there's lots of different ways to build a quantum computer, and this is the kind of latest in our lineup that we wanted to focus on, which is this idea of neutral atom quantum computing.

[00:01:44] Ellie Brown: So obviously I'm not the expert. I'd love Andy to, you know, give us an introduction to neutral atom computing. So, for listeners who may not know the term, or may know the term but not the mechanics, what is neutral atom quantum computing?

[00:01:59] Andy Ory: Ah, well, thank you, Ellie. I might step back and just say one thing, that a quantum computer is not a better, faster computer. It's not a computer in the traditional sense. It's a totally different instrument. It was first postulated by Richard Feynman in the early '80s, where he felt that you could solve problems in the natural world if you could use the physics of the natural world to try and perform computations.

[00:02:25] Andy Ory: And so, quantum computers quantize bits of matter. They entangle them, they superposition them, and then we try and control the interactions along the lines of an algorithm to create, some sort of a solution to a problem that we try to feed into this machine. The thing that all quantum computers share is that quantized bits of information have to be very cold, almost at absolute zero.

[00:02:49] Andy Ory: So, the first question that people had to answer is, "How do I get there?" And there are several different ways. One of the obvious ways is to take a computer chip and stick it in a bath that's at 15 millikelvin, and you see those chandeliers, and you connect wires to each one of those little qubits and now you have really cold units of matter that you can quantize and try and perform calculations with.

[00:03:11] Andy Ory: Neutral atoms are different. What we do is we take rubidium atoms. There are many different species of atoms you could take. We take these rubidium atoms, we put them in a vacuum, much, much, much less material than in space. Several orders of magnitude less material. So, it's a true vacuum. And we use lasers to trap them, and then we arrest their motion.

[00:03:32] Andy Ory: And for those that forgot their high school physics and chemistry, energy and heat is a function of motion. So, if you take motion away from something at the molecular level, it has no heat. So, by arresting these atoms, we super cool them. And so neutral atom machines have a vacuum chamber in the middle of a bunch of laser traps that sit on something the size of a ping pong table, in room temperature in a normal room.

[00:03:58] Andy Ory: So, it's very different. The other thing is that you get errors when you try and create qubits, on manufactured silicon. You get fewer errors with atoms. Atoms themselves are perfect every single time. Every atom is the same as every other atom. It's a big manufacturing challenge when you're trying to make the exact same thing at the nanomolecular level.

[00:04:20] Andy Ory: And so, we happen to really like neutral atoms. We think that it's easier to have perfect qubits. It turns out that it's not just the operational characteristics at room temperature that are a huge advantage and cost a lot less, but the notion of error correction, the techniques in which we try and take errors out of these qubit calculations seems to scale much more efficiently with neutral atom machines.

[00:04:46] Andy Ory: And that's actually really important because when I first started my quantum journey, it was all about the number of qubits. Turns out that you can't have a qubit that's more than 99.9% error-free. 1 out of 1,000 times calculations, it's going to make an error. And that's fine if you're doing six or seven calculations, but if you're doing a billion or you're doing a trillion, an error of 1 out of 1,000 tells you that you're going to get garbage.

[00:05:13] Andy Ory: And so, people figured out that they needed to take groups of physical qubits to create one logical calculation on that logical qubit, and that way they could check after each calculation, is there an error? And if so, they could fix it without tearing down its quantized state.

[00:05:32] Andy Ory: And so, fault tolerance in quantum computing is a requirement to get scalable, useful quantum computing. We think that neutral atoms are the best way to get there

[00:05:43] Eric Hanselman: There's been a lot of discussion about coherence times and the rest of that. Ellie, some of the research that you've been doing on this, we've seen great improvements in some of the, the superconducting environment. But yeah, that's still a challenge.

[00:05:56] Ellie Brown: I mean, the difference is you're working on several different challenges, with quantum computers. You have scaling, which is scaling the physical systems, and then you have that error correction. And sometimes what works well with scaling doesn't lend itself to better error correction and vice versa.

[00:06:13] Ellie Brown: But Andy, again, you probably have plenty of perspective from your angle there.

[00:06:18] Andy Ory: Yeah, I mean, you know, so these different approaches they refer to as modalities or different modal approaches. And I think that the people that like superconducting like it for a few reasons, one of which is it looks like a computer. You can look at Google's Willow chip, it's made in a foundry, it looks like a computer, and I think that makes people feel better because they're trying to build a quantum computer, and they're trying to copy as much of general computing as possible to get to a quantum computer.

[00:06:47] Andy Ory: And I understand that. There is also ease of manufacturing. It's a lot easier to make computer chips than it is to make neutral atom vacuum-based systems with free space laser optics. There’s no question about it. But the downside is that these superconducting systems have to be in these baths, these 15 millikelvin refrigerants that have limit in terms of size.

[00:07:08] Andy Ory: You've got to network them together, and they're very expensive. And so that's one issue. And another issue is that it's really hard when you create groups of qubits to create one logical qubit, to have those qubits as a group interact freely with a bunch of other qubits when they're stuck on a chip.

[00:07:26] Andy Ory: So, you really can only communicate with your adjacent qubit. Whereas with free space optics and neutral atoms, we can use lasers to move thousands of atoms simultaneously to meet other atoms simultaneously anywhere in the matrix. And so, this anywhere-to-anywhere connectivity, is hugely advantageous.

[00:07:46] Andy Ory: And so, I think that we're going to see both modalities coexist for quite some time, but I am quite confident that we're going to see fault-tolerant quantum computing and utility-scale quantum computing emerge first and scale first with neutral atoms. And I think that we're going to see that the neutral atom approach is going to make a lot of sense for things that exist in the natural world, so material and chemical science simulations, interactions, life science simulations, biosensing, sensing in general.

[00:08:20] Andy Ory: I think we're going to see neutral atom quantum computers far exceed what superconducting quantum computers can do for a long time.

[00:08:28] Eric Hanselman: The specifics of quantum computing types are important, but before we get too far down the road, I want to take a step back and really talk about what's fueling this boom in quantum computing.

[00:08:39] Ellie Brown: Why quantum? Why all of the hype? Why all of the excitement? What makes quantum such a simultaneously captured part of the human experience? Why is everyone so excited about this, Andy?

[00:08:51] Andy Ory: It's a great question, Ellie. I've been trying to figure it out. I got involved in quantum back in 2018, helping Harvard create the world's first PhD, or the country's first PhD for quantum computing. And I thought it was a really neat scientific area. And as I got involved, I just saw more and more hype, and now we see a tremendous amount of societal and investor hype.

[00:09:15] Andy Ory: And as best as I can tell, it's responding to this dynamic that's emerged, which is the world is filled with tools that make better tools, and the world is hyper networked so that when a change happens, it can precipitate to everyone very quickly. And lastly, as a society, we are used to taking up new things more quickly.

[00:09:41] Andy Ory: When I started my tech journey 40 years ago, it was very hard to get people to do anything new and different. And today, the people that run our economy are just, they grew up doing new and different things, and they want more, better, faster, quicker. And so, all of this is contributing to a belief that first mover advantage matters more now than ever.

[00:10:04] Andy Ory: And so, reaching into the funnel of scientific discovery for things that can create these tools that precipitate all of these changes is almost bordering on a mania. And I can't say it's wrong. We could see what happened with AI in November of 2022 when OpenAI released ChatGPT. There were 100 million downloads in the first week.

[00:10:28] Andy Ory: We've never seen anything like that. So that's the context in which something like quantum is being born. And so, The world is almost hyper prepared for this kind of change, and this is a very fundamental change. It's like we as a society, as a species, have been locked in a room and have examined the room around us, and when we start building these real large quantum computers at scale, it's as if we're opening the door and we can go out and explore the rest of the universe.

[00:11:01] Andy Ory: And I think that fundamental prospect, coupled to the economic dynamics and the societal dynamics of change, is what's causing this mania, this hype. And again, I don't think it's wrong, but it is unsettling the degree to which everyone's leaning in.

[00:11:21] Eric Hanselman: I think it gets back to Feynman's comment that you were invoking earlier. This is something that is an entirely new tool with which to understand the cosmos. Not exactly insignificant

[00:11:33] Andy Ory: No, it's actually going to be hugely impactful and, I appreciate you all taking the time to try and get folks exposed to this because quantum's going to happen and we as a society need to be ready for it.

[00:11:45] Eric Hanselman: As you said, we've got to get that pipeline filled, so a lot more to do on that front. I'm curious in terms of some of the broader ecosystem that's needed to support these kind of capabilities. We've talked about, some of the software stack that's starting to support this, with some of the players that are starting to produce, quantum capabilities.

[00:12:05] Eric Hanselman: I mean, back in the day, we had the early frameworks, very limited functionality that came out with the IBM Qiskit, way back in the day. But we're now getting to higher levels of sophistication. I'm curious what you see as some of the necessary pieces, to be able to have in place to be able to make quantum as easily applicable and consumable as a lot of the other software tooling that we've got out there today.

[00:12:29] Andy Ory: Yeah, I think at the base layer, we need trained quantum scientists. That’s a non-negotiable, and there just aren't enough in the world to deal with this exploding field. And educating these folks is really important. Now, we can also do a lot of on-the-job training.

[00:12:46] Andy Ory: We certainly do that at QuEra, and I'm sure that all the other quantum computing companies are doing that as well. You do need these machines. When you think about quantum computing and you think about the companies involved in quantum computing, you're thinking about a very small group, less than 1% of the number that are really out there, and they all tend to have one thing in common: They tend to be full-stack quantum computing companies, meaning they're making everything: the hardware, the software, the operating system.

[00:13:11] Andy Ory: They're helping to co-design the applications, et cetera. I think to get to the first fault-tolerant quantum computers, having full-stack leaders is going to be really important. But for this field to really grow, we need to open up various layers of the stack to partners because we need to harness the collective.

[00:13:30] Andy Ory: For example, opening up your compiler, starting to do co-design with companies that are focusing on particular algorithmic approaches is really important. We also need to engage at a lower layer, at the hardware layer, all the quantum computing companies do, with their suppliers to make sure that we have a good, clean, safe supply chain, and that we can strategically invest in the technological leaps that need to happen at the supplier level for us to actually get to where we need to be with quantum computing.

[00:14:01] Eric Hanselman: Well, the supply chain is an interesting angle on that because of course, I think there's this general market perception that yeah, quantum's out there in the future. It’s still yet to really arrive, although we're starting to see such acceleration in terms of development.

[00:14:18] Eric Hanselman: There's a lot that's going on there, and Ellie, you've been covering this. We talk about Q day, that especially on my side of the world and security side of things, the point at which we've got quantum computers that can actually start to break, modern encryption technologies.

[00:14:31] Eric Hanselman: Back in the day we were talking 2035. Now maybe it's rolled into 2030. What’s your take, Ellie, in terms of where we're headed? Are we rolling any closer?

[00:14:39] Ellie Brown: Well, from what we've been seeing, just looking at roadmaps, Q-Air is included, the timeline keeps shrinking, and it's shrinking very rapidly. You know, as early as last year, the consensus was as early as 2030 for some of these fault-tolerant machines. But we have roadmaps now that are saying 2028 for the initial fault-tolerant quantum computers, right?

[00:15:01] Ellie Brown: And so, in order to prepare for kind of that acceleration in roadmaps, we're seeing a lot of add-on effects, in the security space with timelines saying, "Oh, we should swap stuff over before, 2030 hits to make sure that we're prepared." So that timeline then shrinks as well. So obviously, there's a difference, too, between the initial rollout of a fault-tolerant capable quantum computer and widespread rollout, which kind of gets into some of that, supply chain, work.

[00:15:32] Ellie Brown: You need to have a supply chain that supports that expanded rollout, which isn't exactly always the case, right? I think about superconducting qubits and their dependence on helium-3 for supercooling, right? That's a huge supply chain bottleneck, so actually I'd love, on Andy's side, is there anything on the neutral atom space in that supply chain that might cause a bottleneck for you guys?

[00:15:58] Andy Ory: Well, we don't use helium-3. We don't have any of the refrigerant requirements, so happy about that. Probably the single most important element is lasers and specialized laser power. We are dependent upon, lasers. We do have great relationships with our suppliers. We do have advanced laser and optic teams that are working to pioneer the power, the control, and the technology that we need coming online in 2030.

[00:16:26] Andy Ory: Nut I think from a supply chain point of view, we should be good.

[00:16:30] Ellie Brown: That's great news, right? That's not always the case when you talk to some of the other companies who are trying to backfill the supply chain a little bit, right?

[00:16:39] Andy Ory: Part of being the CEO is you have to deal with the challenges, and every company has them. So thankfully, that's not one that we're really focused on or too worried about.

[00:16:48] Eric Hanselman: So, in terms of the overall landscape of quantum capabilities, where does QuEra fit and what kinds of capabilities are you looking towards, and what are you seeing as some of those advantages?

[00:17:01] Andy Ory: Sure. Every business has its own heritage and its own culture, and Quera is a translational business with foundational science coming out of Harvard and MIT, and we are large investors in those labs. We continue to sponsor research, and we collaborate deeply as we translate that science and pioneer it over at Quera.

[00:17:22] Andy Ory: Being science first doesn't mean that we're not interested in commercialization, but it just means that commercialization is also part of our scientific effort to understand how to make, how to deploy, and how to scale these systems. I often say that we have a bike ride advantage. We really can't be more than a bike ride from these labs.

[00:17:43] Andy Ory: We've been very fortunate to have access to both science as well as capital, and that's allowed us to continue to leapfrog what our systems can do. And I would say about a year and a half ago it became increasingly obvious to us that we could get to a true fault-tolerant quantum computer running at scale by 2028, and, we decided that was the most important thing that we could do as a company and for our shareholders and for society.

[00:18:15] Andy Ory: We then thought about what would be the best way to bring this to market, because people need to be ready for it, people need to engage in co-design and algorithm development, and then we need as many people as possible to get access to it. Because a lot of people believe that quantum computers will provide enormous amounts of value, but nobody has shown any algorithms beyond optimization and really esoteric science.

[00:18:41] Andy Ory: We haven't shown any real business and practical applications where there's real quantum advantage, and I don't believe that absence of evidence is evidence of absence. I just think that we haven't gotten enough really smart people access to powerful machines. And so, partnering with Amazon Braket solves a lot of these problems.

[00:18:59] Andy Ory: First of all, Amazon can be very helpful in terms of how our quantum computer gets integrated into the overall landscape of classical computing, the cloud, and just the quantum ecosystem. And it also allows us to segment this machine so that we can expose lots of different people to a machine of this kind of power, and I think we're going to see some really valuable algorithms that drop out of this, and so it's really exciting.

[00:19:28] Andy Ory: That is what QuEra's real focus is. We did announce the following week that we're releasing a machine, or we will have a machine running in our facility that will have orders of magnitude more capability. So, Libra, which is the first machine to go on the Amazon cloud, that's a fault-tolerant quantum computer in 2028, is part of an architectural approach where we see a lot of to continue to evolve this.

[00:19:56] Andy Ory: So, we're very excited.

[00:19:57] Eric Hanselman: It's a big step. We think about some of the changes. Fundamentally, so much of the work that's been done in the quantum realm has come at it with the expectation that you have to manage, fault. You have to manage error rates because error rates were sufficiently high that the expectation was you had to go build in, as part of your entire approach, an expectation that, that you had to account for all of this.

[00:20:21] Eric Hanselman: So, fault tolerance, it seems, gets us into a different problem space in terms of how you really approach, developing algorithms and putting them to work in terms of real, functional capabilities.

[00:20:34] Andy Ory: Without fault tolerance, you cannot build digital gate-based quantum computers at any kind of scale. Full stop. Just can't be done. And there's three things that have changed dramatically in the last 18 months that are really causing us to say the world needs to be ready for these machines, and it's going to happen.

[00:20:55] Andy Ory: Tt's going to happen soon. The first thing is that people, ourselves included, are able to now build these machines at scale in terms of the number of qubits. Not 100, not 1,000, not 10,000, but tens of thousands and more. We can really build these machines that have lots and lots of qubits, and we're able to replenish those qubits as they get consumed so the machine can continuously operate.

[00:21:17] Andy Ory: The second thing that's going on is the ratio of physical qubits to a logical error-corrected qubit. It was initially, when no one had any, people were thinking the ratio could be as high as 1,000 to one. Gosh, if you're going to have 100 logical error-corrected qubits and you need 100,000 qubits to do it, that, that's really a problem.

[00:21:36] Andy Ory: We demonstrated both at QuEra and at Harvard, a much, much more efficient encoding scheme. In fact, I think a paper came out of MIT as well, that we were a part of that shows a two-to-one ratio. Now, I’m not going to say there's a two-to-one ratio. Clearly error correction you need for memory is different than error correction you need for gate-based computation.

[00:21:56] Andy Ory: But suffice it to say, averaging it out, it starts to look like 10, 20, 25. All of a sudden you start to get radically more efficient. So, these quantum error codes can actually be so much more efficient in taking the aggregate number of physical qubits to creating these logical qubits.

[00:22:13] Andy Ory: So that's the second thing. We're going to see logical error-corrected qubits at scale. And the third thing is we have seen continued progress in terms of projected algorithmic efficiency. Algorithms that people thought would take 2 million qubits, then became 200,000 qubits, now look like 20,000 qubits. And I'm talking about error-corrected qubits.

[00:22:36] Andy Ory: And that ratio is going to keep coming down. So when you think of those three force vectors, which is I can build machines at much larger qubit scale, I can have much more efficient ratio between physical to logical, and I can get real efficiency in terms of the algorithms and the requirements for logical error-correction qubits, you begin to see that everything is working in the right direction to get you to a machine that can actually solve real-world problems and have really neat impact.

[00:23:04] Eric Hanselman: Oh, getting to practical capabilities. Although, Ellie, that's a lot of what you've been looking at and, in your perspective, in terms of what that ecosystem needs to, what needs to be composed of, and the pieces that are just starting to get there.

[00:23:17] Ellie Brown: Definitely. I mean, we focus a lot on hardware in the quantum ecosystem, and that's been the historical focus is qubit counts and logical qubits, and all that really does matter. But you could have the best quantum computer in the world, and if you don't have a good algorithm or good access methods or the tooling on top of it, it's just a very pretty piece of hardware, right?

[00:23:39] Ellie Brown: It's not going to actually be useful. And so, as this kind of ecosystem develops, in quantum computing as a whole, the focus is starting to shift and very rapidly shifting into that upper level of the stack. And it's been very exciting to see improvements on the algorithm side, kind of meeting up with the improvements on the hardware and tunneling from both ends to shrink those timelines that we're seeing there.

[00:24:05] Ellie Brown: So, lots of focus, across the entire quantum stack, which is definitely what's going to be needed as we continue.

[00:24:13] Andy Ory: The perspective that's really changed is imagine you wake up one morning and go for a run and you don't know how far you're going to run. You're just told to bring a water bottle. That’s a pretty tough go. Now it's like we're on a football field carrying a football and we're running to the end zone and the end zone's coming towards us at the same time. It’s a totally different mindset in terms of quantum computing and people's conviction that this is really going to happen, and it's going to happen in short order

[00:24:40] Eric Hanselman: And the goalposts are getting smaller, or I guess, given that we're in the midst of World Cup, maybe the size of the goal target is now getting larger. The field is shrinking. But there are fewer defenders in our way. Seems like it's starting to make that path faster and more identifiably achievable.

[00:24:59] Andy Ory: And I have to plug Boston and Boston's embrace of the Tartan Army, during the World Cup. I just have to say, Boston is a great town, and, I love the sensibility. It also is a great place to build a full stack quantum computing company.

[00:25:15] Ellie Brown: We have seen this emergence of like quantum hubs, right? And that is very interesting to watch from, you know, a general ecosystem perspective. And Boston is certainly one of those. So, from your view, what makes make a good quantum hub? Other than, you know, the Tartan Army and the amount of beer available?

[00:25:35] Andy Ory: Resources, culture, will, leadership, capital, community, all of those things need to be present, and it's very hard to create from scratch. And Boston really does have all of those things. I would say that my feeling about leadership is that each of these states, each of these communities have to decide what a win looks like to them, and that's what they really need to focus on.

[00:26:01] Andy Ory: Because quantum itself is this whole new industry that's going to be born. And it's going to be everywhere, so you can't say, I'm going to be the world's number one place for everything quantum." I think that's going to be very hard. But I do think you can say, "Here's what matters to me."

[00:26:17] Andy Ory: So, I'm not a leader in Massachusetts, so it's not up to me, but my guidance and counsel to them is you've got world-class life science and biological research, you have world-class medical schools, quantum, neutral atom quantum computing. We've got the largest, and we're going to be available at scale.

[00:26:33] Andy Ory: I think that the overlay of quantum and life science, simulation, drug discovery, biological research, I think that Boston should be the best in the world at that.

[00:26:44] Eric Hanselman: And I will call out for our readers who aren't familiar with your history, you don't come from this, speaking without a certain amount of experience of starting new ventures in Boston. So, the last time I think we talked, it was 128 Technologies, the networking side. But you do have a lot of experience and an understanding of what's necessary to really get to critical mass in terms of the capabilities that are available.

[00:27:07] Andy Ory: I've been doing it for 40 years through many different businesses, and if you pour enough water on pavement, eventually some seeps in. So, I think I sort of get it. It's taken a long time though.

[00:27:18] Eric Hanselman: This has been great, Andy. I appreciate all of your insights. Thank you very much for being on the podcast.

[00:27:23] Andy Ory: I really appreciate it. Thank you for everything you do, and I look forward to our paths crossing soon.

[00:27:28] Eric Hanselman: Very definitely. And that's it for this episode of Next in Tech. Thanks to our audience for staying with us, and thanks to our production team, including Sophie Carr, Feranmi Adeoshun, Dylan Scheible, and the marketing events team. If you liked this episode, please subscribe or like us. Please keep in mind that statements made by persons who are not S&P Global employees represent their own views and not necessarily the views of S&P Global.

[00:27:52] Eric Hanselman: I hope you'll join us for our next episode, because there is always something Next in Tech.