Stephanie Simmons, co-founder and chief quantum officer of Photonic, a company that Stephanie describes as the largest quantum computing company that you’ve never heard of, is interviewed by Yuval Boger. Stephanie and Yuval talk about their plans for fault-tolerance, their telecom photon spin systems, entanglement distribution, and much more.


Yuval Boger: Hello Stephanie, thank you for joining me today.

Stephanie Simmons: Oh, thank you very much for having me. It’s wonderful to be here.

Yuval: So who are you and what do you do?

Stephanie: Well, my name is Stephanie Simmons. I’ve been in the quantum computing space for quite a while now. First got engaged in 2001 and I’ve been working towards building fault-tolerant or trustworthy quantum technology since I came up through the ranks as an academic, as it were. I lived around the world working on a bunch of different things connected to spin qubits and then came back to Canada and founded a company called Photonic, which is probably the largest quantum computing company that you’ve never heard of by choice. And yeah, we’ve been busy.

I’m the chief quantum officer of Photonic and we’ve been busy behind the scenes really doubling down on something that we think is going to deliver fault tolerance sooner than the market expects. So yeah, it’s a lot of fun.

Yuval: How large is the largest company I’ve never heard of?

Stephanie: We have over 100 people. We have, depending on how you count it, up to 120 people working at Photonic at the moment. So I’m also a professor here in Metro Vancouver and that sets aside the 20 or so people that have working on the research side up there too, which is a separate effort but highly collaborative.

Yuval: So just between us, what’s the secret of delivering large-scale, fault-tolerant quantum computers?

Stephanie: Well, thank you. Yeah, I don’t think it’s a particularly hidden secret. I think in some sense that it’s really exciting to see the breakthroughs that are happening all the way up and down the stack and think about how the breakthroughs, especially on the error correction side, have really moved the goalposts for everybody. It’s kind of like this big brilliant open secret that as soon as you put connectivity into your system, like instead of moving with these planar things, having a highly connected quantum system, you can really do a lot. You could really take those resources that are seemingly insurmountable for current architectures for quantum computing. You could bring it a lot sooner.

So that’s just being slowly realized by more and more people, which is very exciting. And then the other thing that connects to that is of course having a highly connected system is exactly what you would need or get out of a modular quantum computing system or quantum network of quantum computers. So yeah, it’s really exciting to see this shift in the overall industry towards entanglement distribution as kind of like the key unlock for networks and for computing. It’s a wonderful time to be part of the space because that is starting to become a more widely held view and there’s excitement around it.

Yuval: In some companies, interconnects are a four-letter word, right? Because they say, “Oh, well, when we get above a certain number of qubits, we’re going to have to have interconnects, it’s not a good thing because we’re not going to have many-to-many connectivity and it’s more difficult and we have to convert to photonics and what have you. Listening to you sounds like a good thing. So could you explain what that interconnect means and what distances are we talking about? What’s roughly the architecture?

Stephanie: Yeah, so we are in stealth, so I’m not going to go lay out our whole roadmap and architecture, but you’re absolutely spot on that photons are going to be the glue that glues all of these entanglement distribution networks together.

We came at it with the axiom that if you worked natively at telecom, you don’t have this conversion bottleneck. And back to your broader point, interconnects are seen as kind of this thing that you have to attach on later after you have a working quantum system that you do just to modularize and scale beyond a certain box size because all of the quantum systems out there to some level have some environmental controls which give it a box size.

And you have to, at some point, have something horizontally scalable which allows for modular quantum systems. We kind of came at it from the other side is that, well, okay, if you’re going to have to connect lots of these quantum systems, you’re going to have to get really, really good at that, right? You’re going to have to get really good at converting, what we think, converting memory systems into telecom photons that could have formed the glue of this modular, entangled distribution network.

And you’re not going to really be happy with a system that just puts some modular interconnect at the side, as it were, and connects it because that will always be the bottleneck at scale. Just to learn the lessons of say classical, horizontally scalable supercomputing, is the interconnects that drive the performance of the overall system in a big, big way. Learning those lessons, we decided years ago to double down on the search for really, really good telecom photon spin systems because that is the building block of this whole technology of scalable networks, of scalable computers.

It kind of will drive the performance, given that the interconnects are going to drive the performance of these whole things. We got to double down on them and make them as beautiful as we can.

Yuval: So you said telecom, photon, spin. Let’s, let me break that down. So I think we’ve heard about spin qubits in silicon. I understand photons as well. I understand each word separately, not necessarily all together. Maybe you could enlighten me.

Stephanie: Yeah, I’m sure. And I mean, it’s not surprising we are the only ones looking at this particular object at the moment, although we do know of some that are trying to chase us academically. Yeah, the closest sister technology I would say is an NV center, okay? Or an ion trap where the trap is a solid state object, right? So the physics of all of this is about the same, but it allows for essentially hitting both of those buttons in the same unit, right? So with the spin photon interface, it takes lots of different forms, but we wanted one that had really good integration into silicon because of course, if you could get it working in silicon, you have lots of natural advantages commercially. And you wanted it natively working at telecom because then you don’t have any of these conversion things.

So what you have is an object. So in our case, it’s called the T-Center, but for the diamond community, it’s a silicon-vacancy center or an entity center. This is color center quantum computing, which I’m very excited about, where that object, whatever that object is, has both long-lived memory within that physical object, so an ion or whatever, but it can talk efficiently to a particular wavelength of light. So it can be stimulated to emit a photon of a particular wavelength or it can absorb or interact with a particular photon super efficiently.

And that’s basically the interconnect right there. That’s the key building block that kind of unlocks the rest and why we’ve been doubling down on ours. 

Yuval: Is the reason for the entanglement distribution because a lot of error sources are local and then if you spread qubits around and they’re more resilient as a group or is there another reason?

Stephanie: Well, that’s a great question. So if we are to look at large-scale quantum algorithms in a modular architecture, the entanglement is used to do gates between any of those modules. Like entanglement is, you can use it for a two-cubic gate. You don’t need to have proximity to do a two-cubic gate. You can use the photons to do that work for you. Same thing with quantum networks. Distributing entanglement is the challenge to deliver different quantum networking applications to different users. So it’s about getting entanglement where it needs to be. So it’s not just about the errors. It allows you to use the elements you have very differently. instead of having to swap all of your qubits around to do all your operations, you can still just do a direct gate between the two that you want across a room, right? Or across a data center or across the country if you can get the actual quantum networks to work, right?

So it’s actually, it’s a picture of the quantum internet, but it allows you to do all these things without having to do this kind of really local games.

Yuval: The quantum internet and quantum computing, they’re often mentioned in the same sentence because of, “Oh, there’s computing and there’s sensing and there’s communications,” but I think rarely together say, “Oh, well, one day quantum computers will be interconnected in such a way that you could distribute their power and build super quantum computers.” Would someone say that you’re trying to solve too many problems at once, both communications as well as the computing?

Stephanie: Well, that’s interesting. You could absolutely start off with that view. And that’s, I think, the historical view is that these are separate challenges. Let’s try and just solve them with any combination of technology.

But the more you look at it, the more you realize that the problems of one are the solution to the other’s biggest challenge. Or sorry, the opportunities are for one to actually unlock the other. And what do I mean by that? If you have quantum computers where you’re not using these fancy new error correction codes, you’re using the old school surface code, you’re going to need buckets and buckets and buckets of qubits to deliver fault tolerance.

Why does that matter? Well, fault tolerance is trustworthy quantum computing where you can tell it to run an algorithm that has no commercial value and it’ll do it. It’ll generate value and that will move the needle for customers. That’s kind of the point of this whole industry, right, is actually to deliver value.

So we’re going to need to have buckets of qubits that are connected. And really, if you take a look at all of the hardware architectures out there, they’re going to cap out at a certain box size. So you do need some kind of network ability to really hit the scale necessary for a lot of these applications. And so having the ability to distribute entanglement from a networking perspective unlocks fault tolerance in that sense, or it certainly helps.

The other side is, of course, repeaters. Repeaters are the thing that we’ve been really lacking to scale user count and distances on the quantum networking side. And a good quantum repeater is actually a small-scale quantum computer. So in some sense, if you can unlock the intersection in this Venn diagram, which truthfully boils down to these spin photon interfaces to some level at a high-performance level. If you can do that, then you can actually build one technology that has multiple applications, including networking applications and computing applications.

Yuval: If the qubits are imperfect, then one starts to talk about physical qubits and logical qubits and the ratio between how many physical qubits do you need to create a logical qubit. Do you also speak in the same terms or do you look at it differently?

Stephanie: That’s a great question, and I think that’s certainly one of the ones that matters. I think the most exciting thing about these new codes is that those overheads go from 10,000 to one down to 10 to one with similar other parameters which matter because it’s not just the overhead. It’s a whole bunch of other things that go into delivering a useful algorithm. But yeah, it changes the overhead conversation considerably. It just assumes that you have high connectivity, that you can connect any qubit to any other qubit.

So I’m really excited to be in this space because more and more people are recognizing just how far the goalposts have moved in our favor as a field by being able to unlock these codes, which are essentially actually like 5G codes that have been “made quantum.” It’s very, very exciting work. And I wish, what I’m really happy to see is over the past two years, in particular, more and more work has gone into developing these codes because there are a lot of opportunities for improvement already.

They’re still very young as opposed to these older planar codes, which have been kind of worked through in detail for a number of decades now. So yeah, really exciting, but it’s not just an overhead game, right? There’s lots of other aspects that go into delivering something useful at scale.

Yuval: Is the programming model the same? I mean, would your computer just accept QASM code?

Stephanie: Yeah, that’s right, that’s right. Standard gate based, that’s right.

Yuval: How far are you from releasing something to the market?

Stephanie: So I won’t be sharing that with you in this podcast. So we are, despite this conversation, actually in stealth and happily moving forward and keeping our heads down and working to deliver products. So you will hear from us in due course, but not today.

Yuval: Some companies have indeed raised lots and lots of money and could either afford or say, hey, come see us in a few years when we’ll have a fault-tolerant quantum computer. And others, maybe because of desire or maybe because of necessity, say “We have to work with customers today. We got to understand the market, develop initial use cases, and so on and so on.” It sounds like you’re more on the first camp.

Stephanie: We’ve got enough money. We know what we’re doing. We’ll tell you when we’re ready. I’m really excited for the understanding that these codes, and this is partly why I’m excited to have this conversation with you today.

There’s a lot of work to be done on fault-tolerant use case development, right? So what I would say is that there’s use case work happening today around today’s systems. And rightly so, people are trying to find commercial value with the systems that are available today. But I think that’s been to some level at the expense of the use case development and the algorithm development and the application engineering in the fault-tolerant regime.

I think there are some opportunities to go and take a look at the laundry list of quantum algorithms that exist,, and see how that could be mapped and what resources are required, or quantum resources are required to deliver value for different customer use cases. So there’s the regular laundry list of applications that it’s really important to go through and actually figure out how you do it, like how you’d actually grind through that algorithm and figure out the resources and figure out how many. Because what would be very helpful from a push versus pull perspective is to have, as it were, mathematical proof of product market fit. If it requires X qubits and X runtime under these assumptions and that gives your company billions of dollars of advantage, then you should know what that roadmap looks like for your company if these resources are coming online.

How much would you pay for 10,000 logical qubits? What would it be useful for you to use? That I think would be a really helpful shift in the way that people are thinking about use case development. We would love to start having those conversations with people, but I think that it has become clear that the NISQ size systems and the NISQ opportunity is not the same thing as the fault-tolerant quantum opportunity.

And it also would change how people think about the deployment of post-quantum and quantum network changes to our cryptographic system if they understood that fault tolerance is made a lot sooner based upon these new codes.

So yeah, that’s one of the reasons why we’re having this conversation here to try and kind of see what kind of partnerships could exist on that basis and to let people know that it’s actually really, really exciting on the fault tolerance front right now.

Yuval: What do you know today that you didn’t know 12 months ago about the market or anything related to quantum? Before the call, we said no Hamiltonians, so let’s not talk about them.

Stephanie: Well okay, like a week ago, not even a week ago, they came out with an LDPC code that has an overhead of three. So that’s kind of cool. That’s not a market question. I think there’s a big shift in the way that people are thinking about quantum networks and post-quantum cryptography. I think that’s been a big shift over the past year, especially geographically there are different distributions there. Again, people are starting to come to the consensus that the answer is both. We need to deploy software-based post-quantum solutions, but we also should be thinking about quantum networks for critical infrastructure. And that’s a shift I’ve seen. Not a full shift, but that’s moving. It’s an interesting development. And yeah, the broader recognition that modular architectures, entanglement distribution architectures, can unlock a lot. I’m seeing more and more of that. So definitely also the space is becoming more mature, people are becoming a little bit more clear-eyed in terms of what quantum can deliver and when in the NISQ and the fault-tolling regime. And I think that’s all very helpful.

Yuval: As we get closer to the end of our conversation, I wanted to ask you, professionally speaking, what’s keeping you up at night?

Stephanie: I sleep really well, but sometimes my kids wake me up. No, it’s a really interesting dual-use technology, right? This has been known for a while. And so it’s going to be a really interesting set of dynamics that click in at some point. And I don’t know when it’s going to be, but if you go and you take a look at any transformative technology over the past few decades, it follows a very standard pattern where there’s this kind of Cambrian explosion of different approaches and startups when people recognize an opportunity. And then there exists some kind of consensus-forming event where one design ends up emerging as kind of the preferred standard. And then there’s a mass consolidation event and then there’s a big geopolitical aspect to it and talent. It’s really fun to kind of think about quantum in that context because it will go through the same thing. And how that shakes out on a supply chain basis, on a geopolitical basis, just because it’s a dual-use technology is very fascinating. And yeah, it’s a very big game that the whole industry is playing in some sense.

Yuval: And last, a hypothetical. So if you could have dinner with one of the quantum greats dead or alive, who would that person be?

Stephanie: Oh, okay. Okay, I’ll give an answer that probably has been said many times before, but Feynman would be excellent just because it’s a certain echo, right? Especially with his work on the atomic bomb, bringing nuclear physics to a commercial state. It has a dual-use aspect to it. And it’s really interesting to see his views on that, you know, towards the later stage of his life, like what, how you reflected on that.

But if I were to kind of push the definition a little bit and consider semiconductors as basically a branch of quantum to some level, semiconductor physics, it’s a bit different. But if I were, I would love to have dinner with the, what was it, the treacherous eight, the team that formed Fairchild, and just to see what the birth of that whole, like the last time we commercialized a branch of physics. I mean, it’s just, these are human trends, very long trends, and they always have more implications than you can see at the outset. So that would be I think an excellent dinner. 

Yuval: Excellent. Stephanie thank you so much for joining me today. 

Stephanie: Well, thank you I really appreciate it.

Yuval Boger is the chief marketing officer for QuEra, a leader in neutral atom quantum computers. Known as the “Superposition Guy” as well as the original “Qubit Guy,” he can be reached on LinkedIn or at this email.

June 19, 2023