Transcript | Quantum Networking with memQ

Quantum computing requires quantum networking to reach its full potential, but there are engineering challenges to overcome, especially in building repeaters to extend entanglement between qubits. We dive deeply into quantum networking and its future, including how it will affect interconnect, in this episode of The Post-Quantum World. I’m your host, Konstantinos Karagiannis. I lead Quantum Computing Services at Protiviti, where we’re helping companies prepare for the benefits and threats of this exploding field. I hope you’ll join each episode as we explore the technology and business impacts of this post-quantum era.

Our guests today are cofounders of memQ. We have the CEO, Manish Kumar Singh, and we have the CTO, Sean Sullivan. Gentlemen, welcome to the show.

Thank you. Happy to be here.

Thank you.

Before we roll deep into what you all do, Could you tell our audience about how each of you made it into the QIS field in general?

I took a very nonlinear pathway. I did my undergrad in chemical engineering, then worked in the semiconductor industry for almost five years. Then, I decided to get a Ph.D. By the time I came here for my Ph.D., the winds had changed and a lot of opportunity had started coming up in this new field. That’s where I chose to do my Ph.D. It was related with some of the work that I had done, meaning, we were trying to develop a platform that will allow you to take advantage of the same methods that goes into the making of silicon chips and use those to make qubits. My Ph.D. thesis was to develop a method that would allow us to do solid state qubits on a silicon platform, and I worked on it for about five years. Three years in, I started working with Sean. Around the fourth year, we got lucky with the materials that we’re working with. From there, memQ started.

Sean, how did you find your way to silicon qubits back then?

I started out doing materials science and engineering, similar to Manish, but I was more focused — originally, before I started my Ph.D. — on energy conversion for greener technologies. A lot of the materials and things that I was working on, you started getting into the quantum properties to engineer these materials at the atomic level to maximise things like efficiency for energy conversion. I started getting more into the quantum mechanical, quantum engineering, aspects of materials science at that point and using optics and things like that to characterise these things.

I pivoted after my Ph.D. from focusing on the quantum materials for energy conversion to focusing on quantum materials for quantum information processing — going from energy transport to information transport. That’s why I joined my postdoc at Argonne National Lab. There are a lot of experts in the quantum information science field here in Chicago, and so I was excited to join the group of David Awschalom and Joe Herman’s at Argonne. That’s where I started working with Manish, where we are tackling this problem of making materials and devices for quantum communication applications.

Yes. Oh, I’m going to be visiting Chicago on my birthday. I’m going to be staying at Argonne. If that’s not the nerdiest way to spend your birthday, I don’t know what is.

Instead of then taking this research and publishing a paper and going into academia, what made you guys say, “No, we’re going to form memQ”? How did that get going?

The University of Chicago provides a very good ecosystem. It has a lot of departments. It has a great business school. All of that allows you to explore all the ideas that you want to explore. One of the things that I was interested in is, how do you take a technology from the lab to the market? Having the business school right there, I got to interact with a whole number of VCs, professors, MBA students and learn how to think about these game-changing technologies, or these breakthrough technologies, that come about. Like, how do you even go about thinking whether it would become something that somebody will be willing to pay for in five years’ time?

Having been through that framework, by the time we made our discovery, I was convinced that we had made that final breakthrough needed for this platform that we were developing to become the platform for quantum communication. From there on, it was customer discovery, and that led us to where we are today. Of course, we did publish our papers along the way. But what we also did that was important was file for patents.

Yes. That definitely enables the commercial engine moving forward. Let’s talk a little bit about those keywords you said — your discovery, and the platform. Let’s give everyone a sense of what you’re hoping to build here.

One of the toughest challenges right now is, how do you send quantum information over longer distances? Now, when you think about quantum information, it’s not just about sending photons. It’s about this fundamental resource, in a sense — quantum entanglement: How can you send it over longer and longer distances? That is the resource. It’s fundamental to computing. It’s fundamental to communication. It’s fundamental to sensing. By these three, I mean fundamental to quantum versions of these words.

What our platform enables is, we work with what’s called a rare-earth qubit in order to store the quantum state for longer and longer periods of time so that we can extend this entanglement. It has been demonstrated we can connect A and B, we can connect B and C, we can connect C and D. But in order to connect A to D, you need to be able to store information at D for a long period of time, C for a long period of time. By long, it could depend on the applications — it could vary from a few milliseconds to a second, and that’s good enough to send quantum information. That’s the heart of the problem we are trying to solve.

What are those rare-earth qubits?

We’re specifically working with erbium ions. The rare earths are a grouping on the periodic table. When you look at the periodic table, you have those two rows that are at the bottom. The rare earths are in there. We’re focusing specifically on one. None of these are radioactive — we’re talking about the top row here. This is erbium, which has been used a lot in the communications industry for a long time for making lasers and things that are used in fiber-optic networks. It produces light, emits photons that are at a very specific wavelength that’s in the near infrared — just beyond what you can see toward the red that transmits through optical fibers well. They’re not absorbed. They go long distances. No problem in these optical fibers.

What’s exciting about that is, we have this way of producing these photons which is great for quantum information, for communication. But there’s also a cool property about the electron structure of the ions themselves. When they form in matter, in solid-state matter, you can have individual electron spins at different energy levels — we hear things about spin qubits. This is basically a spin qubit that you can interface with light. You can excite and read out and do all of that with light. You can make that light go through telecom networks. There’s some cool stuff that you can do with that. That’s for both communication and memory, storage, those types of things.

When people hear about quantum networking before this, it’s very set things: You’ve got this protocol. You’re doing something on your computer, and you’re sending just pure information out there. It’s not information that’s live. It’s not like a calculation that’s in progress or anything. It’s, like, “I wrote this thing, and I send it out as bits.” That gets moved around to the right place because of the networking stack and all that good stuff. When people try to visualise quantum networking, it’s really different. Do you want to give a high-level to help listeners understand why it’s so different and what its potential benefits are?

When it comes to quantum communication, the heart of it is, how do you send entangled photons over very long distances? That’s the problem. Now, an analogue here would be like, say you can generate a voltage by using hydroelectric plants or something — you can have voltage generation. But you’ve got to send it to the point of consumption, in a sense. For that to happen, there are lots of stairs too. If you use two 20 volts to send it over long distances, you lose all the current that you have along the path. There will be resistance. You’ll lose everything on that. To be able to send that over long distances, you need a transformer to up the voltage to thousands — 33,000 kilovolts is what’s currently used. That is the key thing there that enables you to distribute electricity to longer and longer distances.

We can do wonderful things with entanglement today. Very beautiful experiments have been done. But in order to send it over longer and longer distances, you need two things: You need to be able to store it at intermediate stations. That is important because you can only send single photons, and you cannot copy and amplify it, because that destroys the quantum state.

The same thing that gives us the promise of unconditional security also brings this challenge that it’s very difficult to send single photons directly over long distances because they get absorbed as they propagate. That’s a central challenge there: In order for any platform to solve that problem, you need to be able to transmit at the telecom wavelength, like Sean explained, and you need to be able to have thousands of these signals, or, for the lack of a better word, thousands of these communication qubits at every base station. From our estimates, these base stations would be somewhere between 50 to 80 kilometers apart. In order to get a meaningful rate between them, you need thousands of qubits at each station. That’s where the platform needs to be scalable so that you can have thousands packed on a single chip.

Then, it needs to have this memory capability. You can store it while you connect the other bit. Like I said in the previous example, say A and B get connected. Now, B and C need to get connected, and these qubits at B then undergo this special operation called an entanglement swap. That is what ends up connecting A and C. The magic is that A never sent anything to C, but this entanglement swap ended up connecting A and C. Now, A can send information to C without actually sending information directly. That’s where they can use the quantum, the entanglement, as the resource that gets used up once you transfer information from A to C.

That’s easy to visualise, and you’re avoiding the no-cloning principle this way.

Exactly.

Just to draw an analogue with the digital communication that we think of now, everything is converted into bits and sent over a communication link. This is almost more of an analog communication: You’re sending the raw qubits in a network that’s able to speak the language of quantum physics rather than having any translation to digital packets, and then you can use algorithms to unpack that. You can’t do that in quantum physics. You can do measurements in a measurement basis, at which point you collapse that wave function, but you need to preserve this quantumness over the entire channel. That’s how it’s a little bit different.

Basically, these are like repeaters. They’re simplified quantum computers that do one operation: the entanglement swap.

Well, that’s how we envisioned the first generation. Professor Jeong at the University of Chicago came up with these generations of repeaters about seven, eight years ago where the first generation is only capable of doing what you just said. Then, as you add more capabilities on it, if you’re sending information from A to B, there is a very good chance that you’ll lose photons. It’s a probabilistic process. You’re going to lose some of the qubits. You’re going to get some errors over that channel. Now, you need to be able to correct for those errors. In order to do that, you need more capabilities at these nodes. These nodes now get up to the second generation where you can correct for some of these errors.

Then, there’s a third generation that is defined as where you can correct for all the errors: propagation errors, gate-based errors, loss errors. At the third generation, you’re essentially working with these very specific quantum computers in their own rights. If you can do error correction, and if you can do manipulation with a bunch of qubits, what you have is a quantum computer, essentially. With the third generation, what you have is very specific specialised quantum computers at each node that are dedicated on information at transfer.

That’s where our vision for memQ aligns well. The platform that we are developing, it’s not just memory. It’s not just entanglement. It’s about developing a whole system on a chip where, in generations, we prove a set of capabilities, we add more on it, add more on it, and we complete the full arc with these quantum computers at each base station.

This generation seems to match and map perfectly to what we hope to get out of quantum computers in general: We’re hoping to get lower errors and then error correction. You’re hoping to chart the progress along the way. Networking has always been minicomputers — a router is a minicomputer. You can run Linux on there or something. It’s a similar approach. What other technical challenges do you have to overcome right now?

This has application beyond just quantum networking — interconnect in general, which is moving toward photonics, using light to transmit information quickly. One of the key challenges is taking light that’s in a fiber or in a fiber network and putting it onto a chip which may be running silicon photonics or some sort of photonic circuitry. The reason that’s challenging is, these are different materials. Light propagates them inside them in different ways, like in silicon versus a silica fiber, they have very big different refractive indices, things like that. You have almost an impedance mismatch between your fiber and your chip. You need to come up with ways to carefully and slowly adapt, getting light from the fiber onto the chip, and vice versa. There’s been a lot of progress in that.

One of the challenges in quantum photonics and quantum systems that are using light is, a lot of these operate at cryogenic temperatures, where it’s cold. You stick this in a cryostat, you cool it down to 4 Kelvin, or, for some systems, maybe you have to cool it down below in a dilution refrigerator. You get all these shifts that happen with temperature: Things contract and move around. The refractive indices, materials change at different rates. There’s not a good solution for getting light on and off chips in a way that has a high efficiency at cryogenic temperatures that you could make a package, stick this in, cool it down and expect it to work as you would want. That’s one of the things that we’re focusing on.

Basically, if the efficiency drops to 50%, you lose half of the photons that are coming up. One thing that Sean likes to say is, every photon is sacred, and we need to take care of every single photon that comes up.

The BB84 protocol doesn’t consider everyone is sacred, but that’s fine. What kind of innovations do you feel you’ll be bringing to this? Obviously, a lot of people are looking at quantum networking. What do you think separates memQ here?

The key innovation has been this light-matter interaction that we can engineer and that we are building upon. Rare earths have good spectroscopic properties, and they have been under investigation for 40, 50 years. As Sean said, they have been using optical communication-network lasers and whatnot. In order to get the properties that you get in ball crystals, we have developed engineering methods that allow us to transfer some of those properties and engineer the same properties that we see in balls, or at least approach them. Being able to do that is a key breakthrough.

What memQ aims to do is build on that and engineer this to a point where you have the thousand devices that we were talking about. Right now, no platform can make a thousand devices that are very similar to each other. We believe that our variations are small enough that we can have a thousand devices that are very similar to each other. By adding additional controls, we can make them exactly similar to each other. That’s where the true power lies.

Yes — leveraging all the decades of expertise of the microelectronics industry. We both have some experience with that — Manish, in particular — but that’s something that for large-scale processing scalability, we want to try to leverage. We’ve chosen a system that we think can work well for that.

What about the impact of protocols on a platform like this? Let’s say we want to use this for interconnect one day. Let’s say we want to have two, six, 10, however many quantum computers behave as one — and a few companies are working on this already, like Entangled Networks. They’re trying to figure out what that protocol is that lets these machines behave as one over distances or whatever. How would that impact your devices? Would you have to build something specially for that? Do you have to already start giving thought to how these are going to be used one day to accomplish that magic of entanglement on multiple systems?

The beauty is that at the hardware level, as it exists right now, as long as you bring enough photon, whatever device you have, as long as it brings enough photon, you can plug into this network. If you’re trying to connect a microwave computer like the kind of computers that Google and IBM and Rigetti and all of them are developing, you’ll need what’s called a transducer. You need to be able to convert the microwave photon to an optical photon in order to plug into what would be memQ’s networks. As long as you can do that — the hardware level itself — it’s agnostic to what computers are coming in. It treats every photon as the same.

Interesting. With the right software or protocol, it might be possible to do interconnect between a transmon and a trapped ion or something if they can connect your network ultimately.

Are you guys working on anything in the field of Interconnect? Are you trying to experiment with those aspects of it, or are you leaving that to other folks to use your architecture?

Our code capability is in the light matter and the memory division in that sense. We’re trying to develop that code of our network. That’s the primary problem. The secondary problems are these interconnect problems, which we are not the best at right now. If other solutions come up, we will just make sure that it adapts to our network. But the focus is to develop the heart of the network right now.

What kind of security concerns do you think there will be with quantum networking? You build something like this. What kind of vulnerabilities will there be? I’ve been involved in that side of it for a while, with the early QKD experiments and things. There have been pretty funny hacks in the early days with flooding laser light and things like that. What do you anticipate would be some concerns?

One of the things that you just mentioned is like in any of these networks: If you can flood it with light, you can just stop all sorts of communication. That’s a major challenge. There are some solutions that are under research, but, yes, I’ll acknowledge that as a challenge.

That is a major challenge. QKD in general is seen as vulnerable to a variety of attacks, and like Manish is saying, there are work-arounds that you can come up with. That’s from a purely security standpoint. In terms of enabling things beyond just the cybersecurity aspects, like you were talking about interconnect and distributed quantum computation, this is the way to go, where you’re sending flying qubits over a long distance. We have to think about if we want to avoid someone doing a denial of service on our distributed quantum computation, then maybe we have to come up with some clever ways of shunting things around, changing network typologies — things like that — quickly. Yes, that’s definitely an open problem that a lot of people are working on.

It’s not a privacy concern. You can’t eavesdrop on a qubit and have it mean anything. You’ll damage the computation, but you won’t actually learn any information.

Yes. I’ll just add that one way to think about it is if, instead of a linear topology, you’re thinking about a network that has many connections, you can essentially cut off that region where an attack has happened and find out another that would deliver the entanglement.

These advances that you’re making, you’re hoping to make in these miniature quantum computers that are almost like ASICs. They’re like this one application that they’re built to do. Do you see any of this filtering upstream to other types of computing development? Do you anticipate, like, “We’re doing such a great job at this” that maybe you want to implement this in your general, fault-tolerant systems one day? Do you see anything like that coming out?

Yes. As I alluded to before, there’s a lot of interest in going toward photonics. We’re laying the groundwork for a lot of integrated photonics technologies, using photonic chips for low-power computing, where this is classical computing and you’re doing some sort of Fourier transform operation using photonic elements, but you could do this with low power. Getting light on and off chips there, that’s a place that can get a lot of benefit.

Our methods for getting entanglement, on getting a whole number of devices that are very similar to each other, that maps to the quantum computing effort that focuses on photonics. That section maps to any of those problems that are being solved using this. The ability to error-correct, for example, build maps directly to any photonic platform — like Sean said, it even maps to nonclassical platforms. But especially in quantum platforms, there are companies out there that are developing quantum computers out of these photons. Say, a memory, a module, for example, could map onto that. Right now, if I understand correctly, the memory is just a loop of fiber. That’s good enough to introduce a delay into what they are trying to develop. What we are trying to build has the memory capability at the center of it.

We’re not trying to work around the problem. We’re trying to solve that problem, because it’s very important for the quantum communication application. A memory application that gets developed for quantum communication can also map the quantum computation, because if you’re loading very large data sets, you need to be able to store them. You need to be able to have these registers that can store these quantum results for a while, and then you’d load new data, perform computation, reload and so on.

It could maybe even impact the depths of circuits or something like that going forward in quantum computers?

Yes. Or maybe new types of quantum algorithms that utilise something like a quantum memory, which is not typically used when you look at a normal algorithm today.

When would you anticipate having a basic network running that companies could finally plug into? I know that’s probably the billion-dollar question.

I’ll try and answer a simpler version of that question. We expect to have our devices plug into the Chicago quantum network. There’s a test bed that already exists in the Chicago area. It’s about 124 miles. That would be the perfect place to test out memQ devices. As we build on, if it works on that network, then we can bring on more folks who can use our devices, our capabilities, on this test network in order to build more applications and test out the technology. We see this in three different waves. The first one, we are pretty hopeful we’ll get a prototype of one of the components by the middle of next year, another one in another 18 months or so and a network prototype where we are actively managing entanglement over a network like this, something like 24 to 30 months.

Wow. There could be some real traction here. Are there any other types of projects that you’re working on that you wanted to talk about before we close, or are you focused on that goal right now?

Now, the key for a young, small startup is to stay focused in one problem that we know how to solve very well and solve it. Then, we grow the team and focus on other problems too. Very focused right now.

This is going to be very critical to the success of QIS in general. I’m sure of it — in having networking. I’m excited to see what comes out of this. Thank you so much for both of you taking the time to come on down.

Thanks, Konstantinos.

It was a pleasure chatting with you.

Now, it’s time for Coherence, the quantum executive summary, where I take a moment to highlight some of the business impacts we discussed today in case things got too nerdy at times. Let’s recap. Quantum networking allows for entanglement between qubits to be extended across great distances. Doing so has some significant technical challenges.

One of these is the no-cloning principle of quantum mechanics, which prevents an entangled state from simply being copied. Instead, we have to use entanglement swapping. For example, a qubit from a quantum computer — let’s call that point A — entangles with a qubit in a quantum repeater, point B. That repeater is an application-specific quantum computer of its own tasked with entanglement swapping. The repeater at point B will entangle with another repeater further down the fiber, or point C. This process continues until it reaches another quantum computer sensor, completing the circuit. This could allow a qubit from point A to entangle with, say, point J.

Quantum information is protected from eavesdropping by its, well, quantum nature. An observation made on such a network will destroy the state. We’re familiar with this principle from years of QKD research. However, like QKD, we know that denial-of-service attacks are a real possibility via flooding fiber with lasers, for example. Protecting against such attacks will represent the first type of security control needed in quantum networks and repeaters.

memQ is working on making these repeaters a reality on the Chicago network test bed within two to three years. They’re building the rare-earth silicon qubits that handle the swaps and quantum memory for the device. One day, protocols like interconnect can work with these systems to allow multiple quantum computers to work as one, as regular listeners have heard discussed on this show before.

Excited by the future of quantum networking and want to be a part of memQ? The company is hiring. Check out their website in the show notes for more information.

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