Quantum Progress and Roadmap: How Close Are We?

My name's Andrew Lord. Um, I lead, um, bts, uh, optical and Quantum Research. Uh, and, uh, I'm really thrilled at the moment to be right at the cusp of seeing some of this quantum tech starting to, to really hit reality.

And I want to put a very simple slide up just to demonstrate the power of what's coming. Um, now this is naive, simplistic, so please don't criticize. I'm trying to make a point.

What if I've got at some point, um, an n qubit quantum computer on the left? N might be 2020 qubit quite good. What if I've got another machine on the right, 30 kbit?

What can I do if I put them together? Well, let's talk about how I put 'em together in a minute. But if I could combine them, that's 50 kbit.

Now that hasn't gone up by double or triple, that's gone up by many, many orders of magnitude just by combining them. I, I don't want to labor that point. If I got my laptop and another laptop and put 'em together, I get two laptops.

That's double not the case. In quantum, if I get two quantum computers and put 'em together, it's not double. It is a vastly more powerful machine, okay?

Because the qubits have gone up. It's two to the power N plus M now. So it's a much, much, much bigger number.

So the motivation to connect bits of quantum computer is huge. That connection is an entanglement network. In other words, it's getting entanglement, like I was talking about, um, sending something from me to Tom.

It's getting it between quantum computers now, not people. Okay? Um, it also turns out that building a quantum computer is gonna inevitably involve a network because I can't scale qubits physically in one place.

There's too much stuff. There's super cooling, there's magnets, there's lasers, there's all sorts of stuff. If you go and see these machines, they're big power hungry thing.

You can't just keep scaling. At some point you have to say, stop. Let's just build lots of them and connect 'em together.

And that is an entanglement network. And the motivation for doing this is vast. The power increase is gonna be enormous.

Um, however, that in that interconnection network could well be between computers in a local area, so it could be, um, a local, you know, uh, area near some energy supply. It's not a broad national network at all. What I'm really motivated to do is to give this, um, capability to everybody.

And that's, uh, what I want to talk about next. So here's my naive, but uh, overall picture of what I think of future quantum network will look like. And on the right, my QCs of content computers, and now I've connected them together to get untold power.

I've also put some high performance computing in there because I don't think you always need quantum computing. They're not good for everything. And ultimately, your compute resource will be a combination of quantum and classically compute in the middle.

I've got a cloud where, which is essentially connection. So it's how do I, how does my company, how do my customers, how do people connect into a quantum compute based resource? How do they get benefit from it?

And I've drawn pictures of QKD there because, um, QKD would be the natural vehicle, the connectivity for access. I've also shown a list of our customers on the left and, and also things that we might want to do with it. Um, we might want to optimize our end network.

Um, so I can think of, you know, in 10, 20 years time, all of bts customers wanting to have some, um, access to quantum compute, if it's logistics optimization or all the way through to financial trading, it'll probably just become part of what they do. Um, so how, how do we do that? What, what is the, the network that allows that?

How do we make sure that people get access when they need it without having too much, because these quantum computers cost billions. So what's the charging mechanism for that? You probably only need a few seconds a month.

Um, so how, how does that work? So I'm imagining an orchestration layer over the top that polices and governs access to make sure that people have access when they need it and pay for it. Um, but it's not overused and we don't end up someone using it when they don't need it, and it's not available for somebody else.

So there's a whole efficiency question in this diagram. Um, very quickly, uh, we had a big, um, end of project, um, event today where we are looking at how to put quantum computers into data centers. Um, so this, this project, I've got three slides.

I'm not gonna talk in detail. It's really just to give you some, I guess, flavor of the kinds of things we're doing. Uh, the question is, um, quantum computer and a data center feel like they're very synergistic.

What does it look like when I put the two together? And you can see here lots of examples of thinking of how people are starting to put these things together. Uh, and I want to just go onto this, um, architectural slide, which shows the kinds of things you would expect to see in a future data center that's got some quantum in it.

Now you see there, there's a whole load of stuff in that data center that is there already. Um, but we're now starting to add a quantum layer, adding it to everything that's there already, and making sure that there's resourcing optimization that that makes use of that quantum layer, uh, when it's needed. And the algorithms and the control of all of this are already being developed so that when quantum computers are available, we can slot them in and people can start to take, take benefit of them, um, as soon as possible.

I don't wanna go into, uh, uh, premier Yes. Uh, perspective, the orchestration, uh, and you know, management dimension clearly very important. Yeah.

Do you anticipate that an existing platform from say, an Ericsson or Amdocs will be able to incorporate this? Or do you envision like there will be a need for a new purpose designed orchestration platform to enable, you know, this overall quantum computing vision? What's, what's your perspective on that?

So I don't, I can't speak for specific other than to say Equinix, who we are working with are already trialing it. So I think some DC um, operators are ahead and are kind of seeing the writing on the wall. Uh, some of them already have links with quantum computer companies that, that slide I showed you before has got some, some news points over the last few months.

Um, so some are doing it already and some perhaps aren't. I think we're gonna see some leaders and some laggards the same in the financial industry. You, you've got banks that have whole quantum teams, HSBC have a whole team.

There's other banks that have nobody or maybe just one person. So you're seeing a big differentiation in all in industry, those that are quantum literate and those are hanging, hanging back a bit. Good To know.

Black out. Yeah. Yeah.

I, I recommend to get up to speed here. Here's, um, some work. Andrew, sorry, just a, a question if I could that, my name is Drew Con Murray, I'm with Packet pushers.

Uh, do, can you give us a sense of what you think a quantum network would look like? Are we talking entirely different mechanisms and new, new protocols and new way of thinking about networking? Or are we gonna find a way to get ethernet to do this too?

It's not ethernet. Um, it's completely new, but it might be able to make use of some of the existing infrastructure. Um, but not necessarily, and I've got a a bit of discussion on that, but if you, if you look at the side here, this is, I'm kind of trying to answer that question is what would a quantum network look like?

So this is, um, some work we've been doing recently with EDL, with Cambridge to to say, design a quantum network, do some, uh, optimization. And you can see that we're, we're actually looking at quantum networks from very small scale within a data center to, uh, all the way through to a national network, um, with, with fibers connecting nodes in, in a, in a national grid. Um, can we use the same fibers we have already?

Maybe, but they're quite lossy. There are some very exciting new fibers called hollow core fibers. You might have heard of much, much lower loss potentially.

Um, so maybe we need to install a brand new fiber base to do this. And you might think, gosh, that sounds really kind of tough, but when you think that the end devices cost billions, why wouldn't you spend a lot of money on your interconnect? You know, if I buy a Porsche, I'm not gonna drive it down a B road outside my house.

Uh, I need to go on a motorway. Um, and it's the same with these networks. I expect we would put serious investment into the highways and the con interconnects between quantum computers, do they deserve it?

And they're not gonna work very well without it. And that involves new fibers, new lasers, um, new ways of coupling between the computer and the network. Um, there's a lot of work going on already.

There's places in the US doing this, um, places like MIT universities also in, in the UK and Cambridge. Um, so we're starting to see that whole ecosystem starting to build up, but it's, it's probably behind the compute industry. And I guess that makes sense.

You want the computers first, but, but in terms of the, the networking bits, you know, technologies starting, I mean, one really important one would be a quantum memory or a quantum repeater, something that can buffer, some quantum can say, I've got some quantum coming in, I don't need it yet. Can you hang onto it for a minute? Really difficult.

Um, but a really crucial building block to make this work. But that is being researched at the moment. Can I ask a follow on question?

Yeah. That through the Bruno Walman Eurocom solutions, um, you're talking about the extra power and cooling needed for the quantum computers. Is the expectation that the network infrastructure to support a quantum network would also require, um, larger amounts of power and cooling than traditional network here?

Not really. A little bit maybe, but that's, no, I don't think that's a big issue. Um, so generating entanglement, yeah, it, it needs usually a high powered laser, um, which kind of induces nonlinear effects in the materials that that mean that they send two photons out.

Um, and the two photons are entangled. So some of these techniques are, are a bit more esoteric than regular comms. Um, but quite a lot of the components are still very regular communication based, and it's by no means the same kind of horrendous energy consumption that you would expect in the computers.

And actually things like QKD, if you look under the hood, much of the stuff in a QKD box is just regular. So iOS lasers ERs, standard interferometers, silicon, it's all, it's just configured differently, but a lot of it's same. Andrew, quick, quick, quick, quick question for you.

This is, uh, Jason Ner. Um, so while we're on the topic of conventional networks, is there distance limitations with, with quantum networking versus, uh, the networks we have today? Yes, absolutely.

Um, so if you are sending secret keys, there's an echo on the line, by the way. First on there, if you're sending secret keys, um, the loss of the fiber reduces that key rate. So for example, if I'm sending, I know, uh, a, a million keys per second, um, that after, after a few tens of kilometers becomes, um, 10, you know, a hundred thousand keys and then 10,000 keys per second.

And ultimately, it's not enough keys to, to be worthwhile. So yes, you lose entanglement, you lose keys, um, over distances of tens of kilometers. Um, and that probably means you should focus, at least for now on kind of metro level, shorter distance networks.

Uh, one way around that I talked about satellite is that you, uh, instead of going fiber between countries, which probably doesn't work quantum at all, you, you just go up to a satellite and down again. Uh, and that will help a little bit. But, but that's still very lossy, so absolutely, um, it's a big issue.

Uh, and I think what it will mean is that in the short to medium term, these, you'll get islands, you'll get a quantum island, uh, capability with quantum computers and networking within a kind of a small area. And, and it's gonna be a lot longer before that becomes something that's national scale. So Andrew, a a quick question kind of, um, following on from that a little bit.

Is it, uh, physically impossible, uh, for quantum computing to have a lossless fabric or a lossless network and you, and all the peripheral things have to take care of it? Like ho core fiber and things like that? Um, I mean, how do you solve for that problem?

Um, it won't be lossless, but, uh, some of the losses I've seen for ho core fiber are incredibly low now, um, much, much lower potentially than standard fiber. Secondly, you can kind of overplay. So in other words, uh, if I wanna send some entanglement, uh, between A and B, instead of sending it, I can send double or t triple or 10 times, and I know that some of it will get through.

So there's, there's ways of making sure that I've over-engineered to, to take account of the fact that I'm gonna lose them. I, I put that in really poor physics terms, but, but you can do that. I've talked about qubits being in two states, but you can have qubits in multiple states.

You can set up these atoms or photons in 10, 20 states at once, um, believe it or not. So I can kind of overload my, my network so that it, it it's more resilient to loss, um, so that you can potentially have something that's lossless, but we're, there's a lot of research to, to kind of make that happen. And so I wouldn't, I wouldn't say that's here yet.

And I, and I still think that the distances are gonna be modest for a long time. Um, we're not, we're not talking about, um, quantum across the Atlantic. Uh, and, and for actually one really simple reason, um, making a quantum memory, a quantum repeater is a really difficult thing because I can't clone back to what I said at the beginning.

If I take an atom or a photo, I can't perfectly copy it. And what that means is I can't take an atom and make a hundred copies so that I can go a much longer distance. I'm not allowed to do it.

So I, so I can't, like, like I would do in classical communications when my light level has dropped too much, I just amplify it and off I go again. Uh, and I put amplifiers across the ocean, across the Atlantic Pacific. I build global connectivity using amplifiers that just every time the light level dropped too far, it pull it back.

It doesn't work with quantum because I'm not allowed to claim, I cannot reproduce the quantum state and make a hundred of them. So yeah, some existential questions that will limit, um, the, especially the distance scale. So Andrew, would you see fundamental changes into network topography and topologies, uh, uh, based upon, uh, the arrangements of interconnecting quantum nodes?

Uh, like currently we have right now, you know, flat networks, hierarchical networks. Uh, how do you see interconnecting quantum nodes and how would it change our current routing protocols or current pitching protocols? I do.

Uh, and in the final section, I'm gonna talk about that a bit more. Um, but actually I think that's happening with AI too, because you're starting to see AI, massive data center, a number crunching in regions where there's cheap energy or renewable energy. So I think that network, um, re-architecture will happen anyway.

Uh, and we will just get used to the fact that our heavy compute happens over there where it's, where it's cold or where there's lots of, uh, water or wind or something. Um, and yeah, for sure that means re-architecting. Um, what's the problem with that?

Things that latency. Um, so we need to think about are there situations where my compute has to be near and at the edge rather than far away? And there are situations, I mean, gaming, for example, requires low latency.

So, you know, just one example, uh, financial trading probably you don't want a decision to be made at the other side of the planet when you're trying to trade it and make a, get, get a benefit or an advantage by being quicker than someone else. Um, so yeah, really, really good. I, I'll come to, um, that a bit more in, in the final section.