Paul (00:01.382) Hello and welcome to this episode of the Digital Humans podcast. I'm Paul Ashcroft. This podcast explores what it means to be human in a world where technology is rapidly reshaping our reality. We meet entrepreneurs, startups and industry pioneers at the forefront of innovation, diving into breakthrough ideas and disruptive technologies that are shaping our future. Today's episode is all about quantum computing and we're delighted to be joined by our guest today.

Nathan Phillips from Trek. Welcome Nathan.

Nathan has over 20 years experience managing the development of advanced commercial software products for large corporate and public sector clients in a variety of industries. He holds a degree in computation from the University of Oxford. And with this experience, he's now bringing best practice methodologies, architectures and patterns to the world of quantum enabling classical software. His passion is centred around compilers, domain specific languages and the hardware software interface. So...

Nathan Phillips (00:35.224) Thank

Paul (01:03.641) We're delighted to have you with us today, Nathan. Welcome to the podcast.

Nathan Phillips (01:08.578) Thank you Paul, I'm excited to be here. It's very timely.

Paul (01:11.162) There's a lot in there. There's a lot there even in your bio as I read it out for us to unpack. But first I've got to say for those that are watching the video, I have complete background envy today. Nathan is sitting there with a bona fide quantum computer behind him, aren't you Nathan?

Nathan Phillips (01:30.04) not currently operating and anyone who knows about it will see this and keep it not connected at the moment. But yeah, you're right. This is all the key components of a computer.

Paul (01:36.486) Yeah, you're right. This is all the key components for a computer. Beautiful. Well, why don't we start there? don't you please just tell us a bit about yourself and about Trek.

Nathan Phillips (01:46.744) Now, as I said, it seems like a really timely interview. is actually just most of my team aren't here today because they're all off in America preparing for next week's global physics summit hosted by the American Physical Society. It's International Year of Quantum Science and Technology. So quantum computing is taking center stage there. So we've got all hands on deck over in America. Exciting times. Trek, as a lot of your people you interview.

Paul (01:59.525) Physical Society, it's International Year of Quantum Science and Technology, so quantum computing is taking center stage there, so they've got portions on deck over in America, exciting times. Trek, as a lot of your people you interview, I think is a startup, it raised five million dollars last year on the back of a transistor. The world of quantum computing is reaching an inflection point, and by that I mean it's moving out of academic...

Nathan Phillips (02:14.4) I think is a startup. We raised five million dollars last year on the back of a premise that the world of quantum computing is reaching an inflection point. And by that, mean it's moving out of academic laboratory sort of spaces and into the manufacturing space from toys for doing experiments with heading towards appliances for commercial application. And so we're building a business where we're taking manufacturing techniques to actually manufacture quantum computers built on an open architecture by

Paul (02:26.31) laboratory sort of spaces and into the manufacturing space from toys for doing experiments with paying towards appliances for commercial application. And so we're building a business where we're taking manufacturing table apps. actually manufacturing on some of the reasons built on an open architecture. By open architecture, I mean where different supply chains come together with fits interfaces between them. And so we have this interoperable model where we can.

Nathan Phillips (02:44.154) open architecture I mean, where different supply chains come together with fixed interfaces between them. And so having this interoperable model where we can combine the best technology from a variety of suppliers into useful machines. As for me person, I'm pretty new to this world. I only officially joined Trek last November. One of Trek's principles, maybe you caught it in what I already said, is that the quantum industry has a lot to learn from the surrounding industries in which it's built. So as you mentioned in my bio, thank you very much.

Paul (02:55.781) the best technology from a variety of suppliers into least quality machines. As for me personally, I'm pretty good at this as well. I only officially joined Trek last November. I to trust principles.

Maybe recording what I always say is that the quantum industry has a lot to learn from the surrounding industries in which it's built. So as mentioned in my bio, thank you very much. I bring best practice from the world of software development, these methodologies and things to support developing commercial-grade software infrastructure around the hardware stuff that the physics thinks are doing. It's all part of our vision to move the whole quantum computing experience out of academia, maybe something that fits more naturally in a commercial business world than the value you and I are used to spending our time in.

Nathan Phillips (03:14.074) I bring best practice from the world of software development, those methodologies and things, to support developing commercial grade software infrastructure around the clever hardware stuff that the physics peaks are doing. It's all part of our vision to move the whole quantum computing experience out of academia and make it something that fits more naturally in a commercial business world that people like you and I are used to spending our time in.

Paul (03:37.333) Trek is about trying to, it's in that race to build essentially a practical quantum computer, something that is, whether it's affordable, but certainly something that corporations, governments can actually start to use at a more practical level in the way they want to do so. Is that the goal?

Nathan Phillips (03:57.944) I think...

There's still a reasonable amount of work to be done towards that goal and a lot of that work still is going to happen in those laboratories that I've mentioned. I think what we're trying to do is perhaps democratise things, make quantum computers more accessible so that they're not limited to people who have got a team of PhD students on their staff, but actually you can almost order an off-the-shelf quantum computer where something will be delivered and installed for you and you can be given enough training to get yourself to the point where you

Paul (04:12.085) more accessible so that they're not limited to people who have got a team of PhD students on their staff. But actually you can almost order an off-the-shelf points computer where something will be delivered and installed for you and you can be given enough training to get yourself to the point where you can start to experiment with it now.

Nathan Phillips (04:29.39) can start to experiment with it. Now, who knows how long it's going to be before we get to system where you plug it in and you sit on your desktop. And that's some way off, as you can see by the size of the thing behind me. But yeah, we're certainly hoping to, let's say, democratize quantum computing.

Paul (04:33.124) who knows how quite a long it's going to be before he gets a system where you plug it in, you sit on your desktop. That's some way off as you can see by the size of the thing behind me. But yeah, we're certainly hoping to let's say democratize quantum computing. Brilliant. It's exciting. But let's back up a bit. Maybe dive in first and just tell me and our listeners what exactly is a quantum computer and how how is that different from what we would

Nathan Phillips (04:51.384) Hmm.

Paul (05:02.584) have on our desk today, which essentially is a classical computer. What are the main differences?

Nathan Phillips (05:07.574) Yeah, great question.

I guess the main difference is that they're built on entirely different sets of physics. You think physics is what physics is, but classical computers are built on electronics and physics that we understand in terms of the flow of electrons, whereas quantum computers are built on quantum physics. And in particular, there are two properties of quantum physics that we use to enable a whole new different way of doing computing. One of those is we call it superposition, which enables to literally put an element into a combination of two opposite states at the same time.

Paul (05:16.694) is what physics is, but classical computers are built on electronics and physics that we understand in terms of the flow of electrons. Whereas quantum computers are built on quantum physics, and in particular there are two properties of quantum physics that we use to enable a whole new different way of doing computing. One of those is what we call a summa position, which enables to literally put an element into a combination of two opposite states at the same time. It's not

Nathan Phillips (05:39.53) It's not that we don't know which state it is, it's literally in a combination of those states and that's superposition. And the other property we call entanglement, that enables us to link the state of one particle to the state of another, even when we can't look at the state of either of them. And that combination of those two properties means we can bring a whole class of problems that classically would be too difficult to solve in a reasonable amount of time. These are intractable problems or part of the class of NP problems.

Paul (05:42.99) we don't know which state it is, it's literally in the combination of those states and that's secret position. And the other property, or the tan rate, enables us to link the state of one particle to the state of another, even when we can't look at the state of either of them.

And that combination of those two properties means we can bring a whole class of problems that classically would be too difficult to solve in a reasonable amount of time. are intractable problems or a particular class of NEP problems. It brings those problems into the domain problems that actually are solvable. What would be an intractable problem? What's an example of one of those? A classical one that you always learn at university is the travel itself.

Nathan Phillips (06:09.676) It brings those problems into the domain of problems that actually are solvable.

Nathan Phillips (06:18.52) The classical one that you always learn at university is the travelling salesman problem. I don't know how many people need to solve a travelling salesman problem in real life, but problems where you're trying to optimise a complex grid. And these are problems where you can solve it for a salesman travelling between two cities, you can solve it for a salesman travelling between five cities. With a computer, you can do it for 10 cities. But as you add more stops that they have to go to, the problem doesn't scale linearly, it scales exponentially or possibly sometimes even

Paul (06:24.708) how many people need to solve a traffic in real life, problems where you're to optimize a complex grid. And these are problems where you can solve it for a certain traffic between two cities, you can solve it for a certain traffic between five cities. With a computer, can do it for 10 cities, but as you add more stops that they have to go to, the problem doesn't scale linearly, it scales exponentially, or possibly sometimes even worse, in next year. And that means that those problems

Nathan Phillips (06:48.474) less than exponentially. And that means that those problems, for some size of problem, just become intractable to the point where you can't solve them with classical computing. And computing isn't just faster computing. It doesn't just half the time. It changes the scale of the problem. It takes it from an exponentially difficult problem to a linearly difficult problem, let's say. That may not be true in many actual real cases, but that's the principle of what we're trying to achieve.

Paul (06:54.755) for some size of problem just becoming tractable to the point where you can't solve one with classical computing. Computing isn't just faster computing, it doesn't just pass for time, it changes the scale of the problem, it takes it from an exponentially more problem to a little bit difficult problem, let's say. That may not be true in many actual real cases, but that's the principle we're trying to achieve. And you focus on a specific

area as I understand you focusing on superconducting quantum computers. Why this approach over other approaches and what are the other approaches that you could have adopted?

Nathan Phillips (07:35.576) You

That's not quite true. That track is actually pretty agnostic when it comes to the underlying technology used by the quantum processing units in the machines that we ship. When I say quantum processing units, you can think of them like so it's a QPU, which you can think of like a CPU, a classical computer. And so that's the thing that is superconducting or trapped. And if the technology works and there's demand for it, then track will create a machine using it. And that's part of what I was talking about in the introduction. Until recently, every computer has been built almost as a one-off

Paul (07:37.346) Trump is actually pretty agnostic.

Paul (07:43.331) processing units, it will be shipped. When I say quantum processing units, can think of them like, so it's a QPU, which is like a CPU, that's a computer. So that's the thing that is superconducting or tracked by it. And if the technology works and there's demand for it.

then a track will create a machine using it. And that's part of what I was talking about in the introduction. Until recently, every computer has been built almost as a one-off lab project with companies trying to own the whole supply chain and do everything in-house. Whereas companies like Trek, we're bringing the assembly of these things, but not having spent billions of pounds on investing in superconducting quantum processes or track dynamic computers. So if somebody says, want to track IMOC, we will find the supply chain and do what work needs to be done to make that supply chain work.

Nathan Phillips (08:06.366) lab project with companies trying to own the whole supply chain and do everything in-house. Whereas companies like Trek, we're kind of bringing the assembly of these things but not having spent billions of pounds on investing in superconducting quantum processes or trapped iron cube use. So if somebody says I want a trapped iron one, we will find the supply chain and do what work needs to be done to make that supply chain work with the machines that we're delivering. And so

Paul (08:29.745) with machines that we're delivering.

Nathan Phillips (08:32.721) Our sole purpose is really to build these reliable systems based on open architectures, I mentioned open architectures earlier, that bring components together from the supply chain that's forming our ecosystem. And if those open architectures don't exist, then we want to help create those. And that way, hopefully, we'll be able to create quantum computers more quickly and also provide systems that are more useful to customers that don't have their own quantum research departments in place. But to answer your question directly, you're right. The system we're working on right now, the one behind me, that is a superconduct

Paul (08:33.623) Our sole purpose is really to build these reliable systems based on open architectures, I mentioned open architectures earlier, that bring components together from the supply chain that's forming our ecosystem. And if those open architectures don't exist, we want to help create those. And that way, hopefully, you'll be able to create onto computers more quickly and also provide systems that are more useful to customers that don't have their own quantum research departments in place. Let's answer your question directly. You're right. The system we're working on right now, the one behind me, that is a super

Nathan Phillips (09:02.586) and that's mainly because it's the most established technology with the most mature supply chain at the moment. It's the same technologies that Google and IBM are using, for example, and our team has deployed several of this type of system in previous roles, so we have got a lot of expertise in superconducting, but we're really keen to serve customers that want to more than one modality in the same quantum cluster. In fact, that's probably one of our chief distinctives as Trek. We've built our stack around this idea that actually a single quantum cluster could

Paul (09:04.428) and that's mainly because it's the most established technology, it's the most mature supply chain in the world, it's the same technologies that Google and IBM are using for example, and our team has deployed several this type of system in previous roles, we have got a lot of expertise in conducting, but we really keep to serve customers that want to more than one modality in the same quantum cluster. In fact that's probably one of our chief distinctives as Trek, we've built our stack around this idea that actually a single quantum cluster could

Nathan Phillips (09:32.506) include multiple modalities. You could have a superconducting and a trapped ion system in the same cluster. And that's something I don't think anyone else is doing.

Paul (09:33.204) could include multiple modalities and could have super-leveling and a trapped-variant system in the same cluster. And that's something I don't think anyone else is doing. So at the moment, quantum is not in the world of VHS versus Betamax. And one will win? Is one modality likely to win in the end, do you think?

Nathan Phillips (09:54.208) It's a really interesting question and I think we can't answer that yet. It is possible that, so one of the reasons for wanting multimodality clusters at the moment is because we're still trying to compete these things against each other. But even in the long term, I could see a situation in which we will find that certain modalities, certain technologies for QPUs will work better on certain problems. And so you might end up where you do have a system that has more than one type of QPU in it in order to be able to solve more than one type of problem.

Paul (09:58.562) is possible. So one of the reasons for wanting multi-modality clusters and invariants is because we're still trying to compute these things against each other. But even in the long term, I see a situation in which we will find that certain modalities, certain technologies, the QGEs will work better on certain problems. And so you might end up where you do have a system that has more than one type of QP-irrhenated in order to be able to solve one type of problem. Now, one thing I

I am aware of, maybe our listeners are aware of, is that one particularly special thing about quantum computers, at least today, is they have to operate at incredibly low temperatures, or colder than space, if I'm not wrong. Maybe you could tell us, and you talked a bit about the supply chain, so why do they have to operate at such low temperatures, and then how do you get these things to do that? They have a very complicated-looking, amazing-looking structure that is almost unrecognisable as a computer.

Why is that and how do you achieve it?

Nathan Phillips (11:01.1) So those cold temperatures that you've heard about and you're right, it is colder than space in my office and that's not just because nobody's turned the aircon on today. The really cold temperatures you've heard about are there specifically for semiconductor based QVues which would include the superconducting ones and there's two reasons for that. Firstly, at the moment there just aren't any viable options for materials that we could use to fabricate superconducting chips at higher temperatures or pressures than we're working with right now. And secondly, the big problem

Paul (11:05.938) company space in my office.

Paul (11:14.347) semiconductor-based QBs which include superconducting ones and there's two reasons for that. Firstly, at the moment there just aren't any viable options for materials that we could use to replicate superconducting chips at higher temperatures or pressures than we're working with right now.

And secondly, the big problem that we have with interacting with these tiny silicates, these are atom-sized silicates, the problem we have with interacting with is noise. Noise is always our problem. And at the frequency these devices operate at, thermal noise, that's just the noise emitted by the air molecules moving around you. That significantly impacts operation, even when you're down to a fraction, sort of a Kelvin. So just the noise of the heat emitted by all the atoms moving around you.

Nathan Phillips (11:31.016) that we have with interacting with these tiny circuits, these are kind of atom-sized circuits, the problem we have interacting with is noise. Noise is always our problem. And at the frequencies these devices operate at. Thermal noise, that's just the noise emitted by the air molecules moving around you. That significantly impacts operation, even when you're down to a fraction of a Kelvin. So just the noise of the heat emitted by all the atoms moving around you, we need to get rid of that.

Paul (12:00.051) We need to get rid of that, so we're cooling things down to temperature where even air molecules stop moving. So the whole thing operates in a vacuum and the air molecules tend to end up all stuck to the inside of the vacuum shielding. So the actual space inside it will go to the vacuum. And so, yes, we're reducing noise, we're creating conditions where our superconductors work and the idea is called with the space, just the background radiation in space means that however far you are from the sky, you're still only around for

Nathan Phillips (12:00.89) So we're cooling things down to temperature where even air molecules stop moving. So the whole thing operates in a vacuum and the air molecules tend to end up all stuck to the inside of the vacuum shielding. So the actual space inside is a total vacuum. And so, yeah, so we're reducing noise, we're creating conditions where our superconductors work. And you're right, it is called within space. Just the background radiation in space means that however far you are from a star, you're still around four Kelvin.

Paul (12:30.006) Yeah

Nathan Phillips (12:30.81) operating at less than 50 thousandths of a Kelvin, down to 10 thousandths of a Kelvin. Not quite the coldest place on Earth, but colder than anywhere that's not on Earth. And then I'm on Milton Park, just a couple of miles down the road we've got a fusion reactor which is about 500 times hotter than the hottest star in the universe. Humans are amazing, aren't they? Aren't we not amazing that within three miles we've got the hottest and the coldest places in the universe? It's a crazy world.

Paul (12:59.768) I hope you don't bring those two things too close together. I'd be afraid of what happens.

Nathan Phillips (13:01.912) Well the good news is we're both operating it well the area that we're cooling down is about that that size and the area that they're cooling down is even smaller heating up is even smaller they're heating up very very small bits so high temperatures let's see

Paul (13:15.507) So at the heart of a quantum computer then is essentially a series of fridges to put it very plainly that is sort of allowing you to cool down and cool down and cool down to that level.

Nathan Phillips (13:26.646) Yeah, so in the background behind me is a fridge from a supplier called Mabel.

We're based just out of Oxford. There's another great supplier around here called Oxford Instruments. They do fridges as well. And these fridges, they work on the same principle as the fridge in your house. We use slightly different coolants. So we're using a gas-foured helium three. And obviously, you know, hopefully you know, fridges work by taking gases and compressing them and expanding them. And it's the same principle here. So we have at the top of the fridge behind me, that would be room temperature. And then it would drop down by a couple of hundred degrees.

Paul (13:38.112) these fridges, they work on the same principle as the fridge in your house. We use slightly different coolants. We're using a gas-forked helium three. As you know, I hope for you to know, the fridge works by taking gases and compressing them and expanding them. It's the same principle here. So we have at the top of the fridge behind me, that would be room temperature. And then it would drop down by a couple of hundred degrees C.

Nathan Phillips (14:01.626) and then drop again by another 50 degrees C and the cell will eventually end up kind of four degrees above absolute zero and then a couple more stages and we can get that down to the thousandths of a Kelvin above absolute zero. We can in theory get more than one QPU in the same fridge so rather than being a series of fridges it's more likely that there'll be one fridge with multiple QPUs inside it.

Paul (14:04.864) 50 degrees C and settle between four degrees above absolute zero and

we can in theory get more than one QPU in the same fridge. So rather than doing a series of fridges, it's more likely that there'll be one fridge with multiple QPUs inside it. And as the technology progresses, and as you say, you know, maybe one day we'll have one of these on our desk, is it likely to be a similar, you know, it's going to be the same approach to this? Are we going to have a super cooling device on our desk that's operating our quantum computer, do you imagine?

Nathan Phillips (14:47.308) You're asking me to predict the future? I'm gonna guess... No. I'm gonna guess that in 50 years time we'll either have invented a room temperature superconductor or found a way of using...

Paul (14:49.384) Hmm.

Paul (14:53.694) that in 50 years.

Nathan Phillips (15:02.616) Yeah, I don't think there's going to be a, a, a, cryogenic fridge anytime soon, I'm afraid we'll...

Paul (15:09.857) Yeah, that's a shame. Exactly. I could retire the, yeah, exactly the drinks fridge. Okay, let's talk about what we can do with these things. As I understand it, quantum computers are at the moment mainly used, as you said earlier, in an experimentation way. They're in research labs rather than large scale practical application. what's...

Nathan Phillips (15:11.672) You want it to make some ice cream.

Nathan Phillips (15:20.074) it

Paul (15:36.819) What needs to happen before they become commercially useful? Why don't we start there?

Nathan Phillips (15:43.394) You're right, we are not there yet in some systems that can actually solve commercial problems.

The number one thing that we really are looking to solve at the moment is improving performance by solving the problem of errors in the logical gates that we create, the quantum logical gates, and that's the error correction or error mitigation. And there's some big companies dedicating a lot of resources to that problem. Google have made some interesting discoveries recently. There's been quite a lot of encouraging signs on that. There's another major problem, which is scalability. So that's both in terms of making bigger chips with more qubits per QBE, but also in terms of just making

Paul (15:59.359) and that's the error correction, error mitigation. And there's some big companies dedicating a lot of resources to that problem. people made some interesting discoveries recently. There's been quite a lot of encouraging signs on that. There's another major problem, which is scalability.

So that's both in terms of making bigger chips with more qubits per QBE, but also in terms of just making more chips, more consistently, more reliable chips in batches. And those are problems that we're likely to have to keep chipping away at over many years. That's not something that's going to suddenly be solved with one sudden discovery.

Nathan Phillips (16:19.116) more chips, more consistently, more reliable chips in vatures. And those are problems that we're likely to have to keep chipping away at over many years. That's not something that's going to suddenly be solved with one sudden discovery. And those two things in some ways work together. The more errors that we have, the more qubits we need to implement a logical gates because we use extra qubits to compensate for that. And so for some real world business cases, I think we're going to need tens of millions of qubits. And at the moment, a large QPU is only going to

Paul (16:32.318) And those two things in some ways work together. The more errors that we have, the more qubits we need to influence the logic of against because we use extra qubits to compensate for that. And some real world business cases, I think we're going to tens of millions of qubits. And at the moment, a large qubit has only got about thousand. So scalability would definitely be some way off. Are there companies using it?

Nathan Phillips (16:49.046) a thousand. So scalability would definitely some way off actually usable systems.

Paul (16:58.663) today, are they using this technology today and what sort of applications are they currently at least experimenting with?

Nathan Phillips (17:05.271) So I think the world of quantum computing has opened up a lot of opportunities. There are a couple of ways that companies are preparing for quantum readiness, even though we're not there yet. One of those is they're developing IP ready for when the computers are there to enable them to run those things. And so they are...

Paul (17:14.685) that come visit.

quantum readiness, even though we're not there yet. One of those is they're developing IP ready for when your computers are there to enable them to run those things. And so they are investing in quantum computers in order to develop algorithms.

Nathan Phillips (17:31.978) investing in quantum computers in order to develop algorithms or applications, even though the machines needed to fully run those applications are not yet available, they're using these quantum machines to prove the viability of those algorithms. So they've developed the IP and own that IP, that mean they'll be at the front of the line for exploiting those machines when they do become available. There are other companies that maybe don't have the resources to throw out a team that can start developing those algorithms, but they still want to be up to date on what's going on. And so they're

Paul (17:37.06) even though the machines needed to put in one those applications when it's available, they're using those quantum machines to prove the viability of those algorithms. So they convert the IP and own that IP, that means they're being the front of the line for exploiting those machines that you did not available. There are other companies that...

Maybe don't have resources to throw out a team that can start developing those algorithms, but they still want to be up to date on what's going on. And so they're investing in quantum computers or access to quantum computers in order to make sure they're training their staff to be aware of what quantum computers are and how to use them and what sort of applications might be useful to them. And as a side benefit, those companies that are investing in developing their ID are actually, they are getting some return on investment there because they're developing quantum-inspired

Nathan Phillips (18:01.982) investing in quantum computers or access to quantum computers in order to make sure they're trained, they're training their staff to be aware what computers are, how to use them and what sort of applications might be useful to them. And as a side benefit, those companies that are investing in developing their IP are actually, they are getting some return on investment now because they're developing quantum inspired algorithms. So these are ways of solving problems that can run on the classical hardware we have today. But by thinking about a problem,

Paul (18:25.135) algorithms. So these are ways of solving problems that can run on the fast forward hardware we have today. But by thinking about a problem in the way that you would need to be able to solve it on a quantum device, they're coming up with more efficient solutions that can run on classical computers. It's not as good as treatment in computing, but it's a way that people are getting some ROI now even while we're the intermediate stages of the development of quantum computing. Right, okay. So much like with artificial intelligence, we're learning through our application of it how

Nathan Phillips (18:31.802) in the way that you would need to in order to solve it on a quantum device, they're coming up with more efficient solutions that can run on classical computers. It's not as good as true quantum computing, but it's a way that people are getting some ROI now, even while we're in the intermediate stages of the development of quantum computing.

Paul (18:55.088) to use this in a better way already? Yes. What would you say are some of the impacts that you're seeing on some perhaps of the known use cases around say drug discovery where there's lots of very complex data sets to go through finance, cryptography and so on. What are you seeing how quantum is going to impact some of those industries?

Nathan Phillips (18:58.506) Yeah. Yeah.

Nathan Phillips (19:25.676) Yeah, was in a meeting with some representatives from the FCA last week. They will say they want to be ahead of the curve, they don't want to be a blocker. So that when things do become available, suddenly, if we start thinking about regulation, how it impacts on those sort of things, they're already trying to work out now how they're going to work with quantum algorithms when those are available to be run on real computers. But you asked first about drug discovery. And on that one, if you want to model a molecule, which is what you're doing in drug discovery,

Paul (19:31.197) they will say they want to be ahead of the curve, or they don't want to be a blocker. So that's when things do become available, suddenly we think start thinking about regulation, how it impacts on those sort of things. So they're already trying to work out now how they're going to work with quantum algorithms when those are available to run with computers. But you asked first about drug discovery.

If you want to model a molecule, which is what you're doing in drug discovery, then you are trying to simulate an inherently quantum system. Molecules are quantum systems. And so doing that on a classical computer, it's an exponential problem. It's a problem where as soon as you make anything not trivial, it's at a very long time. So a quantum computer is going to make a

Nathan Phillips (19:55.702) then you are trying to simulate an inherently quantum system. Molecules are quantum systems and so doing that on a classical computer is an exponential problem. It's a problem where as soon as you make anything non-attribute it takes a very long time. So quantum computing is going to make a massive difference to job discovery if it can do that sort of thing. Finance is a different market. They're often just trying to throw computing power at the moment to optimization problems. So that's where they'll see benefit from quantum computers is they're thinking about

Paul (20:13.02) massive difference to drug discovery, and if it can do that sort of thing. Finance is a different market, but often just runs through computing power at the moment, to optimization problems. So that's where the benefit for most computers is, but thinking about things like portfolio optimization. But we're just investigating use cases, and there's no definite kind of this is how it's going to definitely improve things. We are still in that early days of discoveries at the time.

Nathan Phillips (20:25.536) things like portfolio optimization. But yeah, we're just investigating use cases and there's definite kind of this is how it's going to definitely improve things. We are still in that early days of discoveries at the time. Quantum in AI, I think you mentioned, that is interesting. That's very much an area of hot exploration at the moment. We know some possible quantum applications that would help with machine learning or

Paul (20:42.926) I think you mentioned that is interesting. That's pretty much an area of exploration at the moment. know some possible points of applications that would come with machine learning on neural networks. But we expect to see a lot of growth in that application area. It's obviously a lot of people make money into that. The one I do want to click into is encryption technology. The sort of scare story that I hear is that when we do start to have

Nathan Phillips (20:55.456) neural networks but we expect to a lot of growth in that application area because obviously there's a lot of people making money into that at the moment.

Nathan Phillips (21:07.062) Hmm.

Paul (21:12.795) accessibility to quantum computing, then all of the current encryption models are essentially worthless. A quantum computer could work through the best encryption we have at the moment in a matter of seconds, and therefore you're banking your personal data, the encryption that's used to...

behind the military between transport is basically defunct. Is that true? Or to what extent is that true? And should we be worried about it now?

Nathan Phillips (21:46.754) Bye!

Don't panic yet because the QPUs we have at the moment are still too small to crack than the RSA they were using every day. But governments are taking this seriously. So they're investing in quantum ready encryption and there are a few quantum resistance algorithms already become available. If you are transmitting information that you want to be secure in 10 years time or 20 years time, then you should be using those algorithms now so you can find those through the US government's website. They did a call.

Paul (21:50.843) PPs we have at the moment are still too small to crack. The MSA, they're usually every day. But governments are taking this seriously. So they're investing in quantum radio encryption. And there are a few quantum resistance algorithms already become available.

If you are transmitting information that you want to be secure in 10 years time or 20 years time, then you should be using those. So you can find those through the US government's website. get a call quite a while ago now where they've actually thrown

Nathan Phillips (22:17.178) quite a while ago now where they've actually come up with some algorithms that work in a different way that won't be susceptible, but we hope so far won't be susceptible to quantum cracking. But in terms of RSA and the things that we use currently for our banking, yes, I mean, that's not going to survive proper quantum readiness. And so we will need to change things on the way we do that by then. In some ways, that's probably not the scariest area though.

Paul (22:21.435) in a different way that won't be susceptible, but we hope so far it won't be susceptible to points of cracking. But in terms of RSA and other certain things that we use commonly for our banking, yes, mean, that's not going to survive proper.

and so we will need to change things. The way we do that by then, in some ways that's probably not the scariest area. Some people are saying that the underlying encryption behind Bitcoin won't survive quantum readiness and that's going to cause some disruption I think.

Nathan Phillips (22:47.098) saying that the underlying encryption behind Bitcoin won't survive quantum readiness, that's going to cause some disruption I think. But yeah, think the thing to be aware of is just what's the lifetime of your data. So if you're going with banking online, does it matter if somebody gets hold of it and in 10 years time they can read what you did today? What we hear, and I don't know if it's true but it'd be strange in some ways it wasn't, is that governments

Paul (22:58.747) Yeah, I think the thing to be aware of is just what's the lifetime if you're on data. So if you're doing with back-end online, does it matter if somebody gets all of it in 10 years time they can read what you did today? What we hear, and I don't know if it's true, it would be strange in some ways it wasn't, is that governments, again, I've heard possibly the Chinese, by the way.

Nathan Phillips (23:15.22) Again, I've heard possibly the Chinese, although I don't want to offend anyone, and probably I'm sure others are doing this as well without us knowing. They're scraping encrypted information being passed over the internet now. Even though they can't read it, they're storing it because they believe that they will be able to read it once computers have scaled up. And so they're targeting anything they think will still be useful and intelligent to the 10-year timescale, which tells you something about when they expect to have that sort of technology available.

Paul (23:19.033) Wow. Yes, I think

It's one of those big unknowns, isn't it? How a bit like the artificial intelligence world, that's it seems to be changing on a weekly basis, this technology and being prepared for that. I don't know how you prepare for it. You know, it's a wicked problem in and of itself, I suppose. Would you put would you you hazard a guess as a time frame as to when quantum is going to start to become more mainstream? Are there some?

Nathan Phillips (24:16.152) No.

Paul (24:16.866) specific milestones that you're seeing or that you see that we will need to have got through in order to get there.

Nathan Phillips (24:26.136) As to how long it's going to be, I will tell you what other people are working on, guess. That's the best I can do. But, I mean, we don't need to panic, so...

It wasn't that long ago that we upgraded all of the encryption on the internet from TLS 1 to TLS 2. And it takes two to three years to do that just because of the length of the time that certificates are valid for on the web. We could do the same again. It will be a bigger change. It won't just be a step change. It will be a completely different way of doing things. But it is possible. We can upgrade the web in two to three years. And it is going to be more than two to three years before quantum computing causes problems in that sort of way.

Paul (24:42.297) And it takes two to three years to do that just because of the length of time the certificates are valid for the web. We could do the same again. It will be a bigger change. It won't just be a step change. will completely different thing. But it will, it is possible. We can upgrade the web in two to three years. And it is going to be more than two to three years. Four months computing causes problems in that sort of way. So.

Nathan Phillips (25:06.904) So, like I said, you've got probably at least 10 years, given sort of timescales that some people are talking about. Others say 20 years, some people have the difference and say 15 years, but I wouldn't like to say a number.

Paul (25:10.746) Like I said, you've got probably at least 10 years, given sort of timescales that some people are about. Others say 20 years, some people have half the difference, they say 15 years, but I wouldn't say it's the same. If I want to get started today, if I want my own quantum computing, Nathan, can I rent one? I wanted to ask you that. Can I rent space in a quantum computer?

Nathan Phillips (25:34.968) The easiest way, so we've got various levels, you could...

simulate quantum computers. As long as you only want to do one or two qubits or maybe a handful of qubits, you could actually simulate a quantum computer on a classical computer. Like I said, it's an exponentially hard problem, but actually it is a problem that can be simulated on a small scale on a classical computer. So if you want to just do something very basic and just see what it would look like, you can simulate it and you can get those simulators online. If you want to use a real quantum computer without any upfront investment, then a lot of the big clouds

Paul (25:45.753) you could actually simulate what's going be done in a classical computer. Like I said, it's an exponentially hard problem, but actually it is a problem that can be simulated on a small scale on a classical computer. So if you want to just do something very basic and just see what it would look like, you can simulate it and you can get those simulators online. If you want to use a real computer without any upfront investments, then all of it.

a lot of the big cloud services, AWS, your cloud service for brackets, Microsoft this year, one's computing about it as well.

Nathan Phillips (26:10.222) services, so AWS, do a service called Brekket, and Microsoft Azure have got one's computing available as well. People are struggling with those, partly because of the amount of downtime that those systems are having, but it is an affordable way of getting yourself in without upfront investment. And then beyond that, you are either looking at partnering with a university or leasing. There are some places where you can lease them.

Paul (26:19.094) People are struggling with those, partly because of the amount of dams that those systems are having. But it is an affordable way of getting yourself in with that upfront investment. And then beyond that, you are either looking at partnering with a university or leasing. There are some places where you can lease them or buy from them.

Nathan Phillips (26:39.288) or buying one and if you're buying them you can spend anything between 12 million and 120 million depending on where you get it from.

Paul (26:44.664) 12 million and 120 million depending on where you get it from. How long does it take to build your own quantum computer from, I give you an order tomorrow, when could I expect it to be delivered and ready and working for me?

Nathan Phillips (26:54.253) Thank

Nathan Phillips (27:02.092) Yeah, so as I said, our team has got quite a lot of experience in building quantum computers. And so we're looking for our first couple of machines, probably should probably check this with somebody before promising you a quantum computer in six months. But let's say that for you Paul, I'll do your quantum computer in six months. And I think if you were to ask me again in two years, I'd hope that we go into three months. Now,

Paul (27:03.02) Yeah, so as I said, our team.

Paul (27:09.154) So we're in the game for our first couple of machines.

Paul (27:21.761) Accelerated. Fantastic. As long as I've got 120 million to hand.

Paul (27:30.338) Wow.

Nathan Phillips (27:31.99) That's partly just down to supply chain. So like I said, in the past, these things tend to be built as one-off things, whereas we're talking more about a manufacturing way where you keep stock of various components. So there's lead times on lot of these components that are in the many months sort of area. So if you're starting from nothing and you want to build one, it's gonna take you...

I would guess nine to 18 months. But if you're ordering from someone like Trek, then you'd be hoping that we're not starting from nothing, but that we already have.

Paul (27:58.136) we'd guess nine to eighteen months but if you're inputting someone like Trek then you're hoping that they're not sniping nothing that we already have. The components and the process.

Nathan Phillips (28:08.792) Yeah. Do make sure that I check that answer before you publish it.

Paul (28:11.202) Yeah, fascinating.

Nathan Phillips (28:17.944) You said six months?

Paul (28:19.887) Unfortunately, I don't have the means to place my order tomorrow. So you have got time just to double check that.

Nathan Phillips (28:26.284) I don't get a few, I've got seven months in, brilliant.

Paul (28:30.711) A few, we're getting close to our time. A few closing questions for you. What's one question about your industry you wish people asked more often?

Nathan Phillips (28:41.961) Mm.

think that the biggest conception about quantum computers is that they're just extraordinarily powerful, extraordinarily fast computers, whereas in reality they're useful because they solve one or several small specific classes of problems that are just intractable with classical computers. So the intelligent question which people would ask or people could ask would be what sorts of problems do quantum computers solve rather than just assuming it's the same things we currently use computers for? I think you've actually done quite well with your questions, you've captured that.

Paul (28:52.019) in reality they're useful because they solve one or several small specific classes of problems that are just attractive to classical computers. So the intelligent question which people would ask or people could ask...

would be what sort of problems do quantum computers solve rather than assuming it's the same thing to become these computers. Well, I think you've actually got quite a world of good questions and captured that pretty well, but quantum computers in that they kind of run Microsoft Word for you. It's just different sorts of computers. Yeah. And it sounds like my takeaway is that they're not going to just replace, at least not in the short term, the traditional computer. They'll work alongside if anything, or for specialist tasks.

Nathan Phillips (29:14.538) pretty well, but quantum computers are never going to run Microsoft Word for you. This is different sorts of computing.

Nathan Phillips (29:31.468) Think of them.

Yeah, think of it. You know what a GPU is, a graphical processing unit. These are things we've had for a long time. We built them to use salt to accelerate 3D graphics on our computers. The main thing they use for now is AI. So Nvidia are the big makers of GPUs. Think of a quantum computer, not as a quantum computer, but as a quantum accelerator, like a graphical accelerator. They don't replace a CPU. They work alongside a CPU. So your CPU and your QPU

Paul (29:51.767) Cool.

Paul (29:57.793) they don't replace the CPU, they work alongside the CPU. So the CPU and the QPU will together solve the problem.

Nathan Phillips (30:02.874) will together solve a problem. some of the problem will be offloaded onto the GPU, some of the problem will be offloaded onto the QPU, but it's the old traditional CPU that will orchestrate those things together and actually be your interface into that world. And that's why I've got a job, because I come from classical computing background, the non-computing computing background. I couldn't write a quantum algorithm if you gave me...

Paul (30:06.231) Some of the problems we are working on is the GPU, some of the we are working on is the QPU, but it's the old traditional CPU that will orchestrate these things together and actually be your interspaces in world. And that's why I put the job, because I come from classical computing background, but not advanced computing background. I couldn't write a quarter of an with an HDKB, but it gave me a week, I think it would be good. The rest of the day, I'll be good. But I come from classical computing background.

Nathan Phillips (30:26.456) Well, if you gave me a week, I'd probably be good. But if you gave me the rest of the day, I probably couldn't. But I come from a classical computing background. I write the sort of software that we've been writing for years now. And in that world, we're still going to need that. We're still going to need that in order to manage these machines. They're still physical machines. There's magic in the bottom, but the rest of it is all the sort of physics where you can punch it with your hand and it still hurts. It's in one state and it's got a control hardware around it that needs...

Paul (30:35.639) It's sort of software that we've been writing for years now. And in that world, we're still going to that. We're still going to need that in order to manage these machines. They're still physical machines. There's magic in the bottom, but the rest of it is all sort of physics where you can punch a video hand and it still hurts. It's in one state, and it's got a control hardware around it and needs classical software to drive it.

Nathan Phillips (30:56.42) Classical software to drive.

Paul (31:00.446) Yeah, and I love your journey into this. As you said, when we first met that you wouldn't describe yourself as a quantum physicist originally, you're as you say, your path has been into this through your own expertise. It's clearly an exciting and going to be growing industry. What advice would you give to someone who wanted to work in quantum computing? What should they be studying? How do they get into it? What's your advice?

Nathan Phillips (31:24.3) That is an interesting question. I mean, to be honest, at the moment, the industry is small. It does feel like there aren't many people in it who don't have a physics PhD. So if you don't and you want to work in it, then be prepared to learn fast and get used to being out of your depth all the time, which I quite enjoy. But once that's true, and the more you can learn before you make the jump, better off you'll be. Pontum computing governance is inherently multidisciplinary. Only the very small bits at the bottom is the actual pontum bits.

Paul (31:52.63) So it's always at the bottom is the actual quality of the data. It requires many skills plus technical, organisational, business disciplines. So don't feel you need physics, PhDs, industry. Even on the technical side, there's plenty you need for the skills in material science, microwave engineering, FPGA, firmware development, software development, etc.

Nathan Phillips (31:54.186) requires many skills across technical, organisational, business disciplines. So don't feel you need a physics PhD in the industry. Even on the technical side, there's plenty of need for people with skills in materials science, microwave engineering, FPGA, firmware development, software development, etc.

So yes, need to, the more, the more quantum background you can give yourself, the better off you'll be. But don't be scared off if you don't have that already. Perhaps that's a good way of saying it, yeah.

Paul (32:15.03) So yes, you need to do the more quantum background you can give yourself the better off you'll be. don't be scared, ultimately you can go. Open to curious minds by the sounds of it as an industry. One final tricky question for you. How do you think quantum computing is going to change our world in a sentence or two? How do you think it's going to change our world?

Nathan Phillips (32:39.722) spoken about every week. I think if Bitcoin gets broken, we're going to hear about that. I think the things that are going to make more of a difference probably will never be seen by the general public. So accelerating drug development and pharmaceutical, you know, we don't see that behind the scenes stuff. What we're going to see is things like when Bitcoin gets broken.

Paul (32:57.269) we don't see that behind the scenes stuff. We're walking out of cities and like, and then Bitcoin gets broken.

Paul (33:04.991) They think it's been a brilliant conversation. I've really enjoyed learning about quantum from you. Thanks so much for joining us. It's been a real pleasure to hear your story and your insight and all about Trek. Where do we find out more about Trek? The website is trek.tech.

Nathan Phillips (33:22.378) Absolutely. T-R-E-Q, Q, every quantum computing has a Q in its name. T-R-E-Q dot T-E-C-H.

Paul (33:30.271) Brilliant. Well, we really look forward to seeing you guys more in our future. And thank you very much again.

Nathan Phillips (33:37.816) It's been an absolute pleasure. Hope you can do this again sometime.

Paul (33:41.344) That wraps up this episode of Digital Humans podcast. If you enjoy the conversation, don't forget to subscribe on your favorite podcast platform so you never miss an episode. Thanks for listening and we'll see you next time.