Podcasts

Podcast with Rene Schulte, Valorem Reply

17
November
,
2021

My guest today is Rene Schulte, research director at Valorem Reply. Amongst other topics, Rene described several customers that are taking quantum computing solutions into production environments. These are customers in energy, transportation and telecommunications.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval Boger (Classiq): Hello, Rene. And thanks for joining me today.

Rene Schulte (Valorem Reply): Hi, Yuval. Thanks for having me. It's fantastic to be here.

Yuval: That's wonderful. Rene, who are you and what do you do?

Rene: Well, first, my name is Rene Schulte. My role is director of global innovation at a company called Valorem Reply. And what I do there is basically research and development around emerging technology with the focus on, typically, I say three key areas and the first one being spatial computing. Things like mixed reality, augmented virtual reality, but also things like 3D sensors, LiDAR, and that kind of stuff. Like all things that are related to spatial computing and spatially understanding the world around us. And that also comes with the second, which is computer vision, AI, that kind of stuff. And also robotics. And last but not least, of course, quantum computing. And I've been working in that field since two or three years on the applied quantum computing, I would say, and the impact we can all achieve already today. That is the main focus. Less on the scientific research, but more on what can we do already - how can we use applied quantum computing to provide benefit to our clients? 

I'm also a Microsoft Regional Director and MVP. These are awards that Microsoft provides to, like they say, community leaders, that share their knowledge. I do a lot of conference talks, a lot of virtual conferences these days, but also one in-person coming up this week, actually, which I'm super excited about and I'm also on social media and constantly sharing knowledge basically. Doing a lot of things, I also have my own little quantum computing video series. So think about like a podcast, but you can actually see the people while they're talking, which I called QuBites. Not qubits, but QuBites with another E. And so I call it QuBites: Bite-sized Pieces of Quantum Computing. That's what I do. Basically always having my ear on the industry policy in all of these areas, very technical, but also I understand the strategy and business all around. And yeah, I'm always passionate about all of these topics.

Yuval: And in terms of geography, I think Reply is a global business, but do you focus primarily on Europe or other areas? 

Rene: Valorem Reply is a US company. I actually live in Germany working for US companies since almost a decade now. But actually Valorem Reply is part of the Reply group, which is headquartered in Italy and has a strong presence in Europe, in Italy and Germany, all over the place. But like I said, Reply really has offices all over the place, in the US, in the UK, in Germany and Italy all over the place. And I work a lot together with my colleagues, for example, from Data Reply when it comes to quantum computing and the Machine Learning Reply. But also at Valorem Reply, we do a lot of interesting stuff in that space. Yeah.

Yuval: Excellent.

Rene: Basically, what we do is professional services consulting but not just on a strategy level. We can start on a strategy, but then we also support the clients throughout the whole development as well. And so we sometimes do our own product development, but oftentimes it's custom development for clients and it contains all different vendors. It's not just Microsoft, but all of them are part of the, basically, what we developed with and vendor independent, but also lot of different technologies. Reply is around 10,000 people split up into multiple sub companies basically. But as you can imagine, just from the size, it's basically a general professional services, all are related to software, but also a little bit of hardware.

Yuval: When you focus on quantum clients, do you see that in a particular industry, a particular application, is it pure quantum or is it more of the quantum-inspired optimizations and so on? Tell me a little bit about what you're seeing in the market in terms of where quantum is taking hold.

Rene: Absolutely. And so we don't see a particular industry. Of course, there are certain industries that are well suited at the moment. I would say everything that has a strong non-linear challenge, right? Which could be drug discovery, chemical processing in general. If we think about how chemistry actually works, really works, it's quantum chemistry, right? And so using quantum computing to actually be able to simulate how the chemistry, how our universe is really built is of course, very beneficial. And there was a strong advantage in these industries, of course. But it's not just particular industries. We're working with automotive, we're working with telecommunications, with utilities sector as a bunch of our clients. And I can share some examples later of some of the projects we already executed and the things we see there. But first of all, I would say right now we focus on three key areas.

The first one is quantum machine learning, which basically is analyzes of classical data on quantum computers. And you can already achieve some really interesting things. And we are part of a bunch of projects and research that, which unfortunately I cannot mention because of NDA reasons, but with QML, with quantum machine learning, there's amazing stuff that can be done. For example, for image classification. And some of our experiments have shown that certain newer QML algorithms provide not just the faster results than classic machine learning, but actually also more precise, which is crazy if you think about it. It's not just faster, but also more precise. But again, if you think about how is our universe built on the smallest scale that it's all quantum mechanics, right? Applying quantum machine learning maybe also gets closer to the real reality of our analog world which is of course built on this principle of uncertainty, if you will, compared to this digital world.

Getting closer to the nature also helps us to simulate faster, but also better, which is interesting. So QML is one and another one is of course, quantum-inspired optimization, which we developed QUBO: Quadratic Unconstrained Binary Optimization. And the idea behind general quantum-inspired computing and in particular quantum-inspired optimization is using principles of quantum computing and quantum algorithms, but not having them run on quantum computers, on real quantum computers, but having them execute on classical computers. But of course, very powerful ones like think about GPU, where you can execute a lot of parallel operations or FPGA or some of those, right? Where you can really do some interesting workloads loads there. And so you can leverage quantum computing principles, implement these in algorithms and then run them on these kind of classical hardware and achieve better optimization results and faster than classical algorithms, than classical optimization algorithms and classical hardware.

And if you think about the future, once we have enough powerful quantum computers available, well, these algorithms don't have to be redone, because they are already built for quantum computers, but they will just fly if you basically execute them on a true quantum computer. And again, we can talk a little bit more about that later. I can show some use cases here where we're seeing really quite a great return of investment with quantum-inspired optimization. 

But I just want to mention also the third key area, which we see as quantum security and quantum communication. All of that is related to the quantum security fields. Right? I would say, on one hand, quantum security is a threat, as we know, right? With Shor’s  algorithm, you can crack encryption like RSA and once there's enough powerful machines available, once there are enough powerful quantum computers available, it is said that if you have a four thousand stable qubits, you can crack RSA 2048, right? Which is the current standard for a lot of secure communication. The whole internet is built on secure communication these days. And so if you can crack those well, then a lot of bad things will happen, of course. Right? And this is going to happen in a few years, for sure. You might've heard IBM announced that they are planning in two years in 2023 to have a quantum computer with 1,000 qubits. Right? Let's get in closer and with 1,000 qubits might be able to actually already crack, I don't know. I didn't do the math, but let's say RSA 2056 or so already.

Right? And so there's this big threat of the so-called Q-day is coming or in relation to the millennial bug that you might know, it's also sometimes called Y2Q day. It will happen. And so what we provide as a service, for example, consultation. But also advice for clients, what they can do already today. Cause they don't have to wait until it's delayed with the Q-day, but they can already do certain things already today. Right? This is really interesting when it comes to quantum security, because it will happen. It's just a matter of time, but folks should prepare now. And so this is where we are putting a lot of focus. But also there's, with quantum security, there's also the quantum communication, like quantum key distribution, where you can securely exchange keys, symmetric keys and that stuff.

There's also opportunities with also true random numbers and so on that you can generate with quantum computing. And all these three areas like quantum machine learning, quantum-inspired optimization, quantum security, where we see the biggest impact when we work with all clients, where we have achievable impact already today. And even if the infancy of the quantum computing hardware, especially quantum-inspired computing, is running these algorithms faster than classical hardware. That's I think really impressive that we can actually already achieve with the infancy of the hardware we can use the principles to build software with quantum-inspired computing that is outpacing classical optimization algorithms.

Yuval: Given the infancy of the hardware and given that there's a widespread interest in quantum computing, do you see the ROI today? Do you see companies actually moving quantum projects into production and getting real business value from it? Or is it more on the exploratory side, let me investigate this. Let me investigate that. And maybe in a couple of years we will move it into production.

Rene: Yeah. A lot of these projects, of course, and I would say our pilot phase, but one of our clients actually moved the quantum-inspired solution into production already. I cannot mention the name here, but it's one of the largest energy grid providers in the world. And they based maintain these energy grids, as you can imagine. Landlines and transformation stations and whatnot, and a lot of stuff. And they have like 20,000 field service workers, like 20,000 people that every day have a certain schedule where they need to drive with the car or whatever way of transportation they use. But basically go to these certain stops. They need to perform an action. I don't know, repair a transfer mate or whatever it is. Right? And so they have these 20,000 field service worker, which their schedule has to be planned in advance, of course.

And so as you probably know being from the computer science, that's a classical NP complex problem, which is called the TSP, the traveling salesman problem, right? Or a scheduling optimization problem. Right? The more stops you add, the more exponential will grow, right? The more complex it gets because what the optimization goal you want to achieve with that is, of course, you want to maximize the time that the field service worker spend working but you want to minimize at the same time, the time they spend on the road, traveling basically. Right? We want to find the optimal schedule. And long story short, for this client, we introduced our Mega QUBO Solution is quadratic unconstrained binary optimization algorithm. And this can be useful as workforce management, right? And so in a few minutes, the optimization that we developed, can identify a schedule that maximizes the amount of time spent working while minimizing the time spent on the road.

What they achieve is pretty impressive. They achieve a 20% reduction of travel time for 20,000 field service workers, right? And this is an execution time speed of 18 times on a GPU cluster. And so that is I think some pretty telling numbers. This is a lot of time and money saved there. And this is done, again, with quantum-inspired optimization, with our Mega QUBO Solution where you can put in certain things and then it gives you the optimal schedule as an output. And this is one example. I have a few examples where we're working with clients is like a large telecom operator. And again, so the one I just mentioned, the case, this is deployed in production already, right? This is running and it's optimizing the schedule and it gives a lot of ROI. If you can do the math yourself, like 20% reduction in travel time for 20,000 field service workers.

That's big money, of course. Right? But we're also are involved in a bunch of challenges like the Airbus Quantum Computing Challenge. We took part in that and a team of Machine Learning at Reply actually was winning that. The Machine Learning Reply folks won the Airbus Quantum Computing Challenge with a solution. Again, awesome optimization solver and optimizer basically. Well that optimize the loading of a plane, as you can imagine, that is a super complex problems with, I don't know, hundreds of variables maybe. I don't know, you have to sample the mass of the plane and you have all these physical constraints. You cannot go over a certain mass in total. You have to distribute the weight in a certain way. You want to minimize the fuel that has to be used. You want to minimize the time the plane is on the ground and all of that.

There is a bunch of things, if you want to load a plane optimally, that is very complex and challenging. And so we developed this solution and won the Airbus Quantum Computing Challenge, for example. Another example is for a telecom operator, we did a pilot project there where we were using again, optimization solver, Mega QUBO, to optimize the distribution of, basically, the frequency range. As you can imagine, think about a 5G network, like an antenna, the cell tower, if you will, and so, of course, this cell tower can only serve a certain amount of users, right? At some point it's maxed out and so you want to optimize. And so basically they have this range and they want to optimize. It's a little bit fragmented, right? They want to have as many users as possible. And so you need to basically pack them in an optimal way. And this is also showing some really promising speed ups when it comes to using quantum-inspired optimization for this. Right?

Yuval: To grow your business, the quantum side of your business, obviously, you want more customers to move into production and more customers to show positive ROI. Other than stronger computers, what do you think is missing? What is needed to accelerate that deployment of quantum solutions?

Rene: Well, you can split it in multiple areas, I would say. But if we're thinking about, from a technical standpoint, I think there is a need also to have this middle layer on also the programming language available that can be used, not just with one particular quantum computer. Because you need to imagine, these quantum computers are in very early stage, of course. Right? Think about the mainframe area in the the 20th century. In the middle of the 20th century, we had this mainframe area. Right. And so this is probably comparable to the state of quantum computers today. Of course, it's not 100% comparable, but you get the idea, right? It's huge things and they have all specific ways to program them, right?

You actually have to talk to them, to these quantum computers and program them and with a machine language. It's not even comparable to assemble or for classic computers, but really, you need to basically use a language that is not a high-level programming language. And also you're very much dependent on that certain quantum computer. And so the challenge for the development, I think that is being solved at the moment, were things like, for example, Q#, for example. Or although from IBM, they also have some stuff. And then a couple of other companies, of course, but basically you have a high-level programming language, which you can write your quantum algorithms in. And then you have the middle layer, basically, an interpreter or like for Q# it's the QIR, the quantum intermediate representation, which is a representation of your code that can then with a machine specific runtime can run on any of those.

Basically you have a high-level programming language. You have one language and you can target whatever kind of quantum computers are out there, as long as each of them provides a run time that can talk with your QIR basically. And so long story short, what I think is missing is, but I think it's being solved, is this generic and more approachable programming style and programming languages for quantum computers. But beside the technical things, of course. There's also, in general, clients might be hesitant cause they might still think, oh yeah, I heard about quantum computing, but this is a thing that might be happening in 10 years or so. We don't care about it at the moment. Right? That's what you hear oftentimes. And that's why I'm also with my QuBites video series, I try to focus on impact today because there is already achievable ROI with certain solutions.

And I think we need to show more cases where you can see the real benefit that can be achieved already today. Right? But of course, we've also need a more powerful quantum computing hardware. And another thing, which I'm also great that you, for example, Yuval have the podcast here and I'm doing my video series and I know there was a couple of other folks that also do podcasts or video series, is the education, right? And knowledge sharing in that space and getting quantum computing out of the research world and out of this a little bit exclusive club, if you will, to be more approachable for everyone. Because you need not just folks that have a PhD in physics, but you also need people that understand program management or project management, and you need, of course, engineers, you need developers, you need marketing folks.

You need all of this, what we know from classical computing is now needed because we're getting out of the science and research stage and getting into real applied quantum computing, even if its infancy. I think what we need is more folks joining the quantum computing bandwagon, if you will, that are not coming from the pure quantum information or quantum computing background from the university, but also more diverse background and a more inclusive community. And then we also need to show more ROI that can be achieved with these quantum solutions today. And like I said, of course from the technical standpoint, which is already being solved with Q# other things is this not just targeting specific quantum computers, but having more of a generic programming language.

Yuval: I was speaking with a partner from McKinsey the other day, and he mentioned a term that he called Business Translator. Someone who says, I can translate the technical benefits of quantum into a business benefit, into the ROI. This is not just what it does, but also why should you care, how can we deliver value today, and which projects you'll have to wait a couple of years before you can see real value. I like that term. And I think it may be actually what your company does amongst other things is help communicate the business value to the customers.

Rene: Exactly. And actually executing those. Right? We can communicate it, but we can also deliver. And that's, yeah. Important.

Yuval: My last question, you mentioned that Reply is built out of several different companies, several different units. Do you see quantum project as standalone projects or do you see them integrated into a broader enterprise software or hardware architecture? Oh, where's the data center? What's the SLA? How does data come in? How does data get out? Is it isolated or is it integrated?

Rene: Yeah, both I would say. It really depends on the clients. Also, it depends, of course, from the business development standpoint, what is the client relationship, right? Which folks are involved there, but from the typical thing, like the quantum projects themself, there are typically specific things. Cause you cannot just solve every problem with a quantum computer. That's also a misconception. A lot of folks I talk with, not the experts of course, but a lot of folks that are just hearing about quantum computing, there's a little bit of this misconception that quantum computers will be these generic computers, right? And while I'm holding up here my smartphone and showing it to you that we will have a quantum computer running in our smartphone, right?

Or a quantum computer running in our laptop. Well, this is not going to happen, at least not anytime soon, quantum computers are an addition. They're not going to replace classical computers. And they're an addition, and you can compare it with a GPU which is a specific graphics processing unit, right? Which is specific, well-built for certain tasks. That's the same thing with a quantum computer, although they are much larger than just the graphics card in your PC and work a little bit differently, quite a lot differently. But you can think about just another acceleration vehicle for specific problems, right? And so typically it's a very specific problem that needs to be identified first. What we do is our what we call the Reply roadmap to value starts with quantum-inspired optimization. As we go in with the client, we do a workshop, and then we actually select a certain scenario that is well-suited, right.

That actually benefits from the quantum computer. Because like I said, not every problem is well solvable with a quantum computer in a sense that you get actually benefit out of it. Right? And then we identified this problem and then we introduced assault and quantum computing algorithm. It could be QUBO. It could be some other things like there was also Monte-Carlo and a bunch of other, basically quantum-inspired optimization algorithms, that you can use. And it depends on the specific scenario, which is the best one. We select a scenario, introduce the optimal quantum computing algorithm. And then in the short term, we can run these developed quantum-inspired optimization solver on a classical hardware, like a GPU array and then long-term, you can take the exact same algorithm that was developed and put them on a quantum computer and well, on this one, it will just fly, right? It will be even faster, but that's typically how we approach these projects.

Yuval: Rene, that was a wealth of information. How can people get in touch with you to learn more about your work?

Rene: Yeah, well, you can find me on most social media. For example, LinkedIn, just search for my name, Rene Schulte. Feel free to connect also on Twitter. You can find me. My handle is @RSCHU. And I keep constantly sharing lots of stuff on social media. You might want to connect there. And then also my little video series, you can find it called QuBites, like just Qubit, but just with another E and an S at the end. A little bite to eat. So QuBites, you can find that at thereply.com site. And also if you just search with your favorite search engine, you will probably find it. And yeah, happy to connect and thanks for these great questions by the way, Yuval.

Yuval: Well, thank you very much for joining me today. Have a good day.

Rene: Thank you so much for having me. And it was a pleasure to be here.


My guest today is Rene Schulte, research director at Valorem Reply. Amongst other topics, Rene described several customers that are taking quantum computing solutions into production environments. These are customers in energy, transportation and telecommunications.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval Boger (Classiq): Hello, Rene. And thanks for joining me today.

Rene Schulte (Valorem Reply): Hi, Yuval. Thanks for having me. It's fantastic to be here.

Yuval: That's wonderful. Rene, who are you and what do you do?

Rene: Well, first, my name is Rene Schulte. My role is director of global innovation at a company called Valorem Reply. And what I do there is basically research and development around emerging technology with the focus on, typically, I say three key areas and the first one being spatial computing. Things like mixed reality, augmented virtual reality, but also things like 3D sensors, LiDAR, and that kind of stuff. Like all things that are related to spatial computing and spatially understanding the world around us. And that also comes with the second, which is computer vision, AI, that kind of stuff. And also robotics. And last but not least, of course, quantum computing. And I've been working in that field since two or three years on the applied quantum computing, I would say, and the impact we can all achieve already today. That is the main focus. Less on the scientific research, but more on what can we do already - how can we use applied quantum computing to provide benefit to our clients? 

I'm also a Microsoft Regional Director and MVP. These are awards that Microsoft provides to, like they say, community leaders, that share their knowledge. I do a lot of conference talks, a lot of virtual conferences these days, but also one in-person coming up this week, actually, which I'm super excited about and I'm also on social media and constantly sharing knowledge basically. Doing a lot of things, I also have my own little quantum computing video series. So think about like a podcast, but you can actually see the people while they're talking, which I called QuBites. Not qubits, but QuBites with another E. And so I call it QuBites: Bite-sized Pieces of Quantum Computing. That's what I do. Basically always having my ear on the industry policy in all of these areas, very technical, but also I understand the strategy and business all around. And yeah, I'm always passionate about all of these topics.

Yuval: And in terms of geography, I think Reply is a global business, but do you focus primarily on Europe or other areas? 

Rene: Valorem Reply is a US company. I actually live in Germany working for US companies since almost a decade now. But actually Valorem Reply is part of the Reply group, which is headquartered in Italy and has a strong presence in Europe, in Italy and Germany, all over the place. But like I said, Reply really has offices all over the place, in the US, in the UK, in Germany and Italy all over the place. And I work a lot together with my colleagues, for example, from Data Reply when it comes to quantum computing and the Machine Learning Reply. But also at Valorem Reply, we do a lot of interesting stuff in that space. Yeah.

Yuval: Excellent.

Rene: Basically, what we do is professional services consulting but not just on a strategy level. We can start on a strategy, but then we also support the clients throughout the whole development as well. And so we sometimes do our own product development, but oftentimes it's custom development for clients and it contains all different vendors. It's not just Microsoft, but all of them are part of the, basically, what we developed with and vendor independent, but also lot of different technologies. Reply is around 10,000 people split up into multiple sub companies basically. But as you can imagine, just from the size, it's basically a general professional services, all are related to software, but also a little bit of hardware.

Yuval: When you focus on quantum clients, do you see that in a particular industry, a particular application, is it pure quantum or is it more of the quantum-inspired optimizations and so on? Tell me a little bit about what you're seeing in the market in terms of where quantum is taking hold.

Rene: Absolutely. And so we don't see a particular industry. Of course, there are certain industries that are well suited at the moment. I would say everything that has a strong non-linear challenge, right? Which could be drug discovery, chemical processing in general. If we think about how chemistry actually works, really works, it's quantum chemistry, right? And so using quantum computing to actually be able to simulate how the chemistry, how our universe is really built is of course, very beneficial. And there was a strong advantage in these industries, of course. But it's not just particular industries. We're working with automotive, we're working with telecommunications, with utilities sector as a bunch of our clients. And I can share some examples later of some of the projects we already executed and the things we see there. But first of all, I would say right now we focus on three key areas.

The first one is quantum machine learning, which basically is analyzes of classical data on quantum computers. And you can already achieve some really interesting things. And we are part of a bunch of projects and research that, which unfortunately I cannot mention because of NDA reasons, but with QML, with quantum machine learning, there's amazing stuff that can be done. For example, for image classification. And some of our experiments have shown that certain newer QML algorithms provide not just the faster results than classic machine learning, but actually also more precise, which is crazy if you think about it. It's not just faster, but also more precise. But again, if you think about how is our universe built on the smallest scale that it's all quantum mechanics, right? Applying quantum machine learning maybe also gets closer to the real reality of our analog world which is of course built on this principle of uncertainty, if you will, compared to this digital world.

Getting closer to the nature also helps us to simulate faster, but also better, which is interesting. So QML is one and another one is of course, quantum-inspired optimization, which we developed QUBO: Quadratic Unconstrained Binary Optimization. And the idea behind general quantum-inspired computing and in particular quantum-inspired optimization is using principles of quantum computing and quantum algorithms, but not having them run on quantum computers, on real quantum computers, but having them execute on classical computers. But of course, very powerful ones like think about GPU, where you can execute a lot of parallel operations or FPGA or some of those, right? Where you can really do some interesting workloads loads there. And so you can leverage quantum computing principles, implement these in algorithms and then run them on these kind of classical hardware and achieve better optimization results and faster than classical algorithms, than classical optimization algorithms and classical hardware.

And if you think about the future, once we have enough powerful quantum computers available, well, these algorithms don't have to be redone, because they are already built for quantum computers, but they will just fly if you basically execute them on a true quantum computer. And again, we can talk a little bit more about that later. I can show some use cases here where we're seeing really quite a great return of investment with quantum-inspired optimization. 

But I just want to mention also the third key area, which we see as quantum security and quantum communication. All of that is related to the quantum security fields. Right? I would say, on one hand, quantum security is a threat, as we know, right? With Shor’s  algorithm, you can crack encryption like RSA and once there's enough powerful machines available, once there are enough powerful quantum computers available, it is said that if you have a four thousand stable qubits, you can crack RSA 2048, right? Which is the current standard for a lot of secure communication. The whole internet is built on secure communication these days. And so if you can crack those well, then a lot of bad things will happen, of course. Right? And this is going to happen in a few years, for sure. You might've heard IBM announced that they are planning in two years in 2023 to have a quantum computer with 1,000 qubits. Right? Let's get in closer and with 1,000 qubits might be able to actually already crack, I don't know. I didn't do the math, but let's say RSA 2056 or so already.

Right? And so there's this big threat of the so-called Q-day is coming or in relation to the millennial bug that you might know, it's also sometimes called Y2Q day. It will happen. And so what we provide as a service, for example, consultation. But also advice for clients, what they can do already today. Cause they don't have to wait until it's delayed with the Q-day, but they can already do certain things already today. Right? This is really interesting when it comes to quantum security, because it will happen. It's just a matter of time, but folks should prepare now. And so this is where we are putting a lot of focus. But also there's, with quantum security, there's also the quantum communication, like quantum key distribution, where you can securely exchange keys, symmetric keys and that stuff.

There's also opportunities with also true random numbers and so on that you can generate with quantum computing. And all these three areas like quantum machine learning, quantum-inspired optimization, quantum security, where we see the biggest impact when we work with all clients, where we have achievable impact already today. And even if the infancy of the quantum computing hardware, especially quantum-inspired computing, is running these algorithms faster than classical hardware. That's I think really impressive that we can actually already achieve with the infancy of the hardware we can use the principles to build software with quantum-inspired computing that is outpacing classical optimization algorithms.

Yuval: Given the infancy of the hardware and given that there's a widespread interest in quantum computing, do you see the ROI today? Do you see companies actually moving quantum projects into production and getting real business value from it? Or is it more on the exploratory side, let me investigate this. Let me investigate that. And maybe in a couple of years we will move it into production.

Rene: Yeah. A lot of these projects, of course, and I would say our pilot phase, but one of our clients actually moved the quantum-inspired solution into production already. I cannot mention the name here, but it's one of the largest energy grid providers in the world. And they based maintain these energy grids, as you can imagine. Landlines and transformation stations and whatnot, and a lot of stuff. And they have like 20,000 field service workers, like 20,000 people that every day have a certain schedule where they need to drive with the car or whatever way of transportation they use. But basically go to these certain stops. They need to perform an action. I don't know, repair a transfer mate or whatever it is. Right? And so they have these 20,000 field service worker, which their schedule has to be planned in advance, of course.

And so as you probably know being from the computer science, that's a classical NP complex problem, which is called the TSP, the traveling salesman problem, right? Or a scheduling optimization problem. Right? The more stops you add, the more exponential will grow, right? The more complex it gets because what the optimization goal you want to achieve with that is, of course, you want to maximize the time that the field service worker spend working but you want to minimize at the same time, the time they spend on the road, traveling basically. Right? We want to find the optimal schedule. And long story short, for this client, we introduced our Mega QUBO Solution is quadratic unconstrained binary optimization algorithm. And this can be useful as workforce management, right? And so in a few minutes, the optimization that we developed, can identify a schedule that maximizes the amount of time spent working while minimizing the time spent on the road.

What they achieve is pretty impressive. They achieve a 20% reduction of travel time for 20,000 field service workers, right? And this is an execution time speed of 18 times on a GPU cluster. And so that is I think some pretty telling numbers. This is a lot of time and money saved there. And this is done, again, with quantum-inspired optimization, with our Mega QUBO Solution where you can put in certain things and then it gives you the optimal schedule as an output. And this is one example. I have a few examples where we're working with clients is like a large telecom operator. And again, so the one I just mentioned, the case, this is deployed in production already, right? This is running and it's optimizing the schedule and it gives a lot of ROI. If you can do the math yourself, like 20% reduction in travel time for 20,000 field service workers.

That's big money, of course. Right? But we're also are involved in a bunch of challenges like the Airbus Quantum Computing Challenge. We took part in that and a team of Machine Learning at Reply actually was winning that. The Machine Learning Reply folks won the Airbus Quantum Computing Challenge with a solution. Again, awesome optimization solver and optimizer basically. Well that optimize the loading of a plane, as you can imagine, that is a super complex problems with, I don't know, hundreds of variables maybe. I don't know, you have to sample the mass of the plane and you have all these physical constraints. You cannot go over a certain mass in total. You have to distribute the weight in a certain way. You want to minimize the fuel that has to be used. You want to minimize the time the plane is on the ground and all of that.

There is a bunch of things, if you want to load a plane optimally, that is very complex and challenging. And so we developed this solution and won the Airbus Quantum Computing Challenge, for example. Another example is for a telecom operator, we did a pilot project there where we were using again, optimization solver, Mega QUBO, to optimize the distribution of, basically, the frequency range. As you can imagine, think about a 5G network, like an antenna, the cell tower, if you will, and so, of course, this cell tower can only serve a certain amount of users, right? At some point it's maxed out and so you want to optimize. And so basically they have this range and they want to optimize. It's a little bit fragmented, right? They want to have as many users as possible. And so you need to basically pack them in an optimal way. And this is also showing some really promising speed ups when it comes to using quantum-inspired optimization for this. Right?

Yuval: To grow your business, the quantum side of your business, obviously, you want more customers to move into production and more customers to show positive ROI. Other than stronger computers, what do you think is missing? What is needed to accelerate that deployment of quantum solutions?

Rene: Well, you can split it in multiple areas, I would say. But if we're thinking about, from a technical standpoint, I think there is a need also to have this middle layer on also the programming language available that can be used, not just with one particular quantum computer. Because you need to imagine, these quantum computers are in very early stage, of course. Right? Think about the mainframe area in the the 20th century. In the middle of the 20th century, we had this mainframe area. Right. And so this is probably comparable to the state of quantum computers today. Of course, it's not 100% comparable, but you get the idea, right? It's huge things and they have all specific ways to program them, right?

You actually have to talk to them, to these quantum computers and program them and with a machine language. It's not even comparable to assemble or for classic computers, but really, you need to basically use a language that is not a high-level programming language. And also you're very much dependent on that certain quantum computer. And so the challenge for the development, I think that is being solved at the moment, were things like, for example, Q#, for example. Or although from IBM, they also have some stuff. And then a couple of other companies, of course, but basically you have a high-level programming language, which you can write your quantum algorithms in. And then you have the middle layer, basically, an interpreter or like for Q# it's the QIR, the quantum intermediate representation, which is a representation of your code that can then with a machine specific runtime can run on any of those.

Basically you have a high-level programming language. You have one language and you can target whatever kind of quantum computers are out there, as long as each of them provides a run time that can talk with your QIR basically. And so long story short, what I think is missing is, but I think it's being solved, is this generic and more approachable programming style and programming languages for quantum computers. But beside the technical things, of course. There's also, in general, clients might be hesitant cause they might still think, oh yeah, I heard about quantum computing, but this is a thing that might be happening in 10 years or so. We don't care about it at the moment. Right? That's what you hear oftentimes. And that's why I'm also with my QuBites video series, I try to focus on impact today because there is already achievable ROI with certain solutions.

And I think we need to show more cases where you can see the real benefit that can be achieved already today. Right? But of course, we've also need a more powerful quantum computing hardware. And another thing, which I'm also great that you, for example, Yuval have the podcast here and I'm doing my video series and I know there was a couple of other folks that also do podcasts or video series, is the education, right? And knowledge sharing in that space and getting quantum computing out of the research world and out of this a little bit exclusive club, if you will, to be more approachable for everyone. Because you need not just folks that have a PhD in physics, but you also need people that understand program management or project management, and you need, of course, engineers, you need developers, you need marketing folks.

You need all of this, what we know from classical computing is now needed because we're getting out of the science and research stage and getting into real applied quantum computing, even if its infancy. I think what we need is more folks joining the quantum computing bandwagon, if you will, that are not coming from the pure quantum information or quantum computing background from the university, but also more diverse background and a more inclusive community. And then we also need to show more ROI that can be achieved with these quantum solutions today. And like I said, of course from the technical standpoint, which is already being solved with Q# other things is this not just targeting specific quantum computers, but having more of a generic programming language.

Yuval: I was speaking with a partner from McKinsey the other day, and he mentioned a term that he called Business Translator. Someone who says, I can translate the technical benefits of quantum into a business benefit, into the ROI. This is not just what it does, but also why should you care, how can we deliver value today, and which projects you'll have to wait a couple of years before you can see real value. I like that term. And I think it may be actually what your company does amongst other things is help communicate the business value to the customers.

Rene: Exactly. And actually executing those. Right? We can communicate it, but we can also deliver. And that's, yeah. Important.

Yuval: My last question, you mentioned that Reply is built out of several different companies, several different units. Do you see quantum project as standalone projects or do you see them integrated into a broader enterprise software or hardware architecture? Oh, where's the data center? What's the SLA? How does data come in? How does data get out? Is it isolated or is it integrated?

Rene: Yeah, both I would say. It really depends on the clients. Also, it depends, of course, from the business development standpoint, what is the client relationship, right? Which folks are involved there, but from the typical thing, like the quantum projects themself, there are typically specific things. Cause you cannot just solve every problem with a quantum computer. That's also a misconception. A lot of folks I talk with, not the experts of course, but a lot of folks that are just hearing about quantum computing, there's a little bit of this misconception that quantum computers will be these generic computers, right? And while I'm holding up here my smartphone and showing it to you that we will have a quantum computer running in our smartphone, right?

Or a quantum computer running in our laptop. Well, this is not going to happen, at least not anytime soon, quantum computers are an addition. They're not going to replace classical computers. And they're an addition, and you can compare it with a GPU which is a specific graphics processing unit, right? Which is specific, well-built for certain tasks. That's the same thing with a quantum computer, although they are much larger than just the graphics card in your PC and work a little bit differently, quite a lot differently. But you can think about just another acceleration vehicle for specific problems, right? And so typically it's a very specific problem that needs to be identified first. What we do is our what we call the Reply roadmap to value starts with quantum-inspired optimization. As we go in with the client, we do a workshop, and then we actually select a certain scenario that is well-suited, right.

That actually benefits from the quantum computer. Because like I said, not every problem is well solvable with a quantum computer in a sense that you get actually benefit out of it. Right? And then we identified this problem and then we introduced assault and quantum computing algorithm. It could be QUBO. It could be some other things like there was also Monte-Carlo and a bunch of other, basically quantum-inspired optimization algorithms, that you can use. And it depends on the specific scenario, which is the best one. We select a scenario, introduce the optimal quantum computing algorithm. And then in the short term, we can run these developed quantum-inspired optimization solver on a classical hardware, like a GPU array and then long-term, you can take the exact same algorithm that was developed and put them on a quantum computer and well, on this one, it will just fly, right? It will be even faster, but that's typically how we approach these projects.

Yuval: Rene, that was a wealth of information. How can people get in touch with you to learn more about your work?

Rene: Yeah, well, you can find me on most social media. For example, LinkedIn, just search for my name, Rene Schulte. Feel free to connect also on Twitter. You can find me. My handle is @RSCHU. And I keep constantly sharing lots of stuff on social media. You might want to connect there. And then also my little video series, you can find it called QuBites, like just Qubit, but just with another E and an S at the end. A little bite to eat. So QuBites, you can find that at thereply.com site. And also if you just search with your favorite search engine, you will probably find it. And yeah, happy to connect and thanks for these great questions by the way, Yuval.

Yuval: Well, thank you very much for joining me today. Have a good day.

Rene: Thank you so much for having me. And it was a pleasure to be here.


See Also

No items found.

Start Creating Quantum Software Without Limits

contact us