Podcasts

Podcast with Konstantinos Karagiannis, Protoviti

10
August
,
2021

My guest today is Konstantinos, Head of Quantum Computing Services at Protiviti. Konstantinos and I talk about how financial services organizations get started with quantum and what they try to do with it, about his estimate of when quantum computing becomes a production tool and much more.

Listen to additional episodes by selecting 'podcasts' on our Insights page

THE FULL TRANSCRIPT IS BELOW

Yuval: Hello. Konstantinos and thanks for joining me today.

Konstantinos: Hi Yuval, good to talk to you again.

Yuval: Who are you and what do you do?

Konstantinos: My name is Konstantinos Karagiannis. I'm the head of quantum computing services at Protiviti, which is a global consulting company. I'm basically trying to help companies get ready for the post quantum world that we're living in.

Yuval: Excellent. How large is Protiviti and how large is the quantum group, if I may ask?

Konstantinos: The company is about 7,000 people or so, and we're spread out. So, we work in multiple regions. In the quantum computing team, we're tiny, it's emerging. So, there's just a handful of us in there. And we also can partner with other divisions we have. For example, we have a Security and Privacy division, and that way when we're doing something like a post-quantum cryptography audit, we can bring in people who do crypto all day long to do that part. And then we just handle the quantum parts, like the crypto agility and things like that.

Yuval: A board of directors of a large company, why would they care about quantum, in your opinion?

Konstantinos: Well, there are a few reasons. One is, like I mentioned, the crypto agility thing. I think everyone should be worried about the apocalypse that's coming. We all disagree on how many years away that is, I believe. Some people say 30 years and they've been saying it for 30 years so eventually you’ve got to change that date. I think it's a lot sooner than that. It will surprise us, but most companies should be worried about that kind of thing. There is going to come a day where any secrets with a long shelf life, it's probably not a good idea to be sending them around anymore with current methods. So, that's one concern.

And the other concern is: do you want your company to be a leader in some space or a fast follower or a slow follower? Those three kinds of tiers. And I find that financials, in particular, some of them do want to be leaders. They want to start looking now at how they can lift and shift classical problem solving to quantum. So they're ready, in the early days, to have maybe even IP in the space, some kind of intellectual property that's all their own, something they develop that gives them an edge against their competitors. So, that's why it's really not too early to get interested.

Yuval: You mentioned financial services. Are there other industries that are reaching out to you? Do you see the most interest in financial services or do you see it in other places as well?

Konstantinos: So for the crypto apocalypse side, everyone's interested in that, but for the use cases, financials are probably the strongest of interest. So you got to think about our types of customers. We're a consulting company. So, there are forms of quantum technologies that are more fully baked, there's quantum sensing, for example. But it's not really the kind of thing you would go to a consulting firm for. It would be weird also to come to us for help picking which piece of CMOS you should put into a camera. It's just a strange thing to go to a consulting firm for. So, that's where our customers are more curious about how they can apply these in actual use cases, like something they could do to improve their business. And we're ending up seeing the same kinds of discussions we had with machine learning about four years ago or so. Companies knew it was going to be big and knew that it would be an emerging technology that revolutionized their business, but they didn't really know how they can implement it.

They didn't know that they could create a neural network to solve some problem or some other fully baked system. So it's like that with quantum now. That's why financials are the most interested, but I do expect to see some other types of companies start to realize that there might be some benefits to be gained, especially in things like shipping and optimization. Quantum is going to be so great at that. We've already seen proof of concepts for things like traveling salesman problem. Let's say you have a disaster strike in the area and you need to get trucks shipping around supplies as officially as possible. One recent bit of work done on D-Wave's hybrid machine showed that instead of taking 27 kilometers of routing to get to all the points, their quantum approach was able to do it in 20 kilometers, showing that it was just a more efficient use of the traveling salesman problem. I think any kind of company will benefit from optimization really.

Yuval: So let's see if I can get you to spill your secrets. So let's assume I am a CIO or CTO of a large financial services firm, I've heard about quantum, I'm bought into the fact that I should really get into it or at least dip my toe in the water to see what kind of competitive advantage I could get there. How would you take a company like the hypothetical that I presented through the early stage of quantum computing? What would you advise us to do?

Konstantinos: So the first thing we would do probably is either join one of our workshops that we create for a few financial services customers - we do those regularly - or we could do a more one-on-one. And then we take you through what the possibilities are, the kinds of use cases that exist. We do like design thinking, that helps companies see where it might impact the organization. And then it really becomes pure consulting for a moment. It's what are you looking to get out of this. Let's say someone is a champion at a company and they really believe in quantum, they're looking for our help proving to their bosses that this is something that they need to invest in now because there is that of becoming a fast follower only, that your competitor will have something for six months you don't.

So we can show them with a POC, a proof of concept, that they can run internally, that there is some benefit and there is an extrapolation to quantum advantage that they can make for those decision-makers. We could show them that within six months or so, a hybrid annealing-classical approach will show real quantum advantage and optimization. It's basically a done deal. We're certain of it. And then that longer-term, within a year to two years, there are other types of advances in machine learning and things that are clearly going to be also advantageous. So some companies just want that.

That's the early phase. We talk them through it then we show them these POCs. We build them for them. And then we take it from there, do they want ongoing help refining, solving different problems as it goes along. So it's a whole journey. Then they also want advice on things like training. Again, just machine learning is a great example. In the early days, no one had a machine-learning person on staff, now they have dozens or more. So trying to help them figure out what they'll do for the quantum workforce problem, which is a problem.

Yuval: You mentioned the word extrapolation. And that gets me thinking: are the companies today try to solve new problems with quantum or are they just trying to recreate the solution to problems that they solved in other means to see that the solution in quantum makes sense and then you say, "Hey, but in a year or two years or six months, you'll be able to do things that you haven't done in a classical computer?" Which is it?

Konstantinos: That's a great question. I would love to see more creativity, but the reality is they want to know that there's a reason to buy in and usually that's their bread and butter. So if a company is really big on portfolio optimization, it would be strange for them to be like, "Let's try something new." They'd rather want to know that they can do something there. So we do tend to see what I call the big three. It's either optimization, machine learning or some kind of simulation. So you're basically interested in one of those. And then you take a problem you are solving and try and solve it a different way. Some of them have concerns outside of that space. We've had a few customers who said, "Well, what about energy? How will this help us save energy one day or reduce costs or those types of things?"

That's actually harder to help them with than doing something like portfolio optimization because it's so early in the field still that we can't predict how the pricing of machine access is going to go down. It's impossible to say in one year will it be cheaper or more expensive to access these machines? We just don't know. We do know that in theory there'll be using less energy one day, but if we build a whole lot of them, maybe not. I don't know. That's the beautiful thing, if these can do in three minutes what takes 33 hours, which is actually real numbers that we got out of a portfolio optimization, then sure, they will use less energy, but sometimes they're interested in things on the outskirts like that. But I haven't seen anyone come up with a truly new use case yet on the customer side. We're trying to help them figure those things out for the future.

Yuval: We see some of these financial services firm actually publish their work in scientific journals. Do you expect that to continue or is that just because they ended up hiring PhDs from MIT that are just used to publishing their work? Or is there some other reason about doing this?

Konstantinos: Yeah, I think about this a lot actually. I really do. This is one of those things that really bothers me. The fear that it's going to stop. Right now we're fortunate that a few of the big leaders, a few of the big financials, let's say JPMC, they've invested heavily, they have a lot of staff in this space, and so you do see them as a leader. They're actually producing things. I'd be shocked if they're revealing everything they're doing. That wouldn't really make sense. So there's going to come a time soon where I think we are going to start to see that drying up a bit. I think it's too soon for that to happen. So that worries me, but that's basically my thought right now is that I'm afraid this is going to end soon, that you're going to see less and less from the private sector being published. And that's just going to ruin it for everyone in a way, that's going to slow things down. We still need to be doing a lot more sharing.

Yuval: You mentioned AI and machine learning as an example maybe four years back when people were getting into it, but people were getting into AI 30 years ago, and then there was a 25 year “AI winter” where the expectations were excessive and the technology didn't deliver and so on. Are you concerned that something like that might happen with quantum?

Konstantinos: Not really. I think we're going to skip out on a “quantum winter”. I think just from what I see with all the investing and everything, I just don't see that happening. And the AI thing, we could see in hindsight what was wrong. This was the sort of thing that's been discussed when companies get acquired by Google or whatever. This has come up numerous times where anyone doing AI 30 years ago, they hit a wall where they said, "Okay, we can't do AI by forcing it. We need to actually do it in a way that simulates maybe how thinking works." That's where neural networks were starting to become popular. And at the time, the compute was just not there. We just had nothing analogous to it. So whenever someone would say, "Let's build a neural network," everyone was  like, "Yeah, that's cute, but you can't really do anything with it."

So that's really what caused that to dry up. Luckily with quantum, it's a little different. We know what we need and we know that we're building towards it. And yes, you can argue that the true universal gate-based, error-free quantum computer is a few years away, but we do know what we could do within our limitations. And that's why I'm pretty excited about this shallow circuit design work that's being done where people actually getting back to the extrapolation world again. They're getting results that prove that pretty soon we should have these use cases that will have benefits and have a reason for working now. So, that should avoid any “winter”. People are going to get really excited, especially like I said, with optimization. If within six months or so, we can prove absolute advantage there, that squashes any idea of a quantum winter.

Yuval: Do you expect usage to continue on the cloud or would you envision organizations starting to buy their own in-house hardware?

Konstantinos: Yeah, I think about this a lot too. Right now I think cloud, because these machines are so unwieldy that honestly even the cloud providers don't own them. If you go into Azure Quantum or AWS Braket, you're not touching machines that they're housing, you're still going through them as a pass-through all the way out to IonQ or Honeywell or whatever company. But they are also at that same time miniaturizing some of them to make them tabletop. And I laugh when I hear that term because the size of the table you would need for one of the machines they're calling tabletop, it's a pretty big table. Certainly not one I'd have in Manhattan, but they are shrinking. But what's the real benefit too, you have to ask. If you buy one and it costs a lot of money, you're buying one little slice of time.

People used to worry about becoming outdated and that's why the cloud became popular. No upgrading servers, any of that. So why would you force yourself into that too early with quantum when you know the next year you're going to have a machine that's an order of magnitude more powerful? So I think very few private purchases for quite a while until we could figure something out. Maybe with Honeywell's techniques. Honeywell's technology, they're allowing for qubits to actually be added, which is neat. You can actually just add more trapped ions. That's a neat idea that might extrapolate in the future to buying and then upgrading. So maybe. But it would be silly to buy a 10 qubit quantum computer when everyone's going to be talking about 200 qubits soon.

Yuval: Or you can do one of these iPhone plans right, where you upgrade to the latest computer every year when it comes out.

Konstantinos: Yeah. That sounds great. And then put those other ones in usage for students or something back in the cloud again. Yeah, I think that could happen. Or just send them right to a university, use that at a university. So we had a whole quantum reseller thing going. But for the time being, because advanced workloads are moving to the cloud anyway, I don't really see the big deal with time slicing the world's most powerful machines on the cloud.

Yuval: You mentioned manpower and training, and now you've mentioned students. So other than manpower, which seems to be a problem today of getting qualified people in quantum, what else do you see as roadblocks to that quantum future that we're all hoping for?

Konstantinos: Well, a weird skillset. And a lot of people in either consulting or on the customer side, they still have this view that it's all technology and someone who's smart and has been working in technology should be able to learn a few new things, but there's just no analogy for it. If you were a developer in one language in the past, you could learn another language, great. This also comes with it a need to have an understanding of physics, linear algebra, the ability to transcode really complex real-world problems into the quantum realm, if you will, and experience then with the different quantum programming environments, it's a lot. It's a lot to expect to just find. So even if you have a really sharp coder, they've got to go back and learn all the other prerequisites I mentioned.

So, that's a bit of a challenge. So the best way to influence that and curb the problems in the future would be to have better curriculums in schools. I currently actually do talk to a few colleges about this problem, because we have partnerships with Chicago Quantum Exchange, so University of Chicago and University of Maryland. And we're trying to put out the word that we need to focus on you want to be a coder track. Because I don't believe you're going to need to be a Ph.D. to be a quantum coder, which is good news.

For anyone who wants to get started in this, as long as you can get those basic prerequisites and start learning on that path, I think it is a viable way to go with a bachelor's or something and some experience, rather than PhDs. I think that PhDs in this space are mostly going to be on developing truly new things, like a new piece of the stack or a new type of hardware or something on that level rather than just being the next person to encode an optimization for a customer. I think it's going to be overkill to have a Ph.D. at that point.

Yuval: And if you were a betting man, when would quantum computers become production tools and emerge out of the playground and we're just going to try to replicate things that we've done with classical machines?

Konstantinos: I think it's going to be a few stages. So as one of the machines becomes powerful enough to do one of the use cases better than a classical machine can, it's going to be instant success in that use case. So, there's going to be no real reason to go back. This won't be neck and neck. With classical computing, you have Intel and AMD, neck and neck all the time. “My gaming machine is faster.” “My gaming machine is faster.” That's fine, but in quantum, once you pass classical, you then blow it out of the water. It's like the super intelligence theory for AI, it's the same exact thing with quantum. It's supercomputing taken to a new meaning. So I think it would be per use case that it becomes production. Even if there's errors, even if there's multiple shots you have to run into machine and all that.

As long as we can get the sharing right, that's when they become usable. I don't think we need to wait until we have perfect error-corrected qubits in the million range or whatever. I don't think we need to wait until Google's 2030 target for million error-corrected qubits. That would be great. It's just in the interim I think each advancement will bring a new advantage and once we have it, everyone will want it. They're going to want to use that because that'll be the best way to do it. Why would you do it a slower way? But again, we have to get the timesharing right. We have to. Because like I like to say, if you can do something in three minutes that takes 33 hours, that's awesome. But if you have to wait a week to access that machine, it's not so awesome. It just cost you a lot of days to do it. So we have to get the sharing right and have more machines online.

Yuval: Now I think you have your own podcast. Tell me a little bit about that, please.

Konstantinos: So mine is called The Post-Quantum World. And basically, I have on people who are doing things in the space that they want to talk about so we can explore the different technologies that are present today. So obviously I had you once talk about Classiq and I have had on, let's say, Honeywell to discuss their architecture, their machine or Microsoft to talk about their cloud access. So, that's basically what I do. I try to introduce a new technology and business aspect to help the listeners understand how it's going to impact the real world, not just research papers. So it's called The Post-Quantum World.

Yuval: And where can people find the podcast, and more broadly, how can people get in touch with you to learn more about the work that you're doing?

Konstantinos: So the podcast is available everywhere, Apple, Spotify, wherever you go, just look for The Post Quantum World. I guess Twitter is the easiest way to just find me if you want to. So I'm @KonstantHacker. You can also go to protiviti.com and check out what we're doing there. And they could reach out to me any of those ways at protiviti.com/postquantum, you can get to a quantum landing page.

Yuval: Perfect. Well, Konstantinos, thanks so much for joining me today.

Konstantinos: Thanks. It was great talking to you.


My guest today is Konstantinos, Head of Quantum Computing Services at Protiviti. Konstantinos and I talk about how financial services organizations get started with quantum and what they try to do with it, about his estimate of when quantum computing becomes a production tool and much more.

Listen to additional episodes by selecting 'podcasts' on our Insights page

THE FULL TRANSCRIPT IS BELOW

Yuval: Hello. Konstantinos and thanks for joining me today.

Konstantinos: Hi Yuval, good to talk to you again.

Yuval: Who are you and what do you do?

Konstantinos: My name is Konstantinos Karagiannis. I'm the head of quantum computing services at Protiviti, which is a global consulting company. I'm basically trying to help companies get ready for the post quantum world that we're living in.

Yuval: Excellent. How large is Protiviti and how large is the quantum group, if I may ask?

Konstantinos: The company is about 7,000 people or so, and we're spread out. So, we work in multiple regions. In the quantum computing team, we're tiny, it's emerging. So, there's just a handful of us in there. And we also can partner with other divisions we have. For example, we have a Security and Privacy division, and that way when we're doing something like a post-quantum cryptography audit, we can bring in people who do crypto all day long to do that part. And then we just handle the quantum parts, like the crypto agility and things like that.

Yuval: A board of directors of a large company, why would they care about quantum, in your opinion?

Konstantinos: Well, there are a few reasons. One is, like I mentioned, the crypto agility thing. I think everyone should be worried about the apocalypse that's coming. We all disagree on how many years away that is, I believe. Some people say 30 years and they've been saying it for 30 years so eventually you’ve got to change that date. I think it's a lot sooner than that. It will surprise us, but most companies should be worried about that kind of thing. There is going to come a day where any secrets with a long shelf life, it's probably not a good idea to be sending them around anymore with current methods. So, that's one concern.

And the other concern is: do you want your company to be a leader in some space or a fast follower or a slow follower? Those three kinds of tiers. And I find that financials, in particular, some of them do want to be leaders. They want to start looking now at how they can lift and shift classical problem solving to quantum. So they're ready, in the early days, to have maybe even IP in the space, some kind of intellectual property that's all their own, something they develop that gives them an edge against their competitors. So, that's why it's really not too early to get interested.

Yuval: You mentioned financial services. Are there other industries that are reaching out to you? Do you see the most interest in financial services or do you see it in other places as well?

Konstantinos: So for the crypto apocalypse side, everyone's interested in that, but for the use cases, financials are probably the strongest of interest. So you got to think about our types of customers. We're a consulting company. So, there are forms of quantum technologies that are more fully baked, there's quantum sensing, for example. But it's not really the kind of thing you would go to a consulting firm for. It would be weird also to come to us for help picking which piece of CMOS you should put into a camera. It's just a strange thing to go to a consulting firm for. So, that's where our customers are more curious about how they can apply these in actual use cases, like something they could do to improve their business. And we're ending up seeing the same kinds of discussions we had with machine learning about four years ago or so. Companies knew it was going to be big and knew that it would be an emerging technology that revolutionized their business, but they didn't really know how they can implement it.

They didn't know that they could create a neural network to solve some problem or some other fully baked system. So it's like that with quantum now. That's why financials are the most interested, but I do expect to see some other types of companies start to realize that there might be some benefits to be gained, especially in things like shipping and optimization. Quantum is going to be so great at that. We've already seen proof of concepts for things like traveling salesman problem. Let's say you have a disaster strike in the area and you need to get trucks shipping around supplies as officially as possible. One recent bit of work done on D-Wave's hybrid machine showed that instead of taking 27 kilometers of routing to get to all the points, their quantum approach was able to do it in 20 kilometers, showing that it was just a more efficient use of the traveling salesman problem. I think any kind of company will benefit from optimization really.

Yuval: So let's see if I can get you to spill your secrets. So let's assume I am a CIO or CTO of a large financial services firm, I've heard about quantum, I'm bought into the fact that I should really get into it or at least dip my toe in the water to see what kind of competitive advantage I could get there. How would you take a company like the hypothetical that I presented through the early stage of quantum computing? What would you advise us to do?

Konstantinos: So the first thing we would do probably is either join one of our workshops that we create for a few financial services customers - we do those regularly - or we could do a more one-on-one. And then we take you through what the possibilities are, the kinds of use cases that exist. We do like design thinking, that helps companies see where it might impact the organization. And then it really becomes pure consulting for a moment. It's what are you looking to get out of this. Let's say someone is a champion at a company and they really believe in quantum, they're looking for our help proving to their bosses that this is something that they need to invest in now because there is that of becoming a fast follower only, that your competitor will have something for six months you don't.

So we can show them with a POC, a proof of concept, that they can run internally, that there is some benefit and there is an extrapolation to quantum advantage that they can make for those decision-makers. We could show them that within six months or so, a hybrid annealing-classical approach will show real quantum advantage and optimization. It's basically a done deal. We're certain of it. And then that longer-term, within a year to two years, there are other types of advances in machine learning and things that are clearly going to be also advantageous. So some companies just want that.

That's the early phase. We talk them through it then we show them these POCs. We build them for them. And then we take it from there, do they want ongoing help refining, solving different problems as it goes along. So it's a whole journey. Then they also want advice on things like training. Again, just machine learning is a great example. In the early days, no one had a machine-learning person on staff, now they have dozens or more. So trying to help them figure out what they'll do for the quantum workforce problem, which is a problem.

Yuval: You mentioned the word extrapolation. And that gets me thinking: are the companies today try to solve new problems with quantum or are they just trying to recreate the solution to problems that they solved in other means to see that the solution in quantum makes sense and then you say, "Hey, but in a year or two years or six months, you'll be able to do things that you haven't done in a classical computer?" Which is it?

Konstantinos: That's a great question. I would love to see more creativity, but the reality is they want to know that there's a reason to buy in and usually that's their bread and butter. So if a company is really big on portfolio optimization, it would be strange for them to be like, "Let's try something new." They'd rather want to know that they can do something there. So we do tend to see what I call the big three. It's either optimization, machine learning or some kind of simulation. So you're basically interested in one of those. And then you take a problem you are solving and try and solve it a different way. Some of them have concerns outside of that space. We've had a few customers who said, "Well, what about energy? How will this help us save energy one day or reduce costs or those types of things?"

That's actually harder to help them with than doing something like portfolio optimization because it's so early in the field still that we can't predict how the pricing of machine access is going to go down. It's impossible to say in one year will it be cheaper or more expensive to access these machines? We just don't know. We do know that in theory there'll be using less energy one day, but if we build a whole lot of them, maybe not. I don't know. That's the beautiful thing, if these can do in three minutes what takes 33 hours, which is actually real numbers that we got out of a portfolio optimization, then sure, they will use less energy, but sometimes they're interested in things on the outskirts like that. But I haven't seen anyone come up with a truly new use case yet on the customer side. We're trying to help them figure those things out for the future.

Yuval: We see some of these financial services firm actually publish their work in scientific journals. Do you expect that to continue or is that just because they ended up hiring PhDs from MIT that are just used to publishing their work? Or is there some other reason about doing this?

Konstantinos: Yeah, I think about this a lot actually. I really do. This is one of those things that really bothers me. The fear that it's going to stop. Right now we're fortunate that a few of the big leaders, a few of the big financials, let's say JPMC, they've invested heavily, they have a lot of staff in this space, and so you do see them as a leader. They're actually producing things. I'd be shocked if they're revealing everything they're doing. That wouldn't really make sense. So there's going to come a time soon where I think we are going to start to see that drying up a bit. I think it's too soon for that to happen. So that worries me, but that's basically my thought right now is that I'm afraid this is going to end soon, that you're going to see less and less from the private sector being published. And that's just going to ruin it for everyone in a way, that's going to slow things down. We still need to be doing a lot more sharing.

Yuval: You mentioned AI and machine learning as an example maybe four years back when people were getting into it, but people were getting into AI 30 years ago, and then there was a 25 year “AI winter” where the expectations were excessive and the technology didn't deliver and so on. Are you concerned that something like that might happen with quantum?

Konstantinos: Not really. I think we're going to skip out on a “quantum winter”. I think just from what I see with all the investing and everything, I just don't see that happening. And the AI thing, we could see in hindsight what was wrong. This was the sort of thing that's been discussed when companies get acquired by Google or whatever. This has come up numerous times where anyone doing AI 30 years ago, they hit a wall where they said, "Okay, we can't do AI by forcing it. We need to actually do it in a way that simulates maybe how thinking works." That's where neural networks were starting to become popular. And at the time, the compute was just not there. We just had nothing analogous to it. So whenever someone would say, "Let's build a neural network," everyone was  like, "Yeah, that's cute, but you can't really do anything with it."

So that's really what caused that to dry up. Luckily with quantum, it's a little different. We know what we need and we know that we're building towards it. And yes, you can argue that the true universal gate-based, error-free quantum computer is a few years away, but we do know what we could do within our limitations. And that's why I'm pretty excited about this shallow circuit design work that's being done where people actually getting back to the extrapolation world again. They're getting results that prove that pretty soon we should have these use cases that will have benefits and have a reason for working now. So, that should avoid any “winter”. People are going to get really excited, especially like I said, with optimization. If within six months or so, we can prove absolute advantage there, that squashes any idea of a quantum winter.

Yuval: Do you expect usage to continue on the cloud or would you envision organizations starting to buy their own in-house hardware?

Konstantinos: Yeah, I think about this a lot too. Right now I think cloud, because these machines are so unwieldy that honestly even the cloud providers don't own them. If you go into Azure Quantum or AWS Braket, you're not touching machines that they're housing, you're still going through them as a pass-through all the way out to IonQ or Honeywell or whatever company. But they are also at that same time miniaturizing some of them to make them tabletop. And I laugh when I hear that term because the size of the table you would need for one of the machines they're calling tabletop, it's a pretty big table. Certainly not one I'd have in Manhattan, but they are shrinking. But what's the real benefit too, you have to ask. If you buy one and it costs a lot of money, you're buying one little slice of time.

People used to worry about becoming outdated and that's why the cloud became popular. No upgrading servers, any of that. So why would you force yourself into that too early with quantum when you know the next year you're going to have a machine that's an order of magnitude more powerful? So I think very few private purchases for quite a while until we could figure something out. Maybe with Honeywell's techniques. Honeywell's technology, they're allowing for qubits to actually be added, which is neat. You can actually just add more trapped ions. That's a neat idea that might extrapolate in the future to buying and then upgrading. So maybe. But it would be silly to buy a 10 qubit quantum computer when everyone's going to be talking about 200 qubits soon.

Yuval: Or you can do one of these iPhone plans right, where you upgrade to the latest computer every year when it comes out.

Konstantinos: Yeah. That sounds great. And then put those other ones in usage for students or something back in the cloud again. Yeah, I think that could happen. Or just send them right to a university, use that at a university. So we had a whole quantum reseller thing going. But for the time being, because advanced workloads are moving to the cloud anyway, I don't really see the big deal with time slicing the world's most powerful machines on the cloud.

Yuval: You mentioned manpower and training, and now you've mentioned students. So other than manpower, which seems to be a problem today of getting qualified people in quantum, what else do you see as roadblocks to that quantum future that we're all hoping for?

Konstantinos: Well, a weird skillset. And a lot of people in either consulting or on the customer side, they still have this view that it's all technology and someone who's smart and has been working in technology should be able to learn a few new things, but there's just no analogy for it. If you were a developer in one language in the past, you could learn another language, great. This also comes with it a need to have an understanding of physics, linear algebra, the ability to transcode really complex real-world problems into the quantum realm, if you will, and experience then with the different quantum programming environments, it's a lot. It's a lot to expect to just find. So even if you have a really sharp coder, they've got to go back and learn all the other prerequisites I mentioned.

So, that's a bit of a challenge. So the best way to influence that and curb the problems in the future would be to have better curriculums in schools. I currently actually do talk to a few colleges about this problem, because we have partnerships with Chicago Quantum Exchange, so University of Chicago and University of Maryland. And we're trying to put out the word that we need to focus on you want to be a coder track. Because I don't believe you're going to need to be a Ph.D. to be a quantum coder, which is good news.

For anyone who wants to get started in this, as long as you can get those basic prerequisites and start learning on that path, I think it is a viable way to go with a bachelor's or something and some experience, rather than PhDs. I think that PhDs in this space are mostly going to be on developing truly new things, like a new piece of the stack or a new type of hardware or something on that level rather than just being the next person to encode an optimization for a customer. I think it's going to be overkill to have a Ph.D. at that point.

Yuval: And if you were a betting man, when would quantum computers become production tools and emerge out of the playground and we're just going to try to replicate things that we've done with classical machines?

Konstantinos: I think it's going to be a few stages. So as one of the machines becomes powerful enough to do one of the use cases better than a classical machine can, it's going to be instant success in that use case. So, there's going to be no real reason to go back. This won't be neck and neck. With classical computing, you have Intel and AMD, neck and neck all the time. “My gaming machine is faster.” “My gaming machine is faster.” That's fine, but in quantum, once you pass classical, you then blow it out of the water. It's like the super intelligence theory for AI, it's the same exact thing with quantum. It's supercomputing taken to a new meaning. So I think it would be per use case that it becomes production. Even if there's errors, even if there's multiple shots you have to run into machine and all that.

As long as we can get the sharing right, that's when they become usable. I don't think we need to wait until we have perfect error-corrected qubits in the million range or whatever. I don't think we need to wait until Google's 2030 target for million error-corrected qubits. That would be great. It's just in the interim I think each advancement will bring a new advantage and once we have it, everyone will want it. They're going to want to use that because that'll be the best way to do it. Why would you do it a slower way? But again, we have to get the timesharing right. We have to. Because like I like to say, if you can do something in three minutes that takes 33 hours, that's awesome. But if you have to wait a week to access that machine, it's not so awesome. It just cost you a lot of days to do it. So we have to get the sharing right and have more machines online.

Yuval: Now I think you have your own podcast. Tell me a little bit about that, please.

Konstantinos: So mine is called The Post-Quantum World. And basically, I have on people who are doing things in the space that they want to talk about so we can explore the different technologies that are present today. So obviously I had you once talk about Classiq and I have had on, let's say, Honeywell to discuss their architecture, their machine or Microsoft to talk about their cloud access. So, that's basically what I do. I try to introduce a new technology and business aspect to help the listeners understand how it's going to impact the real world, not just research papers. So it's called The Post-Quantum World.

Yuval: And where can people find the podcast, and more broadly, how can people get in touch with you to learn more about the work that you're doing?

Konstantinos: So the podcast is available everywhere, Apple, Spotify, wherever you go, just look for The Post Quantum World. I guess Twitter is the easiest way to just find me if you want to. So I'm @KonstantHacker. You can also go to protiviti.com and check out what we're doing there. And they could reach out to me any of those ways at protiviti.com/postquantum, you can get to a quantum landing page.

Yuval: Perfect. Well, Konstantinos, thanks so much for joining me today.

Konstantinos: Thanks. It was great talking to you.


About "The Qubit Guy's Podcast"

Hosted by The Qubit Guy (Yuval Boger, our Chief Marketing Officer), the podcast hosts thought leaders in quantum computing to discuss business and technical questions that impact the quantum computing ecosystem. Our guests provide interesting insights about quantum computer software and algorithm, quantum computer hardware, key applications for quantum computing, market studies of the quantum industry and more.

If you would like to suggest a guest for the podcast, please contact us.

Start Creating Quantum Software Without Limits

contact us