ポッドキャスト
25
May
,
2022

Podcast with James Sanders, 451 Research

Share the article
Our library

My guest today is James Sanders, Cloud and Quantum Analyst at 451 research. We spoke about co-locating quantum and classical computers in the data centers, quantum hype, the five tenors of quantum analysts and much more.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval: Hello, James. And thanks for joining me today.

James: Thanks for having me. Thanks for spending your afternoon with me.

Yuval: So who are you and what do you do?

James: So the short answer and I'll expand on the long answer in a second, but the short answer is I'm James Sanders. I'm an analyst with 451 Research, which is part of the S&P Global Market Intelligence. And what I do is I talk to the founders of quantum computing companies and see what it is they're doing and write reports about what they're doing and look at the market overall. And the long answer is I kind of came into this as a technology journalist and in a previous career. And in 2015, I was just given the assignment, "Hey, will you write a resource article about quantum computing?" And the kind of amazing thing about this is these articles are supposed to be completely comprehensive. So I have the task in 2015 of all times to look at every single startup and the technology and where these things are going, how people can use them.

It's the Five W's of journalism. Right? It's who, what, when, where, why, and how. And that was a lot of research. That was just me diving into everything. And I just kind of fell in love with it instantly. And I've been lucky enough to keep that as part of my beat as a journalist, and now my research area as an analyst. And it's really the most exciting thing in my job, just because it's an emerging technology that's really cool. So I'm very lucky that I get to do this, that I get to have these fun conversations with founders day in and day out.

Yuval: So this is rare that I meet someone who's from the stone age of quantum computing. And I think that you also cover data centers, right? And observability? You have other items in your portfolio if I'm not mistaken.

James: Yes. So I think part of the reason that I got the quantum computing assignment originally is that I had spent a lot of time looking at different compute architectures, like Arm for data center or the IBM power architecture. These are all kind of HPC things, increasingly Arm. And with that, there's also GPUs, and today I'm also looking at accelerators for AI and ML workloads that are, you know, increasingly deployed to the cloud. And so, we as 451, really see quantum computing as an entry in a class of compute accelerators that are going to make certain workloads faster, cheaper to run eventually. And so I cover what we call cloud differentiated silicon, quantum computing, and now, application and infrastructure monitoring. So I'm kind of all over the place, but it keeps me on my feet. It keeps me learning, and that's really fun for me.

Yuval: Conventional wisdom is that classical computers are going to be used together with quantum computers in the near term. We're not going to be running Zoom on a quantum computer, and because of noise and coherence and other issues, there's going to be some interplay, or some very tight interplay, between classical and quantum computers in quantum-enabled applications. Do you think that quantum computers should be co-located in the data center where the classical ones, or is it okay to have private quantum clouds or really a distributed system?

James: I think the premise of having an on-prem quantum computer is a little bit interesting. And I don't want to be prescriptive about things because there could be a circumstance where this makes sense, especially if you are deploying to a HPC lab, like a national laboratory. For, let's say commercialized quantum computing, I think it makes a lot more sense to have this in the cloud. By the time you have one of these systems deployed, especially if it's something that's quite involved to install in a physical location, like a superconducting system, by the time you've got it deployed, on-prem, it's probably already out of date. And that's a testament to how fast this industry and the advance of hardware is changing. But it's also a practicality thing. Because of the need to have classical systems in conjunction with quantum systems, it makes a lot of sense to have that elastic ability of cloud to give you as much classical compute power, as you need to complement a quantum system.

Yuval: When people move from smaller machines, 10, 20, 30 qubits to quantum machines that cannot be simulated with classical computers, what do you think is going to be needed in the observability and debugging and performance monitoring stage? Is it like a deck of punch cards that we just submit to the machine and see what comes out or are we going to have something more sophisticated?

James: I think for the foreseeable future, a lot of the validation of the results that you get from a quantum computer is going to be inherently manual. It's going to require some human oversight to say, "Hey, if we have this result, does it actually match our expectation? Or is this just wildly, wildly wrong?" An application architected for a quantum computer is substantially different from what you would have for a production workload on, you know, a web server or something of that sort. So the realm of application performance monitoring for classical systems and what you would expect with the quantum computer are probably worlds apart and will be for some time, just because the outcomes that you're getting are going to be more along the lines of scientific computing of high performance computing than you would have for a production web server where you are selling widgets online or running a social media network. They're really two different disciplines. And a lot of the AP and the AIPM and stuff in service for classical systems today I don't think would translate particularly well to a quantum computer.

Yuval: Very good. I read not too long ago, an article by a French analyst that basically was talking about hype in quantum computing. And it made a couple of different arguments. One that hype is not necessarily bad because it increases investments and interest in scientific research, but over-hype of course is dangerous because it creates unrealistic expectations. So do you think quantum computing is hype today? And related to that, if it is, then should a large Fortune 500 company start investigating quantum or wait a couple years until the hype subsides?

James: This is a tricky question because people really like to, and I've been guilty of this in the past as well, they really like to talk about the idea of a quantum winter or quantum fatigue, or all of these bad things. And harking back to the '80s when all of this happened in AI, when there was the AI winter, this sort of thing. The problem is there is a real risk of having quantum computers become these things that the hype and the people trying to make a quick buck on them overtakes the actual science that's happening to build all of these systems. And so I am not the person out there who is the, you know, type of enthusiast that you would have for a less commercially proven technology like Blockchain.

I'm not a Blockchain booster, and I recoil a little bit when people try to be quantum boosters in the way that Blockchain boosters really speak. So I practice caution in that regard. That said, more so with quantum computers and anything classical, the actual capabilities of a system is going to be characterized by the software that's available for it. And the software that you can develop for it is influenced by who's experimenting now. So actualizing what a quantum computer can do requires an investment of time today. And that's the focal point that I want people to look at is, where are you investing your time? What is your research like?

And, you know, how much time can you dedicate to have an engineer to, at your company or a product manager, look at the capabilities of quantum computers today, the roadmaps of where they will be in the future, and map that to business problems you already have. And there's endless numbers at this point of quantum consultancies who want to help you with that journey. And I think that's all well and good, but it does require not just an expertise in quantum computing but also the domain expertise of the problem that you're trying to solve and bringing those two groups together and spending the time to really dive deeply into those problems is what needs to happen today in order for a future where quantum computers can eclipse the capabilities of classical computers to be realized.

Yuval: The argument for getting in quantum would be to say, the cost of doing quantum right now is not high. It's an operating expense, it's not a capital expense. Right? You hire a few people, you give them a little bit of a cloud budget, maybe some consulting budget, and that's it. You don't have to build a factory right now or a new data center or anything like that. The cost of not doing quantum is potentially nothing but also potentially that you are so far behind your competitors that wow, it's like not having gone into the web in time or not having done machine learning in time. It might become a must for the business. Where do you stand on that spectrum of do or not do?

James: There's definitely an opportunity cost that comes with this. And my stance on this is I'm a cheerful pessimist. I think there's definitely going to be a advantage that quantum computers can provide, but we might not actually see that this is an advantage provided by quantum computers until it's been in the market for maybe a year or more. Because if you have a giant competitive advantage, or even a moderate competitive advantage, I don't think you're going to publicize that right out. And yes, there's the counter argument of, "science is done in public."

And I say that a lot, but when it comes to dollars and cents, I think if you've got a business that's identified a problem and they've been able to deploy a quantum computer in production to aid in solving that problem, and if they're able to save money or make more money doing that, they might be a little more tight-lipped about it than I think everyone would want them to be. So, yes, I think the danger is very real that if you aren't getting into quantum computing today, you could be, you know, overtaken by a competitor because of the advantage that a quantum computer or a quantum-realized society - that's so vague of a term that I shouldn't have said it, but it's an interesting one - that a quantum future can provide.

Yuval: I want to ask you two quick questions about things that you're doing today and a little bit on your past. Today, I think you're part of those three tenors of quantum analyst, or maybe you guys are five tenors, right? How's that going? And how much variance is there in the opinion between the five participants? I think this is Andre Konig’s program.

James: Yeah. It's a really fun experience to work with this team, and if we agreed on everything, I think it would be really boring as a podcast. You have to have a little bit of conflict to produce good TV, so it's fun to do. We have different opinions that are informed by the fields that we come from. And I know that that Andre's been on your podcast as well. He's less fond of the idea of different, you know, qubit architectures making that huge of a difference. I think that having all of these qubit architectures is going to drive the science faster, so I'm really fond of, you know, the sort of jockeying the horse race of all of these different qubit architectures because I think it's pretty interesting, and I think that competition, in mostly positive but occasionally negative ways, is driving competition and just making the field more interesting and motivating a lot of people.

Yuval: I think it is interesting, and it's not just horse race. I think it also could influence the strategic choices that end users make. Whether they commit to a fully integrated top to bottom architecture, or whether they want to take best of breed components because they may change over time. So I think you bring a super interesting point and to me, it's beyond just, "I've got more qubits than you." The other thing I wanted to ask you, you said you were a journalist before you became an analyst. And there was a time where the media picked up on the stories, "Oh, quantum's going to change the world," and "We're going to do quantum for chemistry," and "We're going to do quantum for finance," and "I'm going to have a quantum computer in my pocket," and what have you. Do you think that has changed as far - I know you're not in directly in the media right now - what kind of stories do you think - we're not going to tell anyone, it's just you and I speaking and the audience listening - what kind of stories do you think the media is looking for right now on quantum computing?

James: I've seen a lot greater understanding of quantum computing the last couple of years compared to seeing what other journalists, you know, when I was actually a journalist, what my competition was doing, what they were writing, I think the understanding has improved a lot and there are still the weird questions like mobile, quantum computing. That's not a thing. That's not going to be a thing. So the general understanding has improved. I think, rather than asking what stories are they looking for, where the missed stories are, I'm really curious what the first case of quantum piracy is going to be. You talked about, you know, are companies going to want to go with just one vendor or try a best of breed approach? And there's all these quantum software consultancies that will help you on your quantum journey.

And I wonder, if you work with a company that helps you develop an algorithm that works on one quantum computer, if you end your relationship with that company and take that idea, take that work and apply it to a different quantum computer, what's the implication of that? At the time these contracts are defined, who owns the intellectual property that's developed as part of these professional engagements? Especially when you have full stack companies that have their own hardware and their professional software team, I am really curious what happens if you take something from say a vendor blue to vendor orange or to vendor green, and yeah, there could be a little bit of work in translating something from one qubit architecture to another, but this is all things that I think are, for the most part, untested.

And so I'm really curious to see, is this going to be some sort of, you know, public spat in the future? Our company is really thinking about this and who's going to really control the value that quantum computers create. And there's also the aspect of this, which is as a quantum computing company, if you're building hardware, do you want to be a provider of hardware? Do you want to be the next Intel? Or do you want to be a software company that happens to have a hardware backend? And those are two very different questions.

Yuval: I never thought about calling it quantum piracy, but maybe I was just not as imaginative as you are, but I think you bring up an excellent point. We see that a lot of companies think that quantum is strategic, and therefore, they do want to develop their internal competencies. They don't want to just outsource a hundred percent to someone else. This is not like painting my house, I'll just ask someone to do it every 10 years, and we'll be done. I don't need to learn how to paint myself. And part of the reason they want to build it internally is exactly that IP. It's very difficult to ask a consultant to unlearn something that they learned how to do, how to do better option pricing or how to do better chemical simulation when they work with you. And it's easier to keep that IP and knowledge internally. So I hope I understood you correctly, but I think we're in perfect agreement on that point. James, this has been lots of fun. How can people get in touch with you to learn more about your work?

James: So you can go to 451research.com. My email is james.sanders@spglobal.com or you can catch me on Twitter @JAS_NP.

Yuval: That's perfect. Thank you so much for joining with me today.

James: Thanks for having me.

My guest today is James Sanders, Cloud and Quantum Analyst at 451 research. We spoke about co-locating quantum and classical computers in the data centers, quantum hype, the five tenors of quantum analysts and much more.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval: Hello, James. And thanks for joining me today.

James: Thanks for having me. Thanks for spending your afternoon with me.

Yuval: So who are you and what do you do?

James: So the short answer and I'll expand on the long answer in a second, but the short answer is I'm James Sanders. I'm an analyst with 451 Research, which is part of the S&P Global Market Intelligence. And what I do is I talk to the founders of quantum computing companies and see what it is they're doing and write reports about what they're doing and look at the market overall. And the long answer is I kind of came into this as a technology journalist and in a previous career. And in 2015, I was just given the assignment, "Hey, will you write a resource article about quantum computing?" And the kind of amazing thing about this is these articles are supposed to be completely comprehensive. So I have the task in 2015 of all times to look at every single startup and the technology and where these things are going, how people can use them.

It's the Five W's of journalism. Right? It's who, what, when, where, why, and how. And that was a lot of research. That was just me diving into everything. And I just kind of fell in love with it instantly. And I've been lucky enough to keep that as part of my beat as a journalist, and now my research area as an analyst. And it's really the most exciting thing in my job, just because it's an emerging technology that's really cool. So I'm very lucky that I get to do this, that I get to have these fun conversations with founders day in and day out.

Yuval: So this is rare that I meet someone who's from the stone age of quantum computing. And I think that you also cover data centers, right? And observability? You have other items in your portfolio if I'm not mistaken.

James: Yes. So I think part of the reason that I got the quantum computing assignment originally is that I had spent a lot of time looking at different compute architectures, like Arm for data center or the IBM power architecture. These are all kind of HPC things, increasingly Arm. And with that, there's also GPUs, and today I'm also looking at accelerators for AI and ML workloads that are, you know, increasingly deployed to the cloud. And so, we as 451, really see quantum computing as an entry in a class of compute accelerators that are going to make certain workloads faster, cheaper to run eventually. And so I cover what we call cloud differentiated silicon, quantum computing, and now, application and infrastructure monitoring. So I'm kind of all over the place, but it keeps me on my feet. It keeps me learning, and that's really fun for me.

Yuval: Conventional wisdom is that classical computers are going to be used together with quantum computers in the near term. We're not going to be running Zoom on a quantum computer, and because of noise and coherence and other issues, there's going to be some interplay, or some very tight interplay, between classical and quantum computers in quantum-enabled applications. Do you think that quantum computers should be co-located in the data center where the classical ones, or is it okay to have private quantum clouds or really a distributed system?

James: I think the premise of having an on-prem quantum computer is a little bit interesting. And I don't want to be prescriptive about things because there could be a circumstance where this makes sense, especially if you are deploying to a HPC lab, like a national laboratory. For, let's say commercialized quantum computing, I think it makes a lot more sense to have this in the cloud. By the time you have one of these systems deployed, especially if it's something that's quite involved to install in a physical location, like a superconducting system, by the time you've got it deployed, on-prem, it's probably already out of date. And that's a testament to how fast this industry and the advance of hardware is changing. But it's also a practicality thing. Because of the need to have classical systems in conjunction with quantum systems, it makes a lot of sense to have that elastic ability of cloud to give you as much classical compute power, as you need to complement a quantum system.

Yuval: When people move from smaller machines, 10, 20, 30 qubits to quantum machines that cannot be simulated with classical computers, what do you think is going to be needed in the observability and debugging and performance monitoring stage? Is it like a deck of punch cards that we just submit to the machine and see what comes out or are we going to have something more sophisticated?

James: I think for the foreseeable future, a lot of the validation of the results that you get from a quantum computer is going to be inherently manual. It's going to require some human oversight to say, "Hey, if we have this result, does it actually match our expectation? Or is this just wildly, wildly wrong?" An application architected for a quantum computer is substantially different from what you would have for a production workload on, you know, a web server or something of that sort. So the realm of application performance monitoring for classical systems and what you would expect with the quantum computer are probably worlds apart and will be for some time, just because the outcomes that you're getting are going to be more along the lines of scientific computing of high performance computing than you would have for a production web server where you are selling widgets online or running a social media network. They're really two different disciplines. And a lot of the AP and the AIPM and stuff in service for classical systems today I don't think would translate particularly well to a quantum computer.

Yuval: Very good. I read not too long ago, an article by a French analyst that basically was talking about hype in quantum computing. And it made a couple of different arguments. One that hype is not necessarily bad because it increases investments and interest in scientific research, but over-hype of course is dangerous because it creates unrealistic expectations. So do you think quantum computing is hype today? And related to that, if it is, then should a large Fortune 500 company start investigating quantum or wait a couple years until the hype subsides?

James: This is a tricky question because people really like to, and I've been guilty of this in the past as well, they really like to talk about the idea of a quantum winter or quantum fatigue, or all of these bad things. And harking back to the '80s when all of this happened in AI, when there was the AI winter, this sort of thing. The problem is there is a real risk of having quantum computers become these things that the hype and the people trying to make a quick buck on them overtakes the actual science that's happening to build all of these systems. And so I am not the person out there who is the, you know, type of enthusiast that you would have for a less commercially proven technology like Blockchain.

I'm not a Blockchain booster, and I recoil a little bit when people try to be quantum boosters in the way that Blockchain boosters really speak. So I practice caution in that regard. That said, more so with quantum computers and anything classical, the actual capabilities of a system is going to be characterized by the software that's available for it. And the software that you can develop for it is influenced by who's experimenting now. So actualizing what a quantum computer can do requires an investment of time today. And that's the focal point that I want people to look at is, where are you investing your time? What is your research like?

And, you know, how much time can you dedicate to have an engineer to, at your company or a product manager, look at the capabilities of quantum computers today, the roadmaps of where they will be in the future, and map that to business problems you already have. And there's endless numbers at this point of quantum consultancies who want to help you with that journey. And I think that's all well and good, but it does require not just an expertise in quantum computing but also the domain expertise of the problem that you're trying to solve and bringing those two groups together and spending the time to really dive deeply into those problems is what needs to happen today in order for a future where quantum computers can eclipse the capabilities of classical computers to be realized.

Yuval: The argument for getting in quantum would be to say, the cost of doing quantum right now is not high. It's an operating expense, it's not a capital expense. Right? You hire a few people, you give them a little bit of a cloud budget, maybe some consulting budget, and that's it. You don't have to build a factory right now or a new data center or anything like that. The cost of not doing quantum is potentially nothing but also potentially that you are so far behind your competitors that wow, it's like not having gone into the web in time or not having done machine learning in time. It might become a must for the business. Where do you stand on that spectrum of do or not do?

James: There's definitely an opportunity cost that comes with this. And my stance on this is I'm a cheerful pessimist. I think there's definitely going to be a advantage that quantum computers can provide, but we might not actually see that this is an advantage provided by quantum computers until it's been in the market for maybe a year or more. Because if you have a giant competitive advantage, or even a moderate competitive advantage, I don't think you're going to publicize that right out. And yes, there's the counter argument of, "science is done in public."

And I say that a lot, but when it comes to dollars and cents, I think if you've got a business that's identified a problem and they've been able to deploy a quantum computer in production to aid in solving that problem, and if they're able to save money or make more money doing that, they might be a little more tight-lipped about it than I think everyone would want them to be. So, yes, I think the danger is very real that if you aren't getting into quantum computing today, you could be, you know, overtaken by a competitor because of the advantage that a quantum computer or a quantum-realized society - that's so vague of a term that I shouldn't have said it, but it's an interesting one - that a quantum future can provide.

Yuval: I want to ask you two quick questions about things that you're doing today and a little bit on your past. Today, I think you're part of those three tenors of quantum analyst, or maybe you guys are five tenors, right? How's that going? And how much variance is there in the opinion between the five participants? I think this is Andre Konig’s program.

James: Yeah. It's a really fun experience to work with this team, and if we agreed on everything, I think it would be really boring as a podcast. You have to have a little bit of conflict to produce good TV, so it's fun to do. We have different opinions that are informed by the fields that we come from. And I know that that Andre's been on your podcast as well. He's less fond of the idea of different, you know, qubit architectures making that huge of a difference. I think that having all of these qubit architectures is going to drive the science faster, so I'm really fond of, you know, the sort of jockeying the horse race of all of these different qubit architectures because I think it's pretty interesting, and I think that competition, in mostly positive but occasionally negative ways, is driving competition and just making the field more interesting and motivating a lot of people.

Yuval: I think it is interesting, and it's not just horse race. I think it also could influence the strategic choices that end users make. Whether they commit to a fully integrated top to bottom architecture, or whether they want to take best of breed components because they may change over time. So I think you bring a super interesting point and to me, it's beyond just, "I've got more qubits than you." The other thing I wanted to ask you, you said you were a journalist before you became an analyst. And there was a time where the media picked up on the stories, "Oh, quantum's going to change the world," and "We're going to do quantum for chemistry," and "We're going to do quantum for finance," and "I'm going to have a quantum computer in my pocket," and what have you. Do you think that has changed as far - I know you're not in directly in the media right now - what kind of stories do you think - we're not going to tell anyone, it's just you and I speaking and the audience listening - what kind of stories do you think the media is looking for right now on quantum computing?

James: I've seen a lot greater understanding of quantum computing the last couple of years compared to seeing what other journalists, you know, when I was actually a journalist, what my competition was doing, what they were writing, I think the understanding has improved a lot and there are still the weird questions like mobile, quantum computing. That's not a thing. That's not going to be a thing. So the general understanding has improved. I think, rather than asking what stories are they looking for, where the missed stories are, I'm really curious what the first case of quantum piracy is going to be. You talked about, you know, are companies going to want to go with just one vendor or try a best of breed approach? And there's all these quantum software consultancies that will help you on your quantum journey.

And I wonder, if you work with a company that helps you develop an algorithm that works on one quantum computer, if you end your relationship with that company and take that idea, take that work and apply it to a different quantum computer, what's the implication of that? At the time these contracts are defined, who owns the intellectual property that's developed as part of these professional engagements? Especially when you have full stack companies that have their own hardware and their professional software team, I am really curious what happens if you take something from say a vendor blue to vendor orange or to vendor green, and yeah, there could be a little bit of work in translating something from one qubit architecture to another, but this is all things that I think are, for the most part, untested.

And so I'm really curious to see, is this going to be some sort of, you know, public spat in the future? Our company is really thinking about this and who's going to really control the value that quantum computers create. And there's also the aspect of this, which is as a quantum computing company, if you're building hardware, do you want to be a provider of hardware? Do you want to be the next Intel? Or do you want to be a software company that happens to have a hardware backend? And those are two very different questions.

Yuval: I never thought about calling it quantum piracy, but maybe I was just not as imaginative as you are, but I think you bring up an excellent point. We see that a lot of companies think that quantum is strategic, and therefore, they do want to develop their internal competencies. They don't want to just outsource a hundred percent to someone else. This is not like painting my house, I'll just ask someone to do it every 10 years, and we'll be done. I don't need to learn how to paint myself. And part of the reason they want to build it internally is exactly that IP. It's very difficult to ask a consultant to unlearn something that they learned how to do, how to do better option pricing or how to do better chemical simulation when they work with you. And it's easier to keep that IP and knowledge internally. So I hope I understood you correctly, but I think we're in perfect agreement on that point. James, this has been lots of fun. How can people get in touch with you to learn more about your work?

James: So you can go to 451research.com. My email is james.sanders@spglobal.com or you can catch me on Twitter @JAS_NP.

Yuval: That's perfect. Thank you so much for joining with me today.

James: Thanks for having me.

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.

量子ソフトウェア開発を開始

お問い合わせ