Will Oliver, a professor at MIT, and Steve Suarez, an innovation advisor and CEO of HorizonX Consulting. are interviewed by Yuval Boger. Will elaborates on the challenges and opportunities in quantum computing, particularly the need for 3D integration to efficiently control large arrays of qubits. Steve shares his journey into quantum computing and his advice to innovation leaders. They then discuss quantum’s current state and future prospects, the intersection of quantum computing with AI, the role of cloud-based quantum services, and much more.
Full Transcript
Yuval Boger: Hello Will, hello Steve. Thank you for joining me today.
Steve Suarez: Thank you.
Will Oliver: Yep, good morning.
Yuval: So Will, who are you and what do you do?
Will: Yeah, Yuval, thanks a lot for inviting me to your program. I’m Will Oliver. I’m a professor of electrical engineering and computer science and a professor of physics at MIT. I’ve been working for twenty-plus years on quantum computing, in particular with superconducting qubits, but I also work on high-performance cryogenic technologies and 3D integration that we’re going to need to bring these one day to reality, and I enjoy it very much.
Yuval: And Steve, who are you and what do you do?
Steve: Ah, Yuval. Thank you for having me and good afternoon I’m calling in from London I’ve got varied positions: I’m an external advisor at Bain and Company joining them to help them really push the innovation agenda. And one of those topics happened to be quantum computing so I’m excited to be here with you to talk about that. I’m also sitting on a couple of company boards and one of those companies is a quantum computing company out of Israel called Classiq. And lastly and actually it’s got me really excited right now, I’m launching a new consulting company. On September 5 I’m launching a company called Horizon X. where we’re going to help companies work in all three horizons of innovation and really help drive innovation at scale.
Yuval: So wonderful. Will, you mentioned 3D and that got me curious. Is that about qubit connectivity – a 3D arrangement of qubits or is it something else?
Will: Well, it is all of the above. But, where we start is basically if you think about a large array of qubits, you know hundreds of thousands and more, the question is how do I bring in all the control signals to address those qubits. Today, mostly people are bringing them in from the sides laterally, because we’re not at very large qubit processor sizes yet. But of course, that’s a losing battle, because the wires take up more and more room. And the only way to get the wires there is to spread the qubits further and further apart and that’s not an extensible solution. But what is an extensible solution is to bring those signals in from the third dimension, and so this is literally 3D fabrication integration of the wiring and the control technologies needed to address qubits. You could think of the wiring on the ground floor and the qubits upstairs and the stairways are bringing the signals up and down and that is a much more space-efficient way to do it. And then maybe the last comment would be that once you have that, then, of course, you could start thinking about: maybe I’ll put qubits on the second floor and the third floor, and maybe I’ll bring signals also down from the roof, and so there are lots of opportunities to expand this. But where it starts is with bringing the control signals in from the third dimension.
Yuval: So that would be like vias or blind vias on a PCB that’s on a multilayer PCB that’s done today?
Will: Yeah, that’s right, That’s exactly right.
Yuval: What do you think about qubit connectivity? I mean superconducting qubits have all these advantages – speed and so on – but connectivity and requirements for cooling (putting aside LK99) how do you think that’s going to progress in the future?
Will: Yeah, well great question. I mean you know each of these qubit modalities – whether we’re talking superconducting, trapped ions, you know two of the leaders today, up and coming is neutral atoms, semiconductor qubits, and others – today you know each of them has their own strengths and weaknesses. And, you mentioned many of the strengths of superconducting qubits, and of course, that is one of the reasons that they’re in the lead today. But, one of their challenges is going to be connectivity. Currently, superconducting qubits talk very well to their nearest neighbors. So north, south, east, and west. But if I wanted to go two towns away or three towns away, that’s quite challenging. And so one of our challenges is going to be how do we go beyond simply nearest neighbor connectivity. And, why is it important to do that? Well, we don’t have to, but algorithms can become much more efficient in a hardware sense as well as in a software sense if we can achieve something beyond nearest-neighbor connectivity. And of course, the Holy Grail would be full connectivity, where any one qubit could directly talk to any other qubit in the processor. We probably won’t get there. But I think we can get beyond nearest-neighbor connectivity for some advantage.
Yuval: So Steve you and I are probably the only two people in the quantum industry without a Ph.D. in physics. How did you get into quantum?
Steve: it was really interesting because I had the title of innovation. A lot of people were approaching me and saying: Steve, you got to look at this new technology and quantum computing and it was so far out there. And it wasn’t until Nature magazine published that they reached quantum supremacy, and then there was this conversation in the news about RSA’s security broken. Then it really piqued the attention of a lot of boards within the bank and by nature since I had the title head of innovation they were saying: we’re going to get Steve to come to our boards all around the world and Steve’s going to explain what it is and how we’re protected and what’s going on. All of a sudden I had to really quickly first understand what it is and how it works and and then be able to explain to a board why we should be concerned or why we shouldn’t be concerned and I was happy that I was. Explaining why we shouldn’t be concerned at this time and that you know what are the efforts that we’re going to make to look for it. II spent a lot of time educating myself and I could tell you at the beginning I was really like oh is this really going to work? But as I studied it more I got more involved, I became a believer in this technology and I saw the power that this technology can have and so that caused me throughout the years to invest my time to learn the technology and be able to speak to it and look for opportunities both near term and long term. I got involved in quantum computing which was kind of what brought me to meet Will many years ago at MIT and then I became one of his students really learning around Will and Will’s always been an inspiration to me. Ever since then, I’ve just kind of dived in deeper and gone down the the rabbit hole.
Yuval: You mentioned boards, and I think the shiny new object in boards these days is generative AI and GPT. What do you tell boards when they say, oh, we’re gonna take away budget from quantum and we’re gonna move it to this shiny new object.
Steve: Yeah, unfortunately, it’s happening a lot. And I can tell you that in companies that I’m consulting and working with, they’re actually redirecting those employees and saying, okay, you’re gonna spend half your time on quantum and the other half on AI. And the response I get from a lot of the resources is, I’m not an AI expert, that’s not what I do. I’ve been doing quantum and this is, But I think that there’s this misconception that because I’m smart and I know quantum physics, I can go ahead and now be an AI expert and to drive it.
So I think this whole rise about AI and a lot of boards and senior management getting this sense of a FOMO, the fear that I’m missing out because everybody’s talking about it and there are these amazing things is really driving everybody to say, “Okay, let’s just put all of our resources there.” I don’t think it’s the right approach. I think we have to be sensible and see what we can do. I think that there is an intersection between AI and quantum that we should be looking at. But I don’t think we should be retooling quantum physicists to be going towards AI. I think we should focus the people who do that on the experiments on you know, what are the things that we can derive near term, and then we have really good people in the analytics space in that space that can drive that. But right now it seems like everything is being diverted to Gen AI and that’s what everybody wants to talk about.
Yuval: And Will, people have been talking about quantum and AI as Steve mentioned, but AI models have billions of parameters and don’t look like sort of quantum machine learning with today’s qubits in terms of quantity and fidelity is going to get anywhere near that. In your mind Is quantum AI just a new buzzword to get some more budgets into quantum or is there substance behind it?
Will: Well, there’s been work in quantum/AI for about 15 or 20 years and, in fact, folks here at MIT, including Eddie Farhi along with Seth Lloyd and others, have done a lot of work in this area. You raise a good point, and you know 10-12 years ago, Professor John Preskill from Caltech coined a term called NISQ, which stands for noisy intermediate scale quantum, and the idea at the time that he coined it was: let’s try to find an algorithm with the hardware that we have today that will do a useful and hopefully even commercially useful task, which we can do in the interim while we’re working towards doing full error correction to enable these larger systems that you mentioned with millions or even billions of qubits.
So a lot of work has gone into that and there’s been some back and forth between the quantum and the AI or conventional computing scientists saying, “okay, I’ve got an algorithm that seems to run a little better than yours.” And then they (the conventional computer scientists) band together, and come up with an even better algorithm, and it’s gone back and forth. And it’s currently, I would say, in the classical computer scientists camp. They’ve come up with algorithms that they show are pretty darn efficient. And it’s looking more and more like we really do need to get to an error-corrected machine to really see the commercial advantage of quantum computers. Now, I may be wrong on that. It may go back in the other direction, But a lot of really smart people have been thinking about this for a while, and this is where it stands today.
Now, if you think about billions of parameters for an AI, say machine learning, then one challenge that people are thinking about is, “how do I get all of that information into a quantum computer?”, because quantum computers are run or operated by conventional classical computers. And so there’s, as of yet, no quantum speed up on getting a lot of information into a quantum computer, because it’s classical information. We gather it from the world around us and then we stick it into a quantum computer. So that’s called the data loading problem. And that’s something that a lot of people are thinking about – how to address it, or how to work with it – because we may not be able to work around it.
Yuval: You’ve both been in quantum for quite a few years and I’m curious, you know, like a physicist, I wanna look at the derivative. What do you know today that you didn’t know six months ago? What’s new in your mind, new and exciting in quantum?
Steve: I think it’s not so much that I didn’t know, but I think all the time I’m talking to new people, I’m really understanding the different modalities and how the modalities are progressing. And at a certain point I was giving a presentation and a lot of the education that I received was on superconducting. And I’m talking about, you’re never gonna have a quantum computer in your room and all this other stuff. And in the room, there’s a guy having a modality where he’s actually trying to do quantum through diamonds. And so it’s very interesting to me seeing how all of these people are approaching the problem, trying to get to the same solution using these different modalities. How they’re progressing, how they’re driving forward, and how many people in the world are trying to get this done.
From my end, I find that interesting, but then better yet, finding the commercial uses for it and how people find the advantages or the opportunities today. And I’m really excited about that because I think there’s a lot of creativity, a lot of work being done where we can start finding some near-term ways to apply this technology where we can benefit now, we’re not talking about five, 10, 20 years down the line. And that’s kind of where I like coming in from because if I can find that ability to find that value, it gives more investment into this industry that I believe in, and we can keep growing to see how we can benefit the industry. To see how we can benefit the industry.
Yuval: And Will, what’s your viewpoint?
Will: Yeah, well, I’m very fortunate to be working at a place like MIT where we have just really fantastic students, post-docs, and research scientists. I mean, I’m learning something almost every day that’s new. And yeah, it’s one of the reasons that I’m in this field.
I’m trained as an electrical engineer, by the way. And I was thinking, when I was taking those classes at college that, okay, I’m learning about transistors, but transistors are pretty mature and I could make them yet a little bit better, but gee whiz, wouldn’t it have been awesome to live in the 50s and really do this at the very beginning, at the dawn of classical computing?
And that’s exactly what we’re doing right now in quantum. So I get that excitement every day. Just to answer your question, I think there are things that we’re learning at the fundamental level about quantum mechanics and quantum entanglement that we suspected were true, but we didn’t really know. So, for example, quantum mechanics describes the very small – like electrons and protons, that’s where it started. But, can it work with macroscopic objects that you can see, like electrical circuits? That’s really what a superconducting qubit is.
And the answer from 20 years ago is, yeah, it looks like it works. And then you could ask, well, will it work for 10 of those or a hundred of those? Does it work with something I can see with my eye? And the point is that we are entangling larger and larger systems, they behave quantum mechanically, and that’s very interesting.
The question of what happens when you get to many, many bodies entangled, is it different? If so, how is it different? And that can be quite interesting from an intellectual standpoint. From an engineering standpoint, part of the work that I’ve been doing and contributing to is how are we going to think of this as a system – how are we going to build a larger scale and useful quantum computer? It really draws from many disciplines, not just physics and not just electrical engineering, but much broader – material science, fabrication engineering, et cetera, to really build something of a complexity that may be one of the most complex objects that humankind has ever built.
Yuval: If you think about it from a systems approach, as you were starting to describe, I think many people believe that the quantum processor is not gonna be a standalone processor, but it’s gonna coexist with CPUs and GPUs, and maybe every processor is gonna do something else. We’re unlikely to run Zoom on a quantum computer anytime soon, I believe. Do you think it’s too early to start thinking about that integration? In theory and in practice, do you think HPC managers would say, Hey, quantum is coming, what do I do about it?
Will: You know, it is early, and technical development evolves over time. And so if, to use an analogy, if people in the ’50s had said, look, these computers are really wimpy, what we really wanna do is run Zoom, whatever Zoom was in the 1950s, why don’t we wait until we have that technology and then we’ll do it? And if you think that way, of course, you never developed the technology, because it’s the journey, not the endpoint, that is what gets you there. The endpoint motivates us, but it’s the journey that takes us there.
And so with quantum, it’ll be the same thing. So my answer is yes, we should absolutely be thinking about this, but we should also be aware that engineering is hard. This is hardware and it takes time. And so we don’t want to over-hype this or we could end up in a quantum winter, whether it’s, you know, a dark cold winter or even a shallow one, I think we want to manage expectations properly. But at the same time, we don’t want to come across as wet blankets and say, “this is never gonna work and it’s gonna be 20 or 30 years,” because I don’t believe that that’s true either. We need to be highly engaged, we should work towards the applications that we want and solve the problems. And if we do that, we will get to a quantum future.
The other part of your question was about the necessity of having quantum computers operate in tandem with conventional computers. And that is absolutely true the way that we understand it today. If for no other reason, quantum computers will need error correction. So quantum computers run on qubits, those are the logical elements and they’re quantum mechanical, so they’re faulty. And even though we’re quite proud of ourselves as a community for getting them to error rates of one part in a thousand or one part in 10,000, a transistor in our computers today has an error rate of one part in 10 to the 20th power. So orders and orders of magnitude better.
You could also ask, what do we need? We need error rates of something like one part in a billion to one part in a trillion. And so to make up that difference, we will use quantum error correcting codes. Those codes are operated and implemented by conventional classical computers. And so they will always run in tandem.
Beyond that, there are ideas where a classical computer would run an algorithm, maybe 80, 90% of it, but then poll the quantum computer periodically for some advantage to get an answer back quickly. And lastly, of course, we’re not quantum mechanical objects, we sit down at a computer terminal, and type in a program. So of course a classical computer has to be interfaced with a quantum computer in some way.
Yuval: Steve I wanted to ask you, you’re calling from London, I know you’re a world traveler, you advise many boards in many countries. And one of the things that many countries are doing is starting up these national quantum programs where they say, “We want a computer in country as a way to jumpstart the local quantum ecosystem.” Do you think that’s the right strategy or do you think today quantum computers should still remain on the cloud because their useful lifetime is relatively short, they get obsolete fairly quickly, and so on?
Steve: I mean, I think you said it yourself. I think the capex of buying quantum computing and putting it where you, they’re just in my mind doesn’t really necessitate that cost and that investment to do it. And as technology gets better, why don’t you buy it or use it as a service? And I think that’s where using some of these clouds, and today you might be plugging into 433 qubits. tomorrow, it might be 1000 qubits. And if you’re buying it, you’re still stuck with what you bought and the value of that.
So I think being smart about it and leveraging it and using some of these different technologies. Also, if you buy a certain type of computer, you’re stuck to that modality. And I’m really big on being able to understand all the different modalities and seeing how you use the strengths of each one of those modalities to help you process what you’re trying to process. And there might be better strengths on one for photonics or, you know, trapped ions versus, you know, superconducting. So I definitely wouldn’t, again, I’m not trying to rain on any hardware providers that wanna sell their machines, but if I’m sitting there starting up and this is what we do in innovation, is I want to play around and I wanna try it out without making those big investments. And I think using the cloud and being able to access these technologies through that is a good first step to get engaged.
Yuval: Will, you are a superconducting expert and a quantum expert in general. So other than superconducting, what modality are you excited about?
Will: Well, there are many that I’m excited about because it’s clear that, you know, if we think of this as a marathon, we’re in the first five or 10 miles. There’s still quite a ways to go. And who’s leading today likely will not win the race or may not win the race. I shouldn’t try to guess what’s gonna happen.
But again, if we look back at technologies from the last century, we started with vacuum tubes, and they evolved into bipolar junction transmitters, emitter-coupled logic, CMOS, … Technology evolves over time and I would expect the same to happen in quantum.
So the two that seem to be leading today are superconducting qubits and trapped ions. They’re both quite exciting. Superconducting qubits are electrical circuits controlled by microwaves. And trapped ions, generally speaking, are atoms where they’ve had the outermost electron ripped off, an ion, and they’re trapped using electromagnetic fields on a chip, and then they’re controlled generally by lasers. So different technologies, completely different technologies. But both are doing very, very well today, both in the academic as well as the commercial space.
Some other up-and-coming technologies that are quite exciting include neutral atoms. And those again are based on atoms, and they’re trapped with two counter-propagating lasers.
And they’re very good at quantum simulations and some of the largest quantum computers today have been built using neutral atoms. You know, we’re talking maybe 300 qubits at this point.
Photonic approaches, so using photons, the carriers of light, in integrated circuits looks very promising because, you know, they can rely and do rely very heavily on the existing fabrication infrastructure we have for CMOS, because those foundries are also making integrated photonics. And so with a little tweaking, you can just update it. Now you’ve got a photonic chip.
Semiconductor qubits are also quite interesting because they natively exist in those technologies from the foundries. So built from CMOS or silicon or silicon germanium. And so they right out of the box can leverage the last 50 years of fabrication technology.
So all of these different approaches have pros and cons, which we can go through in more detail if you’d like to, but because they each have problems to address, it’s not clear yet who’s going to be the first winner. But the two that are in the lead are trapped ions and superconducting qubits.
Yuval: Steve, you’re coming off a fantastic tenure as a head of innovation in a very large financial services company. And I wanted to ask you two questions about that. First, if you have a friend who’s now starting to be head of innovation at a large company, what advice would you give him? And two, your new company, Horizon XC, I believe, what is that going to do?
Steve: I think if I were to give advice to anyone running innovation at a large multinational organization, Number one is to make sure you have the right support and engagement from senior management. That’s number one at the top of the house. If you don’t have that, don’t waste your time because you’re not going to be able to do it.
Everybody loves innovation because they think it’s a nice shiny job. And you wouldn’t imagine the amount of times people have told me, “Steve, you’re the luckiest man on earth. You’ve got the best job kind of running innovation.” And innovation is, it’s a hard job, but it’s very satisfying for people that love to drive through it. And you’re gonna hear a lot, there’s gonna be a lot of obstacles, a lot of reasons why you can’t do things.
So I think number one is to make sure you have the engagement and the support from senior management because it will be difficult. Make sure you have the budget that you’re gonna need to drive it. So if senior management says, yes, we love innovation, we drive innovation, they have to commit to it. And to commit to it, which means we’re gonna give you the budget and the room you need to be able to innovate, which means you have to be okay with failure. Because if you’re not failing, you’re not truly innovating. And you have to be able to kind of create that culture to where I’m gonna experiment, I’m gonna try to do it very fast, cheap, and frictionless. And if you give me that opportunity, I can bring new things and really do innovation and bring new things to life. And I would say those are probably the main advice I would give to anybody looking to do this.
Yuval: And your new company. What is it going to do and how can it help others?
Steve: Horizon X Consulting, and the reason I called it Horizon X is that I look at it at the three horizons of innovation. Horizon one is core innovation. Horizon two is new products or things that exist in the world, but you’re bringing in new to your organization. And then horizon three is your moonshots.
This is really the disruptive innovation, new business models. And I’m looking to help organizations make sure they’ve got the right balance between the three horizons and get results out of their innovation agenda. A lot of people have innovation teams. And they’re doing what I call innovation, theater or innovation, cheerleading, but they’re not really driving the value from that type of activity. And what I want to do is help organizations truly drive value out of their innovation agendas. Because we’re constantly driving to find new things, new products, new entrances. And either your organization is standing still, it’s going backward, or it’s moving forward, and you can’t do that without innovating.
Yuval: As we get closer to the end of our conversation today, I wanted to ask you a brief question about ethics and quantum ethics. And Will, you work at MIT, you know, in the name, it says Institute of Technology. Do you guys deal with ethical aspects of quantum or other new technologies?
Will: Yeah, absolutely. I mean, we have a school of humanities and many people therein think very hard about these types of problems, whether, you know, for quantum is the new kid on the block, but for a long time, the ethics of AI, for example, or the ethics going back even further with the recent movie Oppenheimer, right? I think, you know, thinking of the ethics of nuclear power and nuclear weapons and therein. So, absolutely we are thinking very hard about it at MIT, and I know people around the world also. And not just the, you know, the technology ethics, but also maybe related to that is the access to these technologies, the potential for disparity that it may create, e.g., economic disparity. And ideas for how we can avoid the negative consequences of technologies as we develop them.
Yuval: And Steve, you talk with a lot of boards and I’m sure the issue of ethics comes up from time to time. Is quantum ethics any different than AI ethics or is it really just, you know, another one of the same?
Steve: I think it’s another aspect of it. I think ethics should apply to everything from business-technology processes. So from us, it’s nothing new. And I think it’s just understanding the technology and making sure that you can use it for a force of good.
And I think I could tell you the reason I’m personally really interested in this technology is that I think it can have, if done right, a significant impact on things like climate change. I’ve got three boys. I want them to live in a better world than I’ve lived. And if I can say, look, I’m getting engaged in something that could have an impact on this world, I think that’s kind of what excites me of people like Will allowing me to kind of get into this industry, allowing me to kind of help push quantum forward to me excites me and looking at how we make things more secure.
I know Will was talking about NISQ with a Q and looking at what NIST with a T, so the National Institute of Standards and Technology is doing around creating security, I think is key. And I think that’s where quantum comes in. So I’m actually really interested in kind of how quantum AI and cyber may come together and have an impact and be able to get into that early enough where I can add value.
Yuval: The last question I like to ask my guests is a hypothetical about if you could have dinner with one of the quantum greats dead or alive, who would that be? Now I know Will for you it’s a little bit of a tricky question because I think that some of my listeners would like to have dinner with you. But putting that aside, who would you want to have dinner with?
Will: Oh boy, that’s a really interesting question and one that I haven’t really thought about before. But you know, off the cuff, I think it would be very interesting to me to have dinner with Niels Bohr, and the reason is that he formed many of the foundations for the intuition of how we think about quantum and quantum measurement. And I think it would be very interesting to have dinner and pick his brain and hear his thoughts on the subject.
Yuval: And Steve how about you?
Steve: I am going to maybe go boring. I’d go with Albert Einstein. And I think when he talks about quantum, I think he described it as spooky science at a distance, or something like that. And I might not be quoting it perfectly. But the fact that Albert Einstein, this great mind that understood so many things, called it spooky science at a distance. Even him, he had a hard time conceptualizing or understanding, and putting this together. And I’d love to maybe pick his brains to understand why you think it was spooky science or, you know, what are, I don’t know. That’d be great. I think I’d probably just enjoy listening to whatever Albert Einstein said. I don’t think I’d understand it, but just to be in his presence, would be pretty cool on my side.
Yuval: I think so too. Steve, Will thank you so much for joining me today.
Will: Thank you.
Steve: Thank you.