← All episodes
Bob Sutor

Episode · Sep 21, 2022

Bob Sutor, ColdQuanta

Why quantum sensors may beat quantum computers to market

Summary

Dr. Bob Sutor, VP of Corporate Development at ColdQuanta is interviewed by Yuval Boger. Bob spoke about leaving IBM, what he has learned about the quantum market in the past year, whether quantum sensors will become mainstream before quantum computing, and much more.

Key takeaways

  • Quantum sensors built on the same cold atom techniques as quantum computers face fewer engineering hurdles, so they can reach commercial use sooner
  • Leaving a large, established program at IBM for a smaller neutral atom company changed how Bob sees the pace and pressures of the quantum market
  • Customers care more about solving a concrete problem than about the underlying qubit technology, which shifts how vendors should pitch quantum hardware
  • The gap between quantum computing hype and near term deliverables has narrowed his own expectations about which applications will show value first

Transcript

Yuval Boger: Hello, Dr. Bob, and thanks for joining me today.

Bob Sutor: Pleased to be here. I'm glad to be back, actually.

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

Bob: Well, I'm Bob Sutor. I'm somebody who's been in the computer industry about 40 years, as it turns out. I guess I must have started very, very young.

I work in quantum computing. For a long time I worked for IBM, in particular IBM Research, and then over the last, let's say, half decade, in IBM Quantum. But I left. I retired, actually, from IBM at the end of February, and one minute later I started working for ColdQuanta, which is a quantum company in Boulder, Colorado, with other sites in Chicago; Madison, Wisconsin; and the UK.

And I'm the VP of Corporate Development there.

Yuval: So IBM employs over 200,000 people, and I think ColdQuanta maybe around 200. So that's about a 1,000-to-one ratio. How do you feel about the move, and what advice do you have for people who make this move from large companies to smaller ones?

Bob: It's been a tremendous amount of fun.

Let me say that through the years, as I've talked to people who maybe tried to attract me to smaller companies, they seemed to have this assumption that at very large companies like IBM, you have all these minions running around doing everything for you. And they say, "Well, you know, in a startup you have to climb under the desk and do what's necessary," and things like this.

Well, guess what? You've got to do that in large companies today as well. This certainly is an opportunity, though, as you said, with this scaling of 1,000, to get to know pretty much everybody in the company, and to really always feel you're at the center of what's going on. Now, IBM Quantum was like that, right?

I mean, it was a few hundred people, let's say, but not all of IBM. I did know probably a few thousand people by the time I was done. But what I'm especially enjoying is living at the core of the startup as it devises its strategy and looks for the best way to use the technology.

I've been working with investors, for example, which is not something I did at IBM. That's a very different type of world. So it's been very invigorating in many ways, including intellectually.

Yuval: And at IBM, these were superconducting qubits, and at ColdQuanta, cold atoms, I think, and then ColdQuanta does other things.

So how does corporate development relate to all these various things that ColdQuanta is doing?

Bob: Well, corporate development's an interesting title, and we devised it as a little bit of, I wouldn't say a catch-all, but kind of an umbrella type of title. That is, yes, ColdQuanta is small, but ColdQuanta doesn't want to stay small.

ColdQuanta wants to grow. ColdQuanta wants to be successful. So part of the reason I'm there is to bring a big-company perspective to this company as it grows. When you are a large company, you do some things very well, and frankly, you do some things not so well, right? There could be extra bureaucracy, there could be slow decision-making, any number of things.

It's the human condition. So how can ColdQuanta grow successfully in this area of quantum, not just quantum computing, but things like quantum sensing and so forth, while avoiding the pitfalls that some large companies have found themselves in? ColdQuanta's very much part of the quantum ecosystem, and I'm here to encourage that, to grow that, and to help them produce great products.

Yuval: As you've mentioned, I've interviewed you before, I think about a year ago, so welcome back, and thank you for doing this again. What have you learned about the quantum market in this past year since we last officially spoke?

Bob: Well, there's a life cycle to it, so let me describe that. When we started seriously talking about quantum computing and future commercialization, we were very clear that this was going to take many years.

Years, and years, and years, and there were going to be little steps, and there would be different breakthroughs, but it would take a while to get to what some people call quantum advantage, or practical quantum advantage, or practical business quantum advantage. And I think people understood that. They spent a lot of time learning about quantum.

But I sense that we've hit this period where people keep asking, "Are we there yet?" It's almost like if you have children and you've ever taken them on a car trip. You say, "Okay, look, this is going to take three hours. We can't get there any quicker. I've given you what you need in the backseat of the car."

And then after 15 minutes they say, "Are we there yet?" We say, "No, we said it was going to take three hours to get there." And then they say, "But I want to be there yet." So what people want, understandably, is quantum advantage. So we're in what I'll call the middle period, right? The beginning period was, let's say, the first five or six years, and now we're in this middle period.

There's an expansion of the understanding of the rate and pace at which quantum computing will become available and will become useful. There's an expansion of the ecosystem and of education. There's maybe a little bit of a resetting of expectations of when it will be here. Now, those expectations may originally have been set unreasonably, but people are getting a more concrete sense of where we are and what these technologies will be used for.

And they're also starting to ask other questions. I think a lot of people view quantum computers as future supercomputers. Now, if you look around at the various supercomputers around the world, these are big. They take up a lot of room, right? They take up a lot of energy. And yes, they're great for solving very hard problems.

But we all use computers in our daily lives. One of the recurring facts about the computer industry is that hardware keeps getting smaller and smaller and more powerful. This was Moore's law. So there's no reason not to think, as long as we're thinking about the future and use cases, that quantum computers will get smaller and they will spread out and become more ubiquitous.

So we can ask more serious questions, not just, "Well, what is quantum computing in a data center?" but, "What does quantum computing mean at the edge?" Wow, okay, now we start talking about what those applications are. What are those use cases? How are they different from the supercomputer ones? Do they involve machine learning, for example?

So in this middle period, what I've learned, to answer your question, is that while progress is steady, the depth of understanding and the depth of thought are increasing around what quantum computing could be as we develop all this technology.

Yuval: And it seems that we've shifted from whether it will ever work to when it will be good enough.

So what's your answer on when?

Bob: Well, I used to describe it this way, and I think it's still valid. Let's talk about this notion of quantum advantage, because that's related to the when. What I'm going to term quantum advantage is when the combination of classical computers and quantum computers can do better than classical systems alone.

All right? Some people might throw in the words "significantly better," but I'm okay with just saying "better" to start with. So I think in the first phase, what we will see in the next two or three years are some really arcane, gorpy examples where quantum computers plus classical computers can do some interesting things, maybe in chemistry, things that we just can't do classically.

Now, people might say, "Well, how does this affect my everyday life?" And the answer is, it doesn't. But it's good progress along the way, and it proves certain points. The second phase, which I think will be in, let's say, the five years after that, so I started with two or three years, and now I'm saying five years, and maybe five and a little bit more, is when we start seeing applications of quantum computing for very specific industry verticals.

And I'm emphasizing the word "computing," and I'm also emphasizing the word "calculation," because quantum computers probably still will not be big enough to deal with large data sets in that second phase. In the third phase, we'll start to see error correction and fault tolerance, we'll start to see quantum memory, and we'll start to see far more widespread use.

Now, that will dribble out over, I guess, what am I up to? Eight or nine years. So beyond that point, 10 years or so, those are the types of things we'll see.

Yuval: So let's pick up on one of the industries you mentioned as an early adopter, where some advantage will come: chemicals. Let's say I'm the CIO, CTO, or CEO of a chemical or pharmaceutical company.

When should I get started with quantum? And related to that, what kind of people should I hire? Should I hire quantum information science PhDs? Should I upskill my existing people? Should I bring in McKinsey or Deloitte? When and how is the best way to go about it, in your opinion?

Bob: First of all, we're speaking very generically here.

You need to answer the question: Will quantum computing be relevant to me? If I'm a food manufacturer, just a straight-up food manufacturer, maybe not so much. If I am actually devising new chemical processes for certain things, though, then I could say, yes, quantum computing. I might ask, do I use high-performance computing right now?

What are the bottlenecks there? Can quantum computing help? So before you do anything, and once you get past the allure of this word "quantum," you have to ask, is it ever going to be relevant to me? And if so, where? And then you can start asking the when, the how, and the who, and things like this.

I think there's a lot of latent talent out there. In my previous life, when we looked around for quantum talent, quantum wasn't necessarily the first thing on people's resumes. But if you go back a little bit, there are an awful lot of physics PhDs out there who may be doing other things, including in business already.

I remember meeting a CFO of a company who was a physics PhD, right? Now, I think he enjoyed being a CFO, but there's this latent talent, the people you may already have in your organization. So if you don't have the talent inside to guide you on quantum, yes, by all means use the consultancies. Use some of the organizations you mentioned to come up to speed and understand where and possibly when.

And then you've got your decisions about skills development and so forth. The educational system itself is shifting. Three or four years ago, most of the people I would say were in quantum computing were physics PhDs. And I used to advise young people, when they would ask, "Well, what should I major in?"

Back then I said, major in physics, minor in computer science. I think we're starting to switch that: major in computer science, minor in physics, right? As we see in most technologies. So yes, begin by understanding your core strategy, and then worry about the milestones and the people.

Yuval: And you touched on education.

Your former company, IBM, does a whole lot of education, I think Qiskit summer schools and challenges and so on. Do you see that as the primary path companies should take to educate the market, or should there be more of, say, quantum for optimization experts or quantum for chemistry experts? Or, as you mentioned, your major is in something else that's not quantum, and now we need to bring you up to some quantum knowledge, not necessarily at the gate level, so you can ask, "How can I solve these problems using quantum computers?"

Bob: I think if you are a serious optimization practitioner, let's say, yes, you certainly should be understanding quantum computing. Now, let me say it's not a foregone conclusion that all these optimization techniques people are speaking about with quantum, either using variational techniques today or even eventual fault-tolerant methods, will be better.

Because remember, quantum computing is only going to really be a serious contender for problems that are too hard classically, right? If it's a relatively easy problem, just do it the way you're doing it, if it's being done in time. So the problems we will be looking at will be hard, maybe very hard problems.

So the setup, if you will, just to run the problem on a quantum computer may dominate the time it's going to take, and what is the trade-off there? So if you are an optimization expert, yes, track that, right? And in the same way, if you are into the computational aspects of logistics, which is a form of optimization in many ways, right?

Yes, you would want to do that. I think quantum chemists are already there, honestly. I think they are probably the closest to quantum computing and how it's being used. The descriptions of the chemistry algorithms are completely within their toolkits of what they do already. So yes, learn what these things are.

Now, there's one other area you didn't mention, which is AI and machine learning. That's a tricky one, because a lot has been written and people are doing a lot of things, but we really have to admit that no one is changing the world of machine learning with 10 qubits or 20 qubits or 30 qubits, right?

We're going to need an awful lot of qubits. We're going to need a serious machine to actually do much better than we can now. And for some parts of machine learning, we will again need things like quantum memory, right? To deal with large amounts of data. We're going to need much better coherence times.

We may need fault tolerance. So don't judge how close we are to commercialization by the sheer number of papers on the topic. Look at the scaling factor, right? Are the systems big enough to solve the problems that are of interest to you? That's a very different question.

Yuval: As we get close to the end of our conversation today, I wanted to ask you a couple of questions about ColdQuanta.

ColdQuanta is doing both computing and sensing. Is that just a byproduct of the same core technology, or do you see an overlap between the ability to make high-performance quantum sensors and high-performance quantum computers?

Bob: Very much an overlap. ColdQuanta started in 2007. One of the founders, Professor Dana Anderson of the University of Colorado Boulder, is still part of the company.

He is the CTO. The idea is that we use natural cold atoms for our qubits or within our sensing apparatus or sensing devices. It turns out you can do a whole lot with atoms and lasers. Our CEO likes to say, "Well, we shoot lasers at small things," and those small things happen to be atoms.

If you put those atoms together into an array using lasers, well, now we think of an array of qubits, and you start using things like Rydberg atoms and Rydberg effects to do two-qubit gates, things like this. So all this language translates over to very natural properties of atoms. Now, atoms are good as qubits because we don't have to manufacture them.

There are no manufacturing defects in the qubits themselves, but we still have to control them. We still have to have this laser technology. And this is going to be the same whether we are using these atoms for sensors, such as inertial sensors of some type. We could use them for atomic clocks. We could even use them as RF antennas, very sensitive RF antennas.

So there's this duality. With quantum computing with atoms, you want a very pristine environment, as you always do with quantum computing. You don't want other quantum effects to disturb your computation. But with sensing, you want to let the outside world in. You want to tell how fast you're going, right?

You want to be able to compute things for time and, ultimately, positioning and things like this. So it's the same technology at its core. Different applications may use different atoms. Now, one thing I do want to mention about lasers is that, the good thing is, lasers aren't just used for cold atoms or indeed things like ion trap technology or photonic technology.

There's an extraordinary industry built around lasers: lidar for cars, barcode readers, and, if you remember, DVD or Blu-ray, things like this. Lasers are pervasive throughout many, many different technologies. So this idea of producing less and less expensive and smaller and smaller lasers and integrated photonic technologies that will be valuable in many industries will feed completely into the use of cold atoms for computing or sensing.

So we are benefiting from the millions of dollars of investment around the world in these other technologies. It's the attractiveness of these natural qubits, if you will — I sometimes joke and call them organic qubits — these natural qubits, or using the atoms as sensors, coupled with this cost-down, scale-down work being done on lasers, that I think will ultimately make this technology successful.

And I think it's extremely versatile, and frankly, I think it's much more versatile than some of the other qubit technologies out there.

Yuval: Most of the people I speak with, and perhaps that's because I come from the computing side, think that ultimately the market potential for quantum computers is going to be much larger than for quantum sensing, in terms of total addressable market and dollar value.

Do you agree? And do you also see this in the short term? We spoke about the timeline for quantum advantage. Will it be the case that in the next couple of years there's actually going to be much more business around quantum sensors, and then ultimately quantum computers will take that over?

How do you see that?

Bob: The numbers I've seen say that if you look at the total market for what we'll call sensors, and remember, we're throwing a lot of things in here. We're throwing in atomic clocks, we're throwing in various types of inertial sensors, right? We're throwing in potential RF applications.

That total market will develop to be roughly the same size as the quantum computing market. It's just that when you pick it apart and say, "Well, this bit over here and that bit over here," you have to add it all up. And in the same way, with quantum computing, you have to ask fundamentally, what is the business model?

Am I just selling time on a cloud, right? Am I selling devices, big, big systems? Or am I going to shrink those down and sell smaller systems and maintain them, or different things like this? So really, I would say even the business model for quantum computing and how people will get paid is evolving, and the final sweet spot is not a foregone conclusion by any means.

So I think they are both quite valuable, but I think we will see quantum sensors become much more prevalent and in production much faster than quantum computing. They're smaller. You don't need 100,000 atoms in a sensor, right? You can make do with far fewer than that.

Whereas you will, in fact, ultimately need thousands, tens of thousands, hundreds of thousands of qubits in a quantum computer that can do something better. So quantum, as it relates to sensors, will be far more immediate. I do think, though, that the market is likely to go through governments first.

I mean, who uses sensors, right? Where do we need sensors? It'll go through government and then a later step of commercialization, although you will see some industries, I think, like automotive, picking up quantum sensors soon. I think it's easy to think of quantum computing as one thing, but quantum sensing is lots of things, and you have to look at it in total.

Yuval: Excellent. So Bob, thank you so much for joining me today. But before we go, how can people get in touch with you to learn more about your work?

Bob: I think the best way is on LinkedIn. I've got a great network of people interested in quantum and all sorts of other things, and I'd love to meet more people who are part of the ecosystem.

Thank you again for having me.

Yuval: Thank you for this second time, and I look forward to the third.