# Talking Quantum Computing with Dr. Bob

Dr. Robert Sutor, a leading technologist at IBM, shares the exciting advances in quantum computing.

Announcer: Welcome to Not Your Father’s Data Center Podcast, brought to you by Compass Data Centers, we build for what’s next. Now, here’s your host, Raymond Hawkins.

Raymond Hawkins: Welcome to another edition of Not Your Father’s Data Center. I’m your host, Raymond Hawkins. Today, we are recording on Wednesday, April 14th. As we continue to pull out of a global pandemic, happy to haw with us today 30-plus-year IBM veteran Dr. Robert Sutor. He is the Chief Quantum Exponent inside of IBM. Dr. Bob, good to have you today.

Robert Sutor: Thank you. Very glad to be here.

Raymond Hawkins: Tell us real quick before we finish introing, what is the Chief Quantum Exponent?

Robert Sutor: Well, sometimes you just have to make up a title when you’re going new things. My training is as a mathematician. I’m a longtime IBM executive, both on the business side, but really primarily in IBM research into stints. Over the last few years, I’ve been moving away from managing people to talking more about technology. In 2019, the end of 2019, I published a bok called Dancing with Qubit, which is an introduction to quantum computing, with a lot of math, but I bring you along on that.

Robert Sutor: I discovered I just really, really like writing. I really like talking and I like somehow making these new technologies make sense to people, so whether it’s through analogies are just, as I said, maybe teaching a little math along the way, helping people understand what they really are so when they’re bombarded with all of this information in the news and in marketing and things like this, they have a little bit of a foundation.

Robert Sutor: We’re trying to come up with a new title here and a lot of times people talk about brand evangelists, and that didn’t quite seem right. Well, it turns out that concept of an Exponent is really important quantum computing, and people talk about exponential technologies, so I jokingly say, “Well, you can’t have an exponential technology without an exponent.” That’s me-

Raymond Hawkins: That’s right.

Robert Sutor: … and I’m the chief one.

Raymond Hawkins: A person who believes and promotes the truth about an idea, so you are the official Quantum Computing Exponent. I like it. All right. Good [crosstalk 00:02:35] stuff. Well, I’m looking live as we record at Dancing with Qubits on Amazon. Comes with a four-and-a-half-star rating, so highly rated. I do have to ask, I love books. I’ll probably push the definition of a bibliophile, but I’m a big hardback guy. Did you ever do it in hardback? I see I can get it in Kindle and paperback.

Robert Sutor: I’ve been talking to the publisher about that and talking in where… I believe plans are underway, but I’m looking forward to that, too. I, too, enjoy having a good hardback book in my hands and particularly in my library.

Raymond Hawkins: Yeah, implied shelf life. There’s a whole section of my library of signed books from the author, whether they are people I’ve met in person and know that are friends of mine and have written books, so would love to if the publisher goes that route. If they don’t, I’m still going to buy a paperback and talk you into signing it, Dr. Bob. All right, let’s get rolling. So-

Robert Sutor: Okay.

Raymond Hawkins: … as our listeners now, we do trivia here. We give out four trivia questions. We’ll do three at the beginning and we’ll do one at the end. In honor of Dr. Bob’s experience at IBM, they will be all IBM-related trivia questions. We’re not giving you the answers today. We’re just giving you the questions, and contrary to popular belief, IBM does not stand for I’ve Been Moved, so the easy trivia question number one is, email us and tell us what IBM stands for. Trivia question number two, tell us what year IBM was founded, and trivia question number three, tell us what IBM’s original name was. It was not IBM. Those are our three trivia questions. We’ll get you the fourth one at the end of the show. Thank you for listening to us, and now let’s get into quantum computing.

Raymond Hawkins: Dr. Bob, before we get into quantum computing, can you back up, and I don’t want to use the right term, whether we call it traditional computing or legacy computing, what’s the right way to think about the way we do computing in an 8088/8086/x86 world where we’re running processing through a microchip? How do we think about that world? Can you set us up a little bit with that before we switch to explaining to people what a qubit is?

Robert Sutor: Sure, and as well as… I mean, not just the Intel line and now, of course, AMD, but the processors and mainframes, IBM’s [inaudible 00:04:58], the-

Raymond Hawkins: Yeah, risk-based processors, too, as well. Yeah, absolutely, yeah.

Robert Sutor: They’re all of a family and the term we use is classical computing. It’s the one that seems to have stuck. I know what you meant when you said traditional computing. This is the type of computing that goes back to roughly the mid-1940s and the people you may have heard of like John von Neumann. In fact, we talk about the von Neumann architecture for these classical computers. You are surrounded by them. The processor that’s in your phone, that’s in your laptop, desktop, maybe thermostat, many of them in your car, these are all classical computers. One way of thinking about it is the information they process, and that all boils down to zeroes and ones. Lots and lots and lots of zeroes and ones.

Robert Sutor: Just to kind of build it up, when you have… That’s a bit is a zero-one. When you have eight bits, you get a byte, and when you get a million of those, you get a megabyte, and then you keep moving on up to gigabytes and petabytes and terabytes and so forth and things like that. Well, I guess I reversed terabytes and petabytes, but it’s all zeroes and ones and it’s absolutely astounding the infrastructure we have built up from what you actually do in the processor to what you’re doing in the onboard memory.

Robert Sutor: Now, of course, the data storage in all of its different forms all comes down to pushing zeroes and ones around in very specific ways. At the low level, they’re what we call logic gates and a lot of it looks like true and false. You look at zero and one and you look at zero or one, and so they have all of these common names, Exclusive ORs. Just a handful of them, but from that we built up addition and multiplication and eventually get to C and we get to Python and PHP and Java and things like that, higher level and higher level languages.

Robert Sutor: It is all classical computing and for many years we had Moore’s Law, which was a hardware statement which roughly said, and you can pose this in different ways and you can put different timescales, but roughly every 18 to 24 months we could jam twice as many transistors on the chip as we used to be able to. What that means is, well, we double the functionality, but if we only did that the chips would just be getting bigger and bigger and bigger and a single chip would take up a room or a house or a football field or something like that.

Robert Sutor: Well, what also happened with this is that the chips every two years were getting half as big, so we were reducing the size and we were also reducing math energy they used. That’s why your iPhone or your Android would have been a supercomputer 30 years ago. It uses very little energy, the amount of storage, of memory, of computing capacity is ridiculously large compared to what had. That was Moore’s Law and, now, there’s always the debate, “Is Moore’s Law still active? Has it slowed down? Is it asleep? Is it dead or whatever?” The fact is, with classical computing, clever people will always figure out ways of thinking [inaudible 00:08:20], but is it enough?

Robert Sutor: Well, the theory says that anything you want to compute, don’t worry, we can compute. We have the so-called “fifth theory” with touring machines and all of the things, theoretical computer science. When you translate that to practice, yes, in theory I can compute something, but it’ll take a million years. That’s not quite so practical anymore. It’s the separation of the great hypothetical, if you will, and really building machines to do this. Or, it would require so much memory, all right, think of RAM, that the number of zeroes and ones and the number of bits you would need would be the same as the number of atoms in the Earth. Well, we’re not going to build storage. I don’t think any of your data centers are that large to fit a copy of the Earth.

Raymond Hawkins: It cannot handle that much storage, that’s right.

Robert Sutor: It can’t do that [crosstalk 00:09:21]-

Raymond Hawkins: Yeah, here, here [crosstalk 00:09:22]-

Robert Sutor: … can’t… so therefore, we hit this very practical limit which says, “Are there certain types of problems that we just will never be able to really handle using classical computers? What are those types of problems? Are there alternative computing models that are not simply based on zeroes and ones that maybe will let us tackle those problems? Make them tangible?” Tractable, rather. Quantum computing is one such method and that’s why we’re focusing on it.

Raymond Hawkins: You know, Dr. Bob, you gave the great example in one of the talks I watched of you around the caffeine molecule and this concept of, “How big?” Can you give us a little bit of understanding about that? I thought it was a great analogy going exactly where you’re going as we transition from classical computing to quantum computing, how we understand the scale of just something simple like the caffeine molecule.

Robert Sutor: Molecules and chemistry are always of interest to people when they think about computing, and even not just chemistry but biochemistry and personalized medicine and pharmaceuticals. Of course, with the pandemic, can we use computers to learn more about COVID to help us come up with perhaps additional vaccines or ways of handling it in the future? You have to start small, though. When you talk about antibiotics and you talk about viruses and you talk about DNA and so forth, those are very large, but let’s roll it way back to something that’s a little bit more manageable to think about.

Robert Sutor: I talk about caffeine because wherever I am in the world, either virtually or in person, I can pretty sure somebody there knows what caffeine is. It’s not a very large molecule and if you look at any of the little stick figures, it’s got a few carbons and some more hydrogens and some nitrogens and some oxygens. It’s really not very big. It’s not tiny. Let’s call it the very small side of medium, but here’s the question. If I have this very modest molecule, yet we hear about supercomputers and, as I discussed before Moore’s Law, shouldn’t it be the case that I could take this molecule and exactly model it in a computer? It’s more sophisticated than maybe my just saying exactly model. I mean if you think of all of the ways that caffeine interacts, so how does it really work in your brain?

Raymond Hawkins: Right.

Robert Sutor: What are the molecular reactions? How eventually does what it does keep you awake? What happens within the molecule itself? Molecules, we’re talking about atoms within atoms. We have electrons. Electrons move around according to the laws of quantum mechanics and things like this. Let’s begin, a very simple question, which is, if I were to put a caffeine molecule in a computer, how much storage would I need to represent it? Is it 50 bytes? Is it 90 bytes? Is it a megabyte? A gigabyte?

Robert Sutor: All of these are easily handled by even your phone. Well, this is what I was hinting at before when I was talking about the number of atoms in the Earth. It turns out to represent one caffeine molecule at one instant in time, and here we’re just talking, if you will, about the electronic structure where it’s electrons and energy and things like this. It would take in number the same size as roughly between 1 and 10% of the atoms in the Earth. To give you a number, that’s about 10 to the 48th bits, zeroes or ones, so imagine one with 44 [crosstalk 00:13:25]-

Raymond Hawkins: 48?

Robert Sutor: Yeah, one with-

Raymond Hawkins: Holy cow.

Robert Sutor: … 48 zeroes. The number of atoms in the Earth are between 10 to the 49th and 10 to the 50th, so one with 49 zeroes, one with 50 zeroes, so 10 caffeine molecules in the worst case, you would need all of the atoms in the Earth. Think about how many molecules are in your cup of coffee, your soda, your tea, whatever it is and, of course, we’re talking trillions and things like that. We’re never going to do that classically, but to nature, and by nature I mean just the way things work and the molecular reactions in your body and around you and in the planet and the universe, one caffeine molecule is nothing. It’s one of many, many, many, many, so nature harnesses an extraordinary amount of information, unbelievably large and accesses a computer in some sense. It controls the processes.

Robert Sutor: Just as we might think of as an application or software running on hardware in a data center, you could think of nature running chemical processes the way molecules interact just to control your own body and, as I said, everything else. That gets you thinking saying, “Well, if nature is a great, big computer, could we build a computer that operates the way nature does? If we could, does that mean we could start to get some of those advantages like being able to deal with things like caffeine?”

Robert Sutor: Well, the answer is yes and the part of physics that we talk about for this part of nature, and there are different parts of physics, is called quantum mechanics. That’s where quantum computing comes from. The quantum mechanics goes way back to the early 1900s and evolved. It’s strange and wonderful and occasionally confusing and it changes the way you think. It’s just marvelous in som many ways and, of course, the potential power as we’re making larger and larger and more powerful machines of the sorts of problems that we’ll be able to tackle, those are quite incredible, too.

Raymond Hawkins: Who do we think of when we think of the original quantum mechanics? Who are some of the original minds that came up with this and help us lay the foundation for what is now becoming quantum computing?

Robert Sutor: They are people like Niels Bohr, Paul Dirac, Walter Heisenberg, Schrodinger. I try to avoid somewhat cliched references to Schrodinger’s cat, but yes, that was actually a thought experiment for quantum mechanics. You know, after a while once people tell you these things a thousand times, it’s like, “Please don’t talk to me about the cat anymore.” Or, if you do, you better know the physics of what he was actually trying to do.

Raymond Hawkins: Right, right.

Robert Sutor: Also, people like Einstein. Einstein wasn’t a big believer in some of the core features of it, in particular there’s something called entanglement, which is the way these quantum systems get correlated, tightly coupled to each other potentially from light years away. It’s a standard process. It happens in your body all of the time. It’s just the way things work. Einstein didn’t like it. A lot of quantum mechanics is fundamentally probabilistic, which means that you can’t necessarily tell something, a certain property, but you can give a probability that it is probably in this state or another. A lot of people are saying, “What do you mean? It’s right there.” I’m saying, “Well, no, it’s not actually right there. It’s got a 75% chance of being there and a slightly different probability of being somewhere else.”

Robert Sutor: The mathematics behind quantum mechanics can be pretty daunting to a lot of people, so with quantum computing and, in particular in my book which you mentioned earlier, I tried to simplify this to the point where people can understand the essentially basic math that goes into this to give them a foundation for quantum computing. Then, they can reason about it as a type of computing as opposed to physics.

Raymond Hawkins: Got it. All right, so we’re transitioning out of classical computing, so I’m going to take the land grant-educated host and take what the Princeton- and Harvard-educated guest is saying. “Hey, frankly, classical computing just doesn’t have the horsepower or the capacity to handle really complex problems.” The great example is nature. Nature handles an enormous amount of information and an enormous amount of calculation and we’ll never be able to replicate it in a classical sense, so there’s got to be another way.

Raymond Hawkins: That transitions us into quantum computing and the concept of, “Hey, is there another way to think about solving problems and looking at problems and studying problems?” That is, “Hey, let’s frankly replicate or look a little bit more like nature looks instead of the ones and zeroes.” Like you said, it’s either zero or it’s one or it’s a combination in a string of eight of ones and zeroes. That’s pretty limiting, but the brute force way of going about it, let’s be a little bit more elegant. Now, we switch into qubits, so with that basic foundation, let’s head down the qubit and quantum computing route.

Robert Sutor: Well, and one thing I want to make clear is, first thing is quantum computers aren’t going to up and replace classical computers. They’re going to work together. Classical computers do many things perfectly well. In fact, you can’t even talk to a quantum computer unless you’ve got several classical computers-

Raymond Hawkins: Great point.

Robert Sutor: … so they’re complementary. The other thing is quantum computing isn’t necessarily suitable for every type of hard problem that we might come across that’s classical computing we have. It’s certain classes of problems. We’re learning much more about that. We’re learning much more about saying, “Hey, we have a feeling that this is amenable to it. We will develop algorithms. We’ll reuse what we have and try to attack that problem.” There certainly are going to be things that quantum’s not going to help you, either. Let’s put it that way.

Raymond Hawkins: Okay, not the answer for everything, but answers for big, complex things. Got it. Okay. At the basic level, let’s start with a qubit and the challenges that come from there and let’s grow from there.

Robert Sutor: Okay, so we started with a bit and a bit, as I said, was zero or one. We’re never going to lose the bits. We’re going to start with bits and we’re going to end it with bits, and in the middle we’re going to use some other things. We’re going to extend the idea to a quantum bit, and that’s where qubit comes from. It’s kind of merger of those two words, quantum bit, and spelled Q-U-B-I-T, just to [crosstalk 00:20:33]-

Raymond Hawkins: Got it. Quantum bit, got it.

Robert Sutor: People often hear about it. Well, a company by its statement, which again bugs me a little bit when they say it because it’s mostly cute if they don’t actually know what they’re talking about, but I’ll explain what it is. That statement is, “A qubit can be zero and one at the same time.” Okay, so let’s dissect that a little bit. Suppose I told you this, so I’m going to translate the problem somewhere else just for a moment. I’m going to say, “Raymond, I want you to walk four blocks north and three blocks east.”

Robert Sutor: You might say, “Well, okay. That’s easy.” When you get there, you call me on your phone. You say, “I’m here.” Then, I’m going to say, “Raymond, the most amazing thing happened. You are north and east at the same time. Isn’t that incredible? Could you imagine anything so special?” You’re saying, “Big deal, okay.”

Raymond Hawkins: Imagine it? I just walked it. Of course I can, yeah.

Robert Sutor: Yeah, and you’re saying, “Okay. Well, this idea of… What I’m essentially saying is two dimensions, walk north and walk east. Walk vertically and horizontally. We get this idea and we get this idea of two coordinates, if you will, so four blocks north, three blocks east. From high school and geometry we could draw lines and parabolas and things like that and xy-coordinates. Quantum computing for one qubit is that with a few extra rules, and so I’m going to replace north and east with things I’m just going to happen to call zero and one. If it helps you not to get confused with the number zero and the number one, write them out, O-N-E, T-W-O. That will replace the words north and east.

Robert Sutor: Now, if I say, “Four blocks one and three blocks east, or three blocks zero,” you’ll say, “Okay, I’m with you. I don’t know why you’re going there, but you know [crosstalk 00:22:42]-

Raymond Hawkins: Gotcha [crosstalk 00:22:42]-

Robert Sutor: … words [inaudible 00:22:43]. I’m going to write them a little differently, so okay, instead of using letters, I’m going to use some fancy notation. In fact, I mentioned Paul Dirac before, beginning part of the 1900s, he came up, what do you call? The bra-ket notation, so I’ll write zero as a vertical bar, a zero, and then a greater-than sign. Mathematicians love to make up new symbols and representations for things. There was nothing handy. He came up with that that allows you to manipulate these. He, in fact, was a theoretical physicist. They’re new symbols and you might see this. Vertical bar zero, greater-than symbol, vertical bar one, greater-than symbol. That’s the zero and one that we’re talking about.

Raymond Hawkins: All right.

Robert Sutor: Okay, so we’re still good. We’re still saying, “Well, you go a certain amount in one direction, a certain amount in the other direction. Well, it turns out that I’m going to put restrictions, and right now I’m not actually going to let you walk four blocks and three blocks. I’m going to make you do some stranger numbers in front of this. I’m going to make you walk square root of three over two in this direction, and one half in that direction. The reason why, if you think of a circle, that point, square root of three over two and one half is actually on a circle a radius one around the origin, the so-called “unit circle.” All of the points next to my zero and one are going to in some sense correspond to points on the unit circle.

Robert Sutor: Raymond, I’m not going to quite let you go the four blocks north and the three blocks east. With quantum computing, I have to put other sorts of restrictions on those numbers, and so, for example, what I’m going to do is for the first number, just in this example, I’m going to choose square root of three over two. Now, bear with me, that number, that seems a little weird, but for the other number I’m going to choose a much more standard number, which is one half.

Robert Sutor: Well, square root of three over two and one half may seem like strange numbers, but they’re not if you look at geometry. Those are points on the unit circle, which you may remember from high school, from trigonometry. You look at all of the points just that live on the circle around the origin and the circles of radius one. There are a whole lot there, an infinite number, and so what I’m really saying is that those numbers in front of my zeroes and ones are going to be restricted in that way. They have to fall on the unit circle.

Raymond Hawkins: That’s an… All right, Bob. I’ll accept that.

Robert Sutor: Why not? It seems like a math [inaudible 00:25:38]. Well, it turns out something else weird is going on here because I went from a bit which was zero and one to a qubit, so remember, I just could be either a zero or a one, but a qubit could be, well, is two numbers as I said. One number’s in front of the zero and one number’s in front of the one. I just said one of them could be square root of three over two and the other one could be one half.

Robert Sutor: Okay, point is a qubit has two pieces of information, two very rich pieces of information and, in fact, it can represent an infinite number of pieces of information. One bit can just be zero or a one, but one qubit can represent an infinite number of pieces of information or the values that it can be. That is what gives you so much power. You’re going from really just zero or one to two whole dimensions worth of space, really.

Raymond Hawkins: I gotcha.

Robert Sutor: Right, and then we puts some [crosstalk 00:26:40] then we have to put some restrictions on this, as I said about the unit circle and things like this. Now, there’s one other thing, which I got to tell you about with this because I said we start with bits and we end with bits. You know, Raymond, I made you walk across the city there and, first, remember four blocks north and then three blocks east. I’m going to make you decide whether you’d like to tell people that you are north of where you started or east. I’m not going to let you, Raymond, say you’re northeast. I’m going to force you assert you are north of where you started or where you are east.

Robert Sutor: I said, “Well, four blocks north and three blocks east,” and you’re going to say, “You know, I’m a little bit more north than east, so probably if I had to choose I would tell people that I’m north.” Somebody else, I might ask… You say, “You know, I really like being east, and I am actually a little bit east, and so what the heck, I’m going to tell people I’m east.” Well, these numbers correspond to probabilities because once we’re done working with qubits, and I haven’t really talked about what we do with them, but once we’re done manipulating them, we have to force them back to being zero or one. These numbers in front of them correspond to probabilities of whether we’ll get zeroes or ones.

Raymond Hawkins: If I roll on back-

Robert Sutor: Right.

Raymond Hawkins: … to first of all the… Let’s just keep it on north and east for a little while, there’ll be more than a 50-50% chance that people will say they’re north versus east. Well, okay. My slightly more mathematical example, square root of three over two and one half, turns out that the probability of choosing that first one is 75% and-

Raymond Hawkins: Oh-

Robert Sutor: … the probability of getting the second one is just one half. For those of you following along with the math, I squared each of those, and that’s how I got those probabilities. Not only is there a geometry to qubits and not only is there this representation of an infinite number of values, but there’s this tie-in with probability, so putting all of this together and adding more qubits, what we’re doing with quantum computing and all of the manipulation is being able to move from just very simple zeroes and ones and lots of those to these gigantic spaces, really huge, huge spaces. In fact, one might even say exponential spaces in which to solve our problems, and the things we do once we do that is we’re nudging the state of the system towards the most likely answer.

Robert Sutor: At the end, we say, “You got to decide, or use zeroes and ones.” This is called measurement and we end up with a sequence of zeroes and ones and there are rules here that tell us that say, “Okay, well, this is a certain probability of being the correct answer. Perhaps we have to repeat the calculations a certain number of times to get to even more confident, but it is this growth with qubits, the amount of information, and every time we put in one more qubit, we double the size of our space. We go from two dimensions to four to eight to 16 to 32. This is why quantum can handle so much. In fact, with 160 qubits, we could represent that caffeine molecule, the one’s that impossible forever, the one-tenth of the Earth.

Raymond Hawkins: Yeah, just can’t get big enough in a traditional computer, right, classical computer.

Robert Sutor: Within a quantum computer with an appropriate amount of work and there’s future work that we have to do that I’m glossing over a little bit here, but the idea is that we’ll be able to actually manipulate problems inside a quantum computer that use a tremendous amount of information.

Raymond Hawkins: All right, so I’m going to go back to the land grant guy from the South trying to dumb it down to make sure I get it. When I think of ones and zeroes, I think of simple like a DIP switch. It’s either zero or one, it’s A or B, it’s up or down, whatever. It’s just this is DIP switch computing and I multiplied across eight to get my byte. What I think I hear you saying is, “Raymond, don’t think of that DIP switch as only having two positions. Think of it of having not countless but a large number of positions in a two-dimensional layout.” I love the unit circle idea of all of the different positions. That’s what I get with a qubit. I don’t get the restriction of a one or a zero, open or shut, A or B. I get a wealth of positions is really what you’re telling me.

Robert Sutor: That’s right, that’s right, and-

Raymond Hawkins: That those positions grow exponentially as I grow my number of qubits in that sequence or processing or computer.

Robert Sutor: That’s right, and another way… I use the word “dimensions” and I do tend to fall back to mathematical terms. Another way of thinking about this, I said a qubit can represent two pieces of information. Two qubit represents four pieces of information. Three, we double that, eight. We double-

Raymond Hawkins: Eight, got it.

Robert Sutor: … that again, so by the time you get to just 10 qubits, that state that you have right there is representing over a thousand pieces of information. In fact, they’re very-

Raymond Hawkins: Whole-

Robert Sutor: … significant numbers. They’re not just zeroes and ones. They’re significant numbers. We’ve estimated that by the time you get up close to 300 qubits, and here I’m talking about slightly future quantum computer. We’re not there quite yet, still a little bit of a theory, but so just roll forward a few years here. By the time you get to about 300 qubits, the amount of information it can process is greater than the number of… corresponds to… greater than the number of atoms in the observable universe.

Raymond Hawkins: At 300?

Robert Sutor: At 300, and by the way, we’ve got to make a lot more of them than 300. We have to get up. These are the so-called “logical qubits.” We were going to need tens of thousands of these and-

Raymond Hawkins: Got it, got it.

Robert Sutor: … so it just blows your mind. This is why at some point you got to let go of these traditional analogies to classical.

Raymond Hawkins: Right, linear ways of thinking of how it compares to classical to me. All right, let’s go down two different tracks, and I’m going to lay them all out because I want you to weave them together however you want. I’ve got two or three questions that burn in my brain as I think about computing and I’m going to lay them out and you take them how you want, Dr. Bob. Question number one, customers say to me all of the time and come to us and ask, “Hey, how does this do this? How do we futureproof it? How big does it need to be because computers keep getting smaller and faster?” Quantum computers are going to be here next summer and when they get here, everything’s going to change, so sort of that. Where are we on the roadmap? Are they coming next summer and going to change the world? That’s question number one.

Raymond Hawkins: Question number two, I don’t understand the temperature component of quantum computing. I understand… I mean, in our business we provide a building, we provide electricity, and then we provide heat rejection to cool the computers. I don’t understand why quantum computers where they have to be so cold, I mean, ridiculous cold, so that question. Then, third, where are we as a race? Where are we in how many qubits we can do? What’s practical? Where do you see it going? That’s three kind of questions all rolled into one, but if you’ll take those three themes and help me understand all three of those.

Robert Sutor: Quantum computers are primarily cloud-based and they are contained in a unit. You might call them a pod, if you will. The electronics and the cooling, which I’ll return to in a moment, are all contained in a unit. The IBM Q System One that we talked about, we’ve seen a few. If you’ve ever seen a photo of this, it’s a glass-enclosed quantum computer. It has a stainless steel cylinder in the front. Just to give you an idea of the dimensions of that, it’s about 9′ x 9′ x 9′, or 3 meters x 3 meters x 3 meters. That’s about the size of it.

Robert Sutor: Most people will want to access quantum computers via the cloud because over the next few years we’ll be significantly increasing their capacity. We will be adding more qubits. The qubits will be better. We’ll be doing miniaturization. We’ll be doing lots of the normal things that happen with hardware. There’s really nothing anyone has to worry about in terms of existing data centers right now. In the middle of a calculation, wherever it’s happening, it could be on the cloud, it could be running in a container, anywhere in the world, you could reach out across the cloud, talk to a quantum computer, get your result back, go on with your business. It’s not going to impact you that much.

Robert Sutor: We don’t anticipate quantum computers being powerful enough to do better than what classical computers can do until roughly mid-decade. I’m fudging on that because they’re going to be small examples of this, but people are going to want to know about their use cases and things like this. This is what we call quantum advantage. It’s a really-

Raymond Hawkins: Mid this decade, Dr. Bob?

Robert Sutor: Yeah, that’s right, and the idea is that you’ve got to be able to get to the point of doing significantly better than classical computers. If you’re just doing about the same, why bother? Just keep doing what you’re doing, and so once we comfortably start doing as I termed it significantly, whatever that means, I’m not defining it, then we’ll start seeing quantum computing, again, integrated into work streams.

Robert Sutor: Now, one thing I will say is that the Cleveland Clinic, we had an announcement with them a couple of weeks ago. We will be installing a quantum computer in Cleveland for their use because they will be doing a lot of direct research related to, of course, healthcare and biochemistry and things like that for which they want full access to. Primarily, most people should think of it as cloud-based.

Robert Sutor: The second thing I believe was about temperature, so I started by talking about atoms and electrons and molecules. Well, there are other things, photons. If you think of a particle of light, a photon, quantum computers work by very careful, very low-energy manipulations of quantum states, and so forgetting for the moment just the cold temperature, if I were to hit a quantum computing chip with one photon, just one photon of light, think of one single photon coming out of a laser, that would have so much energy it would completely destroy the quantum computation. What this means is that any stray RF interference, any variations in temperature that could somehow affect the quantum state inside the physical qubits would disrupt the calculation and cause them to be wrong. It would introduce noise. You could think of it as static.

Robert Sutor: Kind of in a perfect world, what you do is you completely isolate this. You chill it down to close to absolute zero, so this means everything else slows down. The only thing that’s happening is what you are intentionally forcing through that device. Stray, as I said, photons, RF, whatever, just isn’t happening, but you can’t completely isolate it because you have to talk to it, and so you have to expect there’s going to be a certain amount of noise with it. We try to minimize that. We do noise mitigation it’s called, and eventually we will get to the point where we’ll be able to error correction, make it fall tolerant. There’s a lot of research on that. Toward the end of this decade you’ll probably start seeing even some of that.

Robert Sutor: That’s why there are different qubit technologies that say they are more or less tolerant or that you don’t need temperatures this cold or that, but there’s always a gotcha, there’s always a gotcha. The big thing with quantum computers and, as I said, you can build them using different technologies. We do what’s called superconducting. There are ion trap, there are photonic, there are a few others like this. It’s scalability. To be honest with you, I don’t care about anybody who has a two-qubit quantum computer. You can’t do anything using only two qubits. You have to start getting more and more qubits and you have to start breaking a hundred. You have to eventually break a thousand. You have to think of ways of connecting more and more qubits.

Robert Sutor: When you add qubits, it’s not like just adding memory. You don’t walk over and stick in a card and say suddenly your device has more memory. All of the qubits have to work together. You have to have very sophisticated control systems to do this. Our roadmap, the IBM roadmap, is that later this year, in 2021, we will offer a quantum computer that has more than 100 qubits. We will roughly triple that to a little bit more than 400 next year, and for the first time ever we anticipate 2023 we will have over a thousand qubits in a quantum computer.

Robert Sutor: At that point, we will be moving from… If you’re thinking of a classical circuit board, you move from individual chips to multiple chips talking to each other to quantum motherboards, and these will all be kept very cold by virtue of the way the freezers or the refrigerators look. They will be round, and you can imagine them almost being stacked like pizzas with many qubits and control systems that allow us to talk to qubits on one being able to interoperate with the qubits on another. That’s how they will grow. The next three or four years are going to be really critical as we push up and beyond to a thousand qubits.

Raymond Hawkins: Dr. Bob, a couple of quick questions around that. If my Q One today is 9′ x 9′ by 9′ cubed, that’s pretty significant. What do you imagine a thousand-qubit machine looks like? That’s question number one. Question number two, you gave me that notion that at 10 qubits with thousands and thousands of analysis at a thousand qubits, what are we talking about as far as compute power or computational power?

Robert Sutor: Well, I’ll let you do the math. It’s two to a thousand.

Raymond Hawkins: It’s a bunch, yeah.

Robert Sutor: It’s a whole bunch.

Raymond Hawkins: Yeah.

Robert Sutor: Yeah, so it will physically get bigger, so the complete system now pretty much fits in that 9 x 9 x 9. The actual fridge today, so it’s an enclosure that’s kept at vacuum that, as we said, the bottom of it’s kept at absolute zero. You want to think of that cylinder as maybe two feet cross by about four feet high.

Raymond Hawkins: Okay.

Robert Sutor: For the so-called “super fridge,” which will house thousands and thousands of qubits, these so-called “pizza-shaped quantum motherboards,” that you’re going to be thinking about something… just the fridge about five feet across and 10 feet tall.

Raymond Hawkins: Oh, wow. Okay.

Robert Sutor: There’s a huge difference between 27 qubits and thousands or tens or hundreds of thousands. That’s not counting the extra refrigeration units and things like that, so it will take up a little bit of room, but room is not the problem at this point.

Raymond Hawkins: Okay.

Robert Sutor: Even at that, for the extraordinary computational power, that’s cheap space-wise because remember, we’re not just replacing classical computers for the heck of it. These are going to be reserved for the most difficult and certain classes of problems. We’ll need more of them, but you’re not going to imagine filling your entire data center with them and you’re concerned about footprint.

Raymond Hawkins: As I look at the pictures of quantum computers today, I see lots of what I would call coiling or tubing. What’s the right term? What’s going on there?

Robert Sutor: Yeah, if you see them, and I encourage anyone, if you just look IBM Quantum Flicker, there are many, many photos out there if you’re just curious what some of these things look like. There’s something… it looks like a chandelier, first of all, to be clear, a golden chandelier. It’s proper name is a cryostat. Cryostat comes from the Greek meaning cold and stable-

Raymond Hawkins: Cold, yeah.

Robert Sutor: … and the quantum device is actually way, way down at the bottom of that.

Raymond Hawkins: Okay.

Robert Sutor: That is what is enclosed in the dilution refrigerator, the freezer, if you will. Everything you see there is really… it’s either structural to hold it together, it’s either the electronics to control what’s happening in the quantum device, or it has to do with the refrigeration. If you look at one, you can see some golden cables that come down, and then they loop around in a circle and they continue on their way down. Those are microwave control cables and that’s how we talk to the qubits, through microwaves, and so signals are brought down to the qubits and signals are returned from the qubits at the end of the [crosstalk 00:44:13]-

Raymond Hawkins: What does the loop do for us in that path?

Robert Sutor: May I ask just where you’re calling in from geographically?

Raymond Hawkins: Yeah, yeah. Oh, I’m in Dallas, Texas.

Robert Sutor: I’m calling in from Upstate New York. Typically, we have more of a winter than you do, though you had some [crosstalk 00:44:31] this year [crosstalk 00:44:30]-

Raymond Hawkins: We got one this year.

Robert Sutor: … this winter and one this year. If you are in the north, you are very used to seeing on telephones loops of extra cable, and what they are is they’re thermal protection because when temperatures get very low, below freezing, wires shrink.

Raymond Hawkins: Aah.

Robert Sutor: If you didn’t have those loops, the wires would snap, and so you can imagine as we chill that down close to absolute zero, the loops, which are maybe the size of a nickel right now shrink down. I don’t know how small they get, but maybe you could imagine they’re shrinking down to about the size of a dime, right? Because-

Raymond Hawkins: I gotcha, okay.

Robert Sutor: Also, within the cryostat, there are many different types of metals. There are exotic metals, but there is brass and there is gold. It’s actually gold-plated. Gold reflects infrared radiation, which corresponds to heat very well. If you look very carefully, you’ll see some hinges and some other structural devices, so that the entire cryostat as we chill it down from room temperature to close to absolute zero, it just very nicely shrinks slightly and doesn’t rip itself apart [crosstalk 00:45:46]-

Raymond Hawkins: I gotcha. I see. Fascinating. Well, Dr. Bob, this is incredible stuff. I think you debunked a few myths of quantum computing’s going to take over the world and take over the data centers next year, and help us understand a little bit better. I’ll be candid, Dr. Bob, I think if you’d be willing to come and do another episode with us, that I’ve got about 65 more questions that I just didn’t get to.

Robert Sutor: I’m happy to [crosstalk 00:46:09]-

Raymond Hawkins: If you’re willing to join us again, our listeners, there’s sort of three or four continents’ worth of them, but it’d be mostly data center folks and I think they’d be fascinated by what you know and what you understand. I’m going to get us one more IBM-centric trivia question in, and thank you for your time. Then, we’ll get out of here. Once again, thank you for joining us for Not Your Father’s Data Center. Dr. Bob Sutor, the Exponent for Quantum Computing at IBM.

Raymond Hawkins: Our last IBM trivia question. IBM has had five Nobel Prize winners to do qualify for our Amazon $500 Amazon gift card. Answer our first three questions and then give me one IBM Nobel Prize winner. Again, you can email that to me at rhawkins@compassdatacenters.com, or you can tweet us, @compassdcs. Question number one, what year was IBM founded? Question number two, what was its original name? What does IBM stand for? Name any one of five of IBM’s Nobel Prize winners.

Raymond Hawkins: Dr. Bob, you’ve been awesome. This is so, so good. I really, really appreciate it. Would love to have you again. Thank you for demystifying a little bit of quantum computing and stay safe. I think you told me you got both vaccines, so you’re [crosstalk 00:47:19] in good shape, and I hope our whole world is that way soon enough. Thank you.

Robert Sutor: Well, thank you for inviting me, Raymond. It’s been a lot of fun.