#CounterpunchWithSVG E04: IBM’s Love Affair With Quantum Computing Is Age-Old Yet New

3,521 views Reading Time: 12 minutes

Catch Robert S. Sutor (Bob), Vice President – IBM Q Strategy & Ecosystem, in an exclusive conversation with Sanchit Vir Gogia (SVG), Chief Analyst & CEO, Greyhound Research on our executive Q&A series, #CounterpunchWithSVG. As the name suggests, SVG does really pull some punches and then, of course, gives his opponent an equal opportunity to either counterpunch or dodge it in style. Take a read for yourself…

SVG: Bob, what I would love to start with is the comment you made yesterday on LinkedIn. You said that quantum computing involves both quantum and classical computers, and we should avoid the hybrid world of quantum-inspired computing. Now, if you can, please explain yourself a little bit more.

Bob: We’re certainly trying to get across to people exactly what we mean by quantum computing. So, what we call classical computing, which is now embedded in our phones or laptops, servers on the cloud, and so forth, goes back to the mid-1940s. It’s all anyone knows, and so that’s their frame of reference. They have grown up with an understanding that ‘oh look at my watch, it’s a computer’ or ‘look at my car it has 75 processors in it’ or things like this. 

They’re used to having more classical computers to add to their life and their work. This notion of something that’s an entirely different model of computing doesn’t exist for many. Most people don’t think of any other models of computing.

We have this explanatory problem, where we’re dying to compare it to something else when people don’t have a lot of knowledge about what that something else is. So, we have started talking a lot about quantum computing.

For us, at IBM, we’ve been doing quantum information science now for about 50 years. When we first put it up on the web in 2016, that was when it kind of exploded.

But it’s still clear that many people don’t know about the working relationship between a quantum computer and a classical computer. Will a quantum computer replace a classical computer? Will, my phone in four generations, be ultimately quantum? 

I was trying to say on LinkedIn that we tried for a while, stating that quantum computing is this hybrid world where there are some quantum computers and some classical computers. Then this word “hybrid” got a little bit confusing and muddied the waters, and so I wanted to be very specific about this. 

So starting with classical computing, there can be many different types of processors, for example, in our Power systems, Intel chips, AMD, ARM, or they can be our Z chips. It can even be in your phone. That is all classical computing. 

Now what we’re going to do is we’re going to augment that world. And we’re going to add a new type of computer. These are called quantum computers or quantum devices. You’re not replacing the classical world. You’re augmenting it with quantum computers that can do things we can’t do well with classical systems alone. Once you introduce actual quantum devices, real hardware, you get to call yourself quantum computing.

If you are only simulating a quantum computer or doing classical computing but calling it “quantum-inspired,” I’m sorry, you’re not doing quantum computing.

SVG: Thanks, Bob, that was extremely elaborate and very well explained. So, ahead of this call, I reached out to our end-user clients, and one of the questions I got from a CEO who asked, “I’m a business guy, and I quite don’t understand quantum Computing. So, help me understand the principles that will help quantum make a real-life impact on my business”. This question is pertinent because quantum is not a piece of shiny new technology that IBM or Google alone can own. It’s going to be a combination of principles of physics, math, and technology that will come together to make quantum real. So, you know, let’s say if I was that CEO talking to you about the principles that sort of are helping quantum be real, what are those principles?

Bob: So, let me start by saying what we shouldn’t do. Especially for someone who does not see themselves as a scientist or hasn’t delved into how computers work. They maybe have never even taken a computer science class. They want to use an app or use the systems on the cloud to get things done. Now all of a sudden, we have this other phrase (quantum computing) that comes into their discussions.

What we don’t do and shouldn’t do is be so excited about the low-level physics, that we just can’t help ourselves and talk about entanglement and superposition and constructive interference simply because these terms sound smart and sciency to say it out loud. 

One way of looking at this is that nature is one big computer. Everything around us (atoms and molecules) and how everything interacts leads to chemical reactions that ensure results are driven in some way.

Said another way, this world and the universe around us does not operate as a digital computer. To get technical about it, the physical model by which nature operates is called quantum mechanics. We’re trying to say, “Nature, you seem to be able to solve certain problems very well but also classes of problems for optimization or that deal with an incredible amount of information in many different dimensions. Nature, we love those features of yours, and we are working on trying to capture that so we can put them in actual computers paired up with classical systems.”

So, it’s all about augmenting our existing computing capabilities for some problems to emulate how nature does and get to a more powerful computing environment. 

SVG: Interesting. So let’s talk about IBM and its journey with quantum. At what point did this love affair get started? At what point did IBM wake up to this fact that quantum is an area you need to invest it.

Bob: Well, one answer to that is February of 1970. Classical computing goes back to the 1940s. But starting in the 1950s, and especially in the 1960s, people like Claude Shannon started developing what was called information theory. Because of all the work in the early 20th century around quantum mechanics, people started thinking about the basis of a quantum information theory. 

Charlie Bennett, who’s an IBM Fellow, was the first one to write out the phrase quantum information theory in 1970, based on his work and work of others at IBM and other places in the 1960s. So, that was 50 years ago. 

There was a lot of work done in many great research organizations and universities worldwide, including places like Yale, to create the fundamental technology for real quantum computers. By 2016, we at IBM realized we had quantum computers with excellent uptime. We decided to start putting quantum systems on the IBM Cloud in 2016. Since then, over 240,000 people registered to use the 20 systems now available.

SVG: We’ll come to the ecosystem conversation because that’s a conversation I’d love to have with you as well, but for now, let’s talk about qubits. Many of our readers are business technology readers, and they don’t understand the world of qubits well. So, they’re used to this entire binary conversation, because that’s the conversation they’ve appreciated over the years. So, translate one qubit for our readers. What is the power of one qubit if you were to sort of draw some parallels to that same binary world?

Bob: Let’s jump back to the classical world briefly. I just built a PC gaming machine with 64 Gigabytes of memory, What’s a Gigabyte? What’s a byte?

Well, a byte is eight bits, and each bit can only be zero or one. We tend not to think so much about the zeros and ones, but we think of massive numbers of information units, these bytes, and Gigabytes. And we know more is better. It means we can probably compute faster and we can store more videos and all sorts of things like that. Classically we have been able to build all this structure on this straightforward notion of a bit, something only being zero or one. 

At the very lowest level, there’s not a whole lot you can do with one bit. If it’s zero, make it a one. If it’s a one, make it a zero. It’s kind of like a light switch, you can turn it on and off, and that’s it. Now, when you start having more than one bit, you can store a lot more information. You can start building up operations like addition and multiplication and other sorts of things. From the most primitive notions of a zero or one, we get everything we see with classical computing.

I am in New York State in the US, and luckily it’s a sunny day. Now, if I were to go out driving, I would put on my sunglasses, my polarized sunglasses. Well, polarized sunglasses seem to do something better than just not wearing sunglasses at all. The units of light, photons, can be represented like a wave, but the wave can be rotated around in different ways. We can represent this as the wave having a horizontal part and a vertical part.

When I put on those polarized sunglasses, it is cutting out the horizontal part because that’s the glare coming off the ground. That’s why with polarized sunglasses, I’m getting rid of that component of light that causes glare and makes me squint. But I still can see without that part. It turns out that when I think about this, there are two dimensions. So instead of having just horizontal or vertical, I’m dealing in two dimensions, and I’m just removing part of the information. If we get down more deeply into the math, the light photons have quantum states. So, also in a quantum bit, a qubit, we can create quantum states. This is a much more sophisticated way of representing information than just bits, and we can compute with it.

It gets especially powerful when we start using more qubits and having them work together. We can deal with things that aren’t simply zero or one, and we can make it work together with classical computing.

SVG: Very interesting. So let’s talk about the ecosystem now. For them to use a quantum computer in real life, what configuration of a quantum computer do they need to get started? What sort of capital investments in terms of both human assets and financial assets would be required to get started?

Bob: A quantum computer is a system that lives on the cloud. If we’re talking about the IBM Q System One, a system we introduced a little bit more than a year ago, in total, it measures about three meters by three meters by about three meters high. It and the 19 other quantum systems we have are accessed via the IBM Cloud.

We measure the overall system performance by a metric called Quantum Volume. We are now at Quantum Volume 32, and we’ve been able to double that every 10 to 12 months. Think of it as a quantum version of Moore’s Law.

SVG: Okay, so we come back to this entire concept of Quantum Volume. But I want to go back to what you mentioned about the cloud. Now one thing is if you see how the progress on GPUs and TPUs was made, most of the use cases today are currently on the cloud because clearly, the very high failure rate in terms of POCs. The actual use cases are so small that organizations don’t find the value of investing in a GPU or a TPU on-premises. It’s just not viable clearly. So even from a quantum standpoint, will it be fair to sort of think about quantum always in the cloud, or would there be scenarios in your mind that you think on-premise will be the way to go?

Bob: If the power of your quantum computer is doubling every year, this means twice as fast next year, four times compared to today in two years, eight times faster in three years. If you purchase your quantum computer at this stage in the development, it will fall behind pretty quickly compared with what we can put up on the IBM Cloud for you. 

SVG: So, speaking of cloud, what’s happening is that we are moving from a more centralized cloud to a decentralized cloud with the ultimate aim of going to a more distributed environment of the cloud. Now, if you would sort of followed that trajectory just from the optimum performance of assets that an organization has in terms of what it can achieve from those assets, Would it be fair to say that quantum computers in an in a distributed future world will be the reality. And secondly, just because distributed is going to be a big thing a few years from now, would IBM have, let’s say, quantum computers set up in different regions and states like it, just has that a cloud setup or say a zone?

Bob: Today, we have 20 quantum computers on the cloud. Those computers, though, are all in New York State. We announced early in 2019 that we will be putting a system in Germany this year and another in Tokyo.

Ultimately, quantum computers will live in data centres around the world where it make sense, where there’s a business need, where it works with legal and government policies and things like that—the typical considerations for cloud computers.

SVG: Bob, thank you. While you have more than wonderfully spoken about quantum, but I think the relevance and the establishment of what’s happening from a distributed future and distributed computing environment have not been established very well. Because ultimately, the idea is even a decentralized environment isn’t as efficient because there are nodes which are lying underutilized. So ultimately, what people are looking at doing is to use a distributed computing environment where all of your nodes are fully utilized, and matched to the right use case, to ensure that you can solve your unique problems or whatever you have from a computer standpoint. So, I think there’s an interesting conversation to be had about quantum computing as part of a distributed future. 

Bob: We talked before about how what runs on a quantum computer sits along and is called by what runs on classical computers. Let’s say I have a Python program, and I call a method or function. Under the covers, it goes off and talks to a quantum computer and comes back. Well, where is that quantum computer? It depends on how important and how time-critical the application is.

If this thing that would run on a classical computer would typically take six years to complete, but it takes a few seconds on a quantum computer, a little time in traversing the network isn’t going to be a problem, right?

SVG: So, let’s go back to what we spoke about IBM’s having doubled the Quantum Volume every year since 2017. And that’s probably the only measure of success of performance so far. So let’s say, you have existing architectures and existing investments and on the side and parallel going out and doubling up QV every single year. What happens to these existing investments and architectures? And please answer this question for an ecosystem partner who is building some skillsets, capabilities and investing internal architecture, how does it impact me?

Bob: Hardware-wise, all you see are faster quantum computers. So, it’s like the old days in the 90s with Moore’s law you would get a new computer, and your Excel spreadsheets would run much better from an ecosystem. There may be low-level optimizations, but for applications, the software will hide the difference in hardware architecture as we improve them.

Ecosystem-wise, I think they care the most about the software that runs on that hardware. The software has been evolving rapidly, too. Out approach, there is we started the open-source effort and community called Qiskit, and this industry-leading software is now in its 17th release. It has been put on top of not only our machines, but we also on top of an ion trap computer, and that took only three days. It’s a very powerful layer that sits on top of multiple hardware types.

For the most part, these days, developers are more affected by changes, improvements, and so forth in the development environment, rather than the hardware.

SVG: That’s why it makes sense for it to be on the cloud because you can optimize hardware at your end, whereas the developer community focuses on the platform.

Bob: That’s right. Qiskit does let you develop at all levels of the stack from the hardware all the way up to the application.

SVG: Do you think Qiskit is going to be the industry de-facto standard and probably be what DOS was to classical?

Bob: Yes, but I see it much more as the Linux analog. That’s a much better comparison because it runs on multiple platforms. It’s getting broad support, and many people are contributing to it. People are teaching it in schools already.

If you go to GitHub, and you look at the number of times people have forked the code or given it stars, you would see a graph that shows overwhelming Qiskit leadership over the last two years compared to anybody else. I’m certainly hopeful and optimistic that it will become the de-facto standard.

SVG: Sure, but that also means it’s going to be a battleground between the big boys, which is IBM, Google, and probably Microsoft at some stage, right? So what happens to standards because ultimately, what people care about what organizations care about is the investments that they make are future-proofed and applicable across environments. So, from a standalone standpoint, and I know it’s a topic you love very dearly, talk to me about how do you think standards should evolve for quantum? One, is it too early? Second, who should lead the mandate? Cause there are so many bodies around. And how would you sort of fight the classical Google versus IBM debate when are you talking standards and Qiskits?

Bob: Well, so I was in standards groups in the W3C in the late 1990s. I’m a co-author, of some of the early standards, like the Document Object Model and the Mathematical Markup Language. What you don’t want standards to do is to stifle future innovation by being a handcuff instead of a good thing to enforce uniformity.

When technology is changing very rapidly, I believe it’s better to let the hardware and the software and the open-source software evolve naturally. Real user experience influences what becomes standardized versus hypothetical use.

SVG: But quantum isn’t any faster (yet) when compared to a classical computer, right?

Bob: That’s right, and that’s okay. That is completely as advertised. We have been very straight with people from the very beginning. 

SVG: In summary, talking broad strokes, quantum computing between 2022 and 2030. What’s the sort of vision that you paint for people, organizations, and the ecosystem? Where do you think the journey’s going to lead us to?

Bob: We believe that we will reach Quantum Advantage within this decade. That when we will see a significant improvement over what we can do classically. Some problems that cannot be done at all classically will fall to quantum computers.

We will remain in the so-called NISQ (Noisy Intermediate Stage quantum) for most of the decade, and we will start to see error correction introduced in some way, during this decade.

SVG: Okay, and let’s say if I was to ask you five key problems that quantum will solve. What are they going to be in your view?

Bob: The first general area will be in chemistry because of the mapping of quantum computing to quantum mechanics, which seems to be the model by which nature, atoms, molecules, and so forth works. Another area is Financial Services. There’s a lot of work going on now, and it is accelerating. Improving the computational speed and finding new algorithms for AI is another rich research area. Though we may start in certain industries, the techniques and advantages should spread to others quickly.

SVG: Bob, thank you. This has been a wonderful conversation, and I hope to catch-up soon for another, deeper round of conversation on quantum. Thanks again for your time.


<strong>Analyst: Sanchit Vir Gogia</strong>
Analyst: Sanchit Vir Gogia

Sanchit is the Chief Analyst, Founder & CEO of Greyhound Research, a Global, Award-Winning, Digital & Technology Research & Advisory firm.


Disclaimer. This #ONTrigger executive dialogue is brought to you by Greyhound Research. Please note, Greyhound Research holds complete editorial control of this content, and the featured executive (and their employer) has had no influence on the content quality and production process. All content shared herein is the copyright of Greyhound Research and you may share this using the options made available. Please don’t download this content (complete or parts) and distribute it over the web and emails. Email us at connect@greyhoundg.com if you need clarifications.

Copyright © 2020 Greyhound Research. All rights reserved. You may share this research note using the options made available. Please don’t copy this research note (complete or parts) and distribute over the web and emails. Connect with us if you need clarifications.

Legal Note. The contents in this portal do not necessarily reflect the opinions, ideas, thoughts, points of view, and any other potential attribution to any contributor’s, commentator’s, or author’s past or future employers. No NDA’s have been broken. Email our Client Centricity Team to learn more on our NDA policy. Sources may come from both publicly available information and private individual conversations. In case of publicly available information, attributions have been added.

Leave a Reply