Podcast with David Shaw, Director at Fact Based Insight

My guest today is David Shaw, Director at Fact Based Insight, a quantum analyst company. David and I talk about how to best counter China’s quantum strategy, the role of consultants in creating and implementing a quantum computing roadmap, hardware elasticity, and much more.

Listen to additional podcasts here

THE FULL TRANSCRIPT IS BELOW

Yuval: Hello, David, and thanks for joining me today.

David: Hi, Yuval. Nice to be here.

Yuval: So who are you and what do you?

David: This is David Shaw from Fact Based Insight. Fact Based Insight provides quantum market advisory content for interested business and investors in the sector.

Yuval: I saw a couple weeks ago that you published a really comprehensive report about sort of what I would call the state of quantum. And we're recording this in early February 2022. So if you look at 2021 versus 2022, what do you think is going to happen on the market dynamics? Where are enterprises in 2021, and where do you think they'll be in 2022?

David: I think it's been a great transition for the sector. The awareness of quantum technology, and quantum computing, in particular, with large interested potential adopters has really moved forward. And I think it's probably the norm rather than the exception now that major corporations are aware that there's something significant that they need to work out what they're doing with. I think that's a big development.

Yuval: Sometimes we think about the world in three categories: people who are quantum curious, maybe they heard about quantum, maybe they read about it in the BBC reports or Wall Street Journal. Then there are people who have maybe started setting up small quantum teams to explore. And then, of course, there are people who are deeply engaged in quantum and have larger teams and are closer to moving into production. Where do you think the bulk of the market is? On the first, quantum curious, or so dipping their toes in the water, or fully committed?

David: I must say, while we clearly, when we all go to conferences, or join conferences online these days, we hear the players that have started to dip their feet in the water, but I suspect that the bulk of organizations out there, they're still at the quantum curious stage. I think there's a lot of people out there that have maybe looked at their own careers and thought, "Well, actually, is this something that I can learn more about, because it's going to be part of building out my own personal growth journey?" And clearly they're then wanting to go along with that in their own organizations. I do think it's noticeable.

When I look at which of the engagement activities that have been a success so far for players in the market, there's often a very strong educational component of that. In terms of what IBM's great success with IBM Q, a big element that we didn't really strictly speaking need to have access to real quantum devices on the cloud, really small quantum devices on the cloud, to all start learning how to do some quantum programming. But it was really a key point of traction that got a lot of people interested in giving this a go. And I think it's been a big success for IBM, and owning that educational space, I think, has set them up with a very nice momentum around their view of what a quantum software framework looks like.

I think it's very interesting to look at D-Wave's, rather a contrasting approach, which played to its own unique position and unique strengths. Being a small organization, it deliberately looked to a strategy which pulled in early end users to drive user-led trials. And I think that's been really instrumental in alerting a wide population of potential business users to the whole notion that you can start to think about business problems in a different way, and that when you get bright people doing that, it generates very interesting results. And they're not always what you expect, but it drives business benefit in a more general sense.

I think it's interesting. You see some of the current initiatives that have come out of the sector, and there's some very interesting variations on this approach. You've got people that are perhaps have more of a quantum internet view of the world deliberately launching tools which can help people learn about that way of thinking of things. And QuTech are very actively... I think it's a very interesting strategy to try to get a quite different view of the future of the market across. Very interesting also to see what you would've considered a very specialist quantum software shop, QQ-Ctrl, a highly specialized activity, but it spots that it's got an opportunity to roll out an educational tool, Black Opal, which I've really been very impressed with and think it could have really great wide application.I think it's a great starting point for people that haven't come across quantum concepts, but really want to get their head around those basics before they tackle something more substantive like a Qiskit tutorial.

Yuval: So let's assume I was a CIO of large financial institution. I'm curious about quantum. I want to get started. I had my people go through a couple of courses or online education. Maybe I've even identified a couple of candidate use cases where I would like to compare quantum performance with classical performance. Then I think I get to a fork in the road, which is the consulting versus internal development. What do you see happening, and what would you recommend to customers? Is it build your own expertise internally, or is it turn it over to a consultant?

David: Well, I probably should admit my own background in this. My own career started in management consulting. I was for many years a consultant with what was then Anderson Consulting, now it's Accenture, and then many years with AT Kearney, a strategy-oriented management consulting house. And the thing that I'd reflect on, I have a lot of sympathy with the way consultants can add value to an organization overall. I don't think it's an us or them decision. I think organizations should clearly want to identify and understand the in-house capability that they need and want to build, and they should look for how professional service companies, and I include here the active management consultants, whether it's Boston, whether it's McKinsey, et cetera, whether it's, yeah. But also the consulting arms of organizations such as IBM, and look at how they can potentially work with your organization to build that capability you need.

I think it's also very interesting to look across the widest quantum software space. And I always return to an observation at the moment that rather than walk out into that market and look at a particular feature on a particular platform, a particular algorithm for a particular application that someone happens to have developed now, or a particular feature in a particular tool that lets you link... Clearly, that kind of capability's important in the long term, but at the moment, at this early stage, I think it's more important to look at picking the right partner organizations that you're going to want to work with.

And I think there's a tradeoff there between is that a general management consulting player, or one of the more specialist quantum algorithm expertise oriented players, or a special platform capability oriented player, and really ask which of those potential partners to start and then continue on this journey is the right one for what your organization's seeking to do? And the actual skills to do new work in quantum algorithms, new quantum algorithmic work's a very scarce skill. And so you have to be realistic in the access to partners' resources with that skill that you'll be able to get. You have to welcome the ability of large scale consulting organizations to deploy knowledgeable resources in country on specific projects.

And you also have to be aware that working with a partner that provides a tool to assist and choreograph that progress, that doesn't just bring an immediate benefit, but it also potentially brings a way of turning what can be the expertise of point individuals into a more organizational knowledge by using the platform to build that out, whether that's taking a piece of applications and algorithm insight and turning it into something that can then actually be programmed at scale, by having a library of functions, or whether it's something that gives you a head start in taking an innovation and plugging it into what will ultimately be a business process workflow.

And those types of questions, ... These are not new quantum questions. These are questions that business leaders grapple with across a number of contexts. And they need to realize that they have to solve all these normal questions, but yes, in the context of a new long term opportunity, such as quantum.

Yuval: Following up on the management consulting, you could look at that in two parts. One is strategy, and you mentioned BCG and McKinsey. What use case should you attack? Where have we seen this work? Are the expectations realistic? What hardware would you recommend? And so on and so on. And then there's the implementation part. And if organizations believe that quantum is strategic enough, then I don't think that they would want to outsource 100% of the implementation to someone else because there would be a day after tomorrow where they start to say, now we do have a unique quantum algorithm, or we do have a unique data source, or what have you. How do you see that division between the strategy and the implementation?

David: Yeah, I think it's an excellent question. And a basic thing I would have in my mind in responding is that this question of how do you, on the one hand, identify good strategy, and on the other hand, how do you ensure that strategy will be well implemented across an organization, is a classic, and one of the most profound challenges in normal business. And different organizations will have their own style and approach to that, that they're comfortable with. Myself personally, I've always put a premium on the ability to work with partner organizations, whether that's consultants, whether that's others, that can help you along that whole trajectory.

If you, as an end use organization, really want to compartmentalize that, and insist that you're going to work with one highbrow strategy shop here, and then we're going to do this planning internally, and then we're going to build this ourselves... then that's great, but you've got to realize that your organization's got to be oriented around being able to deliver the continuity of that overall arc. And some organizations, I think, they will feel that they can do that. For others, I see a role for consultants, general consultants, that can help join that up across the arc.

Yuval: You mentioned the difficulty of recruiting people to develop algorithms and so on. Do you think that's a function of education? Just need to learn more about quantum. Do you think that's a reflection on the state of the market with regards to the development tools and platforms? Maybe they're just too difficult and not abstract enough. How do you see this gap between supply and demand for people who can implement quantum closing?

David: I think, to start responding to that, I think I would wind back slightly. And there's a key point when we talk about quantum algorithms that it covers a wide range of specifics. At the one end, there are end use applications. And I think we're clear we want to talk about something different, this is specific to a particular business need. And then at the other extreme, you can say, well, what underlying quantum algorithms are there? Well, really, there's only a few underlying primitives that are really where this difference is coming from. And even there's an argument that really they're all a version of one similar thing. And so the research for genuinely new quantum primitives is a very exciting and interesting area, but a highly specialist area.

But then there's a complete middle ground, which is about working out how to assemble underlying quantum primitives, algorithmic primitives, into useful mid-level algorithms. And that's actually the bulk of the activity that I think I see going on, that you see archive papers coming out with great frequency on. And it is itself still a very specialist activity. And so you returned to how to recruit against that backdrop. And I think as an organization, I think you first got to be clear which part of that market you're trying to address. And for some large organizations, yes, they will have deep R&D activities that want to be involved with and fund research right at the algorithmic primitive cold phase.

There'll be other organizations that are very comfortable working at that mid-range, and that's still a challenging recruitment task. And there'll be others that want to work at the applications-oriented end. And of course, that's a massive activity. We know from the history of the conventional digital revolution that that becomes an enormous opportunity and an enormous business activity. So there is a massive re-skilling activity to take place there, but of course, the potential recruits, the way they need to be developed, the new skills, how old skills can be leveraged, is different for these different compartmentalizations.

Yuval: I wanted to ask you about the execution of the program. So first, is it safe to say that the majority of quantum programs that will be executed in the near term will be on the cloud as opposed to truly on-premise?

David: I think if you take a straightforward, quantitative answer to that, then I would suspect it would be on the cloud, because it's simply easier. I think though, that there will be organizations where they perceive that their use of these devices is sensitive, whether that's because they have sensitivities around security and defense applications, around their proprietary materials, research, they'll want to look at a secure on-premise installation. I know it's in the back of my mind, though, one of the bits that I see as very unjoined up at the moment in the field, is on the one hand, the quantum computing-led discussion, which looks at quantum computers delivered through a conventional cloud and potentially on-premise installations.

And then equally, you've got the discussion coming from the quantum communications community that says, "Oh yeah, well, at the moment we're doing quantum random number generators, and QKD, and everyone should be interested in that." But actually the next big thing that community wants to talk about is blind quantum computing on the cloud. And it doesn't get talked about so much and it doesn't get brought together so much, but actually for me, it's a brilliant combined use case.

It's not one of those use cases that has tremendously demanding implementation requirements, but it is a use case that really does address that requirement for a high degree of security and confidentiality. We're not quite there in terms of implementability, but we're certainly there in terms of demonstration level experiments that have been done. And I see a big future for that specific use case.

Yuval: And if we look at specific quantum computers, say on the cloud, today in... Not to pick on particular vendors, but just as an example, today, if I wanted to use an IonQ computer, I could use it directly with IonQ and their API, or I could use it through AWS, for instance. And AWS might say, well, look, if you work for us, you can use different computers at different times of the day and compare them and so on. Whereas if you work with IonQ, maybe you have deeper access and better way to interface with the computer, but it's on the IonQ. How do you see that evolving? Do you think people will be selecting a quantum computer of choice, or will they prefer to work through the cloud providers?

David: I think end users are certainly going to want, in principle, they would desire a high degree of hardware angoricity. They don't want to be tied into one approach. And particularly, when you're learning, I think in the introductory activities, they're very comfortable that they're going to use a standard circuit model and they're going to just want to be able to experience that in the easiest possible way. However, when we ask, where are we going to get real end use applications over the line to commercial advantage in the NISQ era, you have to think it's going to be more difficult. It may not even be possible, but assuming we do manage to do it, then I kind of think we're going to need to be leveraging all of the sources available at the most granular level that we can access them.

And at the moment, in terms of the state of how frameworks have managed to standardize, well, really, that level of standardization isn't there yet, unless I go specifically to one of the manufacturers and go deep into their stack and work with them about what exactly they're going to open up to me. Now, initiatives like Open QASM 3.0, it's trying to extend that, it's trying to push that standardization to a lower level, and that may change the dynamic. And I think you can see why IBM are keen to be part of that. You can see why other platform-oriented service providers, such as AWS Braket are going to want to support them on that journey as well, and you can see that in terms of the collaboration in that standard. But in terms of where we're at, I kind of think to get a gate model machine over the line, you're going to have to work closely at a low level at the moment.

Yuval: As we get close to the end of our conversations today, I wanted to ask you a hypothetical question. You spend most of your time monitoring the industry. You see what vendors are doing and you try to make sense of it and then explain it to your customers. But if you were able to control what the industry is doing, so I give you complete control over the quantum industry for the next 18 months or so. What would you have us do?

David: I almost want to take that question and turn it around slightly to a different angle, because one of the biggest questions we face, I think, as a community, particularly in the West, is how will the growing geopolitical rivalry in the sector play out? And in reality, that scenario you just painted to me, if I could just dictate how this should go according to a more central plan, that's kind of how the Chinese program is likely to develop. It has that advantage, it's a part of the system. China's deliberately leveraged long term technological foresight to develop a commanding place, for example, in solar panels or in batteries, and clearly seeking to do the same in a series of quantum technologies. And so the strength of the Western response to that, well, I don't think it really should be to try to be like the Chinese system.

I don't think we're going to beat the Chinese at their own game, even if we wanted to. But the question is how can we develop the right patterns of collaboration across the sectors, whether that's in the rest of Asia, in Europe, in the EU versus wider Europe, in North America? How can we get the right patterns of collaboration that play to our strengths, but also capture the traditional benefits of our market economies, which is that we don't try to dictate everything from the center and that we allow a variety of approaches to be explored on their own merits? And ultimately the market and that wider ecosystem selects what deserves to win. And so really, for me, that's the challenge for the next 18 months. How can we recapture the strengths of our economic system to drive the industry forward?

Yuval: So for the record, you turned down the opportunity to be the quantum emperor, and it's fine. David, this has been fascinating. How can people get in touch with you to learn more about your work?

David: Yeah. Please come to factbasedinsight.com, or look me up on LinkedIn.

Yuval: Very good. Thank you so much for joining me today.

David: Thank you.