Podcast with Olivier Ezratty, Author and Consultant

My guest today is Olivier Ezratty, a prolific author and consultant. Olivier and I speak about the risks and rewards of hype in quantum, the difference between the German and French quantum ecosystems, full-stack vendors and more.

Listen to additional podcasts here

‍

THE FULL TRANSCRIPT IS BELOW

‍

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

Olivier: Hello, Yuval.

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

Olivier: I'm Olivier Ezratty. So I'm freelance technology consultant and analyst, and I'm specialized in quantum tech. And I've been doing this for about four years now.

Yuval: So you have a book that I think now is in the fourth edition, am I correct?

Olivier: Yes. It's an 836 pages book. It's free to download as a PDF, but you can also buy a paperback edition on Amazon in two volumes, and it covers about anything on quantum tech. I mean, from science to technology, even geopolitics and ethics, social issues, whatever.

Yuval: Wow. So right now it has more pages than we have qubits. So, but maybe qubits are catching up. Right?

Olivier: Well, it depends if you take into account D-Wave, but yes

Yuval: And then you also have a podcast and you recently published a paper aside, almost an academic paper on quantum hype. Is that correct?

Olivier: Yeah. Couple weeks ago.

Yuval: So tell me about it. What was the thesis in the paper and what did you find?

Olivier: Well, first, why did I write this? I started to think about it when I met with a couple corporate customers who were starting to worry about this hype and they were worried that it could be very similar to the artificial intelligence hype from the late eighties, early nineties. And they told me, which was really worrying, that if we are not able to deliver a substantial improvement with algorithms and actual hardware, some customers would pull the plug on the quantum investments.

So I was willing to understand what was behind this hype, what were its different aspects and how was it different from the various technology hubs that we've seen for the last three decades? And I found out some similarities, but also a lot of difference.

Maybe I can elaborate, as you like?

Yuval: Please.

Olivier: So what I found out is there's a bipolar hype. You've got a significant over hype in quantum computing, but on the other hand, you've got an under hype phenomenon with other fields like quantum sensing. It makes a big difference. And I think that probably we should talk more about quantum sensing, which is more ready, higher TRL (technology readiness level), a lot of use cases in various industries, so it could balance the field with quantum computing.

Well, the quantum computing hype started a while ago, probably with Peter Shor's algorithm back... It's a long time ago. 20, 28 years ago. Then you had the quantum supremacy three years ago, then the big chunks of large startup funding. I mean the PsiQuantum, IonQ, Rigetti, and now we've got the D-Wave spec.

And then on top of that, you had all the quantum investment plans from most governments, I would say. US, China, Russia, Israel, France, Germany, UK, Singapore, all these countries have a significant quantum plan, and all this mixed together creates the hype. And what's very interesting is what comes out of the analyst and the consultant scene, with those crazy market forecast, sometimes confusing the created value and the real addressable market. You've got Boston Consulting Group and McKinsey who talks about $1 trillion market, which is just value generation with customers, but it's not the market for hardware, software and services. So this is creating a big confusion.

And on top of that, so where are the differences? The differences with other hypes, like the ones we have right now on the crypto countries or NFTs and Web Three, and so on is the science is very different. We've got a very low technology readiness level, even for startups.

I would say it's more or less similar to nuclear fusion, very low trials. I have never seen such high scientific uncertainty and also confusion because it's very hard to evaluate the technology. If I was a customer, I would have a very hard time to figure out which kind of computer could I use to do what with what performance? What benchmark could I rely on to compare solutions? How do I compare a quantum annealing, quantum simulation, and gate-based quantum computing? It's really hard to evaluate the current scene. But on other hand, what I would say is where the science is strong, even though it's not perfect, we don't have such high investments, even though PsiQuantum raised more than 700 million bucks, it's not that much. I mean, if you look at the total investment in the quantum startup scene, it's less than five billion.

Five billion is pocket money. I mean, I would compare it to the 625 billion money that was poured into startups last year. It's nothing. It's nothing compared to the business benefit that could be created by the whole quantum field.

So all in all, it's not that bad, I would say, but it has a lot of consequences. The quantum scene is really changing everything. It's changing the way public research is being conveyed. You have so much fight for public funding and private funding that it creates a lot of competition between both fields. In some cases, you have bigger research teams, fundamental research teams in the vendor space than the public sector space. Even in US, when you look at the number of people who work at PsiQuantum, even IBM, it's hundreds of people. It's more than any lab, even in the MIT, Harvard, Princeton, the Maryland university, it's more. So we have a fairly new situation, that's part of some kind of bubble. So it's creating a new situation. So then what do we do? That's the question.

Yuval: So let's decompose this a little bit. First, hype is not entirely bad, right? Because I think that you're making the case that hype encourages research, hype encourages students to go into the field, hype encourages investments. So some level of hype, we can argue whether it's too much or too little is a good thing. Is that correct?

Olivier: Yeah. Yeah, exactly. I think there's always been some hype in most of the technology waves. And even those who did succeed, the danger is after the hype. If you have some over hype at some point, and you've got this through of disillusion, and if the disillusion is too big, too large, then it can stop the funding for public research. It can stop the funding even in the private sector, so that's the danger. So if you have a too big a gap between the expectations from the market and from everybody, every stakeholder and what can be actually delivered. And it's not a problem of passive option. It's not a problem of marketing. It's a problem of science reality. If you've got too large a gap, then you are in trouble. That's what happened twice with AI. It happened in the seventies, then in the eighties and nineties.

So I think we should avoid that. So we should probably lower a little bit the tone of the hype and make sure that we deliver, make sure that some safe bet are done on one hand, make sure that people, all stakeholders understand that we are in a very long term scientific journey. We won't have scalable quantum computing in five years, whatever people say, it will take a long time. We have to make a lot of efforts to deliver actual quantum advantages with the hardware and related software and tools. So we should soften the tone, probably. We should probably also have more engagement and involvement of scientists in the debate. It should be more visible. We should maybe have some debate between the skeptics and the optimist in that field. That's the thing. And probably, everything is about how do you implement a so-called responsible innovation process? So how to make sure that all the stakeholders are responsible in their behavior, starting with the vendors. So some changes is required, I would say.

Yuval: I think Bill Gates said that revolutions take longer than you think, and then they are bigger than you think. So when you put it in historical perspective and I mean, we see the potential of quantum, right? And if you have systems with a hundred qubits, I think it's believable that you will have 200 and 400 and 800 and so on and so on. It just might take a little bit more time. In your opinion, how long before quantum becomes something that's not just research and hype and becomes truly useful?

Olivier: Well, that's the hundred dollar question we always hear. The honest response is to say I have no idea. Because it's so complicated, it's about how do you create an entangled state with a large number of quantum objects? Nobody knows. Nobody knows if it's possible. The theory says yes, probably more the mathematical theory than the real quantum theory. So it's not that easy. It's not as easy to move from 10 to 100 and 100 to several hundreds and thousands, because it's about controlling an n-body system, and this is far from being easy, whatever the tools you use.

What I believe is we probably will have some surprises from original architectures. For example, I don't believe a lot in the scalability of superconducting qubits. I think we may have some surprises from photonic qubits, maybe from silicon qubits, maybe from very strange architecture based on topological qubit, even though the Microsoft path seems in bad shape right now. I believe we may have surprises from original solutions. Typically, the measurement-based quantum computing solution with photonic system may be interesting, but so far we don't have the proofs. So we have to be very open. That's why we can't make a bet of what's going to be the winning hardware architect.

Yuval: Earlier in our conversation, you mentioned that your book also covers geopolitics and we've seen large government investments, maybe not huge, but still, 2 billion here, 5 billion there, 3 billion there, soon enough you get to real money, as they say.

Olivier: Mm-hmm (affirmative)

Yuval: Do you think all these governments are wrong in investing, or do you think they look at what will it cost to invest and what might the cost be if we don't invest and we don't have quantum technology? What do you think is behind all these investments?

Olivier: Yeah, what's behind is the quest for sovereignty, technology sovereignty. The goals are different from one country to the other. I'd say that for the large countries, like China, the US, and even to some extent, Russia, it's about controlling the fate of their information systems. It's about the potential to control cyber security. That makes it an attractive position.

I'd say that the European position is a bit different. I don't expect the European countries to be willing to control the cyber space as maybe the Chinese and the Americans would like to do, but Europe is willing to bet that a new technology revolution is an opportunity not to lose again against the Americans. I mean, so much technology is being controlled by large American companies, from Intel to IBM, to Microsoft, Google, Facebook, and others, and Chinese from manufacturing and Taiwan for semiconductor production. So Europe wants to have its fair share of a new cake. And since there is a huge uncertainty in the science field to create the cake, they say, "Why not us as well?" So the rationale is a bit different. And the other countries I would say Israel, Netherland, Singapore, they also want to be a player. They know they will never be as big as the US and they want to be part of this new ecosystem that's being built.

Yuval: Now in Europe, there are investments on multiple levels, right? There are investments at the EU level.

Olivier: Yes.

Yuval: How is the French ecosystem different in your opinion than the German ecosystem? Because on one hand in Germany, there are a lot of industrial companies that are doing quantum, BMW and others. And certainly also Fraunhofer Institute and many academic organizations. In France, there seems to be a hub of quantum activity in Paris. How would you compare the French and German approach to quantum?

Olivier: Well in France you have many hubs, a bit like Germany. I mean Germany, you've got Munich, you've got Dresden or Hanover or Berlin, you've got many places. Even around Stuttgart with the IBM site, and Hanover. So in France, we have at least three main hubs. It's Paris, Siècle, near Paris and then Grenoble with a huge track around the silicon qubit. And then you've got Bordeaux, Strasbourg, Toulouse, Nice, many, many different places.

Well the difference with Germany, I would say, is twofold. One is, surprisingly you have more hardware startup in France than in Germany. You would expect, given the manufacturing tradition of Germany, to have more hardware startups there. And actually in France, we have already five startups in a quantum computing hardware like Pasqal, Alice&Bob, C12. You've got Quandela. You're going to have soon a new one on silicon qubit.

So, it's five. In Germany, I don't know, we've got maybe only two and one just one bankrupt. And they have more startups in the software scene, and upside down, the large corporations' ecosystem is more dynamic in Germany on the demand side. So we've got all the automotive industry, all the pharmaceutical industry, some part of the financial industry that looks to be more dynamic in Germany than in France or even in the UK. So I would rate that. And when you look at research, from the amount of research coming from public research labs, we have a lot more funding in Germany than in France, but there is some teamwork happening. So some teams in France team up with German teams in various fields. There's for example, one partnership that has been launched on creating a hybrid computing platform, teaming up with a Pasqal, the quantum simulation startup.

So Jülich, on one end, the Jülich Supercomputing Center on one end in Germany, and it's equivalent at the Siècle in the Paris region, teaming up on that with the European Commission. So we have to team up, whatever happens. That's the lesson, because Europe, as a fragmented region, can't compete efficiently with the US or China. So that's why we need more partnerships within Europe. We recently have a very, very interesting situation with a software company from the Netherland called Qu & Co, that was acquired by Pasqal in France, or it's an M&A, and now we have a larger company. It's not as big as the merger between HQS and CQC in the UK, but still, it's interesting to have some European companies that are merging together.

Yuval: So actually let's talk about that, the hardware and software companies that are merging together. You mentioned that there are different many different modalities of hardware, and we don't know exactly who will be the winner. So is it wise for customers to go to a monolithic vendor that says we have everything hardware and software and applications, or would your recommendation typically be to select the best of breed in every part of the technology stack?

Olivier: Well, right now there is a lot of vertical integration. I mean, most of the large vendors want to have the full stack solution. IBM is one example, even D-Wave is entirely full stack. So the trend is being full stack, probably being full stack is a tradition with emerging markets and when the market get consolidated, somebody like Microsoft or Intel or Android with Google in the smartphone space is horizontalizing the market. Right now, we are more in verticalization approach.

And so if we take the case of Qu & Co and Pasqal, it maybe makes a lot more sense than for IBM and others. The reason is the software ecosystem for quantum simulation is not that major. I mean, you have much fewer software vendors working on quantum simulation than on gate-based quantum computing. So it makes sense when you are in a more agnostic or less used hardware platform, it makes sense to be vertically integrated. But hopefully this kind of company, those guys doing quantum simulation, they would love to have more software vendors working on their platform, but when it's small, they have to start with their own platform, so that's why the acquisition made sense.

Yuval: So if you were a master of the quantum universe and you, all of a sudden, are not just monitoring, but now you control everything.

Olivier: Yep.

Yuval: What would you want us technology vendors to be working on between now and say the end of 2023?

Olivier: Oh. Well first I would like to have a larger projects, like a Manhattan project with larger teams, more coordination. I would love also to have more coordination around the enabling technologies. If you take semiconductor qubits, whether it's a silicon qubit or super conducting qubit, you need more joint work on enabling technologies like cryoelectronics. You need also to have some work being done on the energetics. I am, myself, involved in such a project with my friend, Alexia Auffèves from Grenoble. We need to have more efforts that are transversal to all the track.

I would also like to have a more joint work between all the vendors on benchmarking. So far, I see the competition is so hard and also the obfuscation of the real capacities of existing qubits, that there's no accepted benchmark that's used everywhere. It's very hard to figure out what are the real performance of each and every hardware solution. So I would love to have more collaboration. And that's contradiction, because there's a contradiction between the usual collaboration and open source attitude that you have in public research and the fact that now there's a lot of research being conducted in the private sector. So we have to manage at some point, this contradiction.

Yuval: Olivier, how can people get in touch with you to learn more about your research?

Olivier: Well, in all my books and all my papers, my email, and even my phone is indicated. So I'm very easy to join. You can Google me, you find everything. I'm easy. I'm easy to find.

Yuval: Very good. Well, thank you so much for joining me today.

Olivier: Thank you. Yuval.