- 5 months ago
Advances in hardware, algorithms, and investment are bringing quantum computing closer to industrial real-world applications. From optimizing financial portfolios and supply chains to accelerating drug discovery and materials science, quantum technologies promise to tackle problems that remain out of reach for classical computers. But how close are we to practical impact? This session explores how quantum computing is moving from the lab to the enterprise, and which industries may see the first transformative breakthroughs.
Category
🤖
TechTranscript
00:12We'll figure out sort of where to sit. I'm fairly sure.
00:18And first of all, thanks a lot also for your kind attendance during our session today because I
00:23think we're also discussing one of the most interesting topics that we could currently
00:28have in terms of when we would actually do a quantum leap and when quantum advantage would
00:32also come. And I'm very pleased to be here today together with Axel, Jerry, as well as
00:40Loic also on stage and discussing this very interesting topic. We'll first give also an
00:46introduction to ourselves. I'll quickly introduce myself, Henning Soller, partner from McKinsey,
00:51also from the office in Frankfurt. Been with McKinsey for the last 13 years and since the
00:56last five years also our global lead for our activities in the quantum space. Maybe Loic,
01:01you want to go ahead with a bit of an extended introduction, also saying a bit about what
01:05you're doing, what your company is doing in this interesting space, and then we take it
01:09from there.
01:11Sure. Thank you very much. I'm very happy to be here with all of you for this discussion.
01:16I am Loic Henriet, CTO at Pascal. Pascal is the French company building quantum computers
01:23and operating quantum computers, implementing algorithms on those quantum computers for our
01:29partners and clients. Our technology is based on neutral atoms, so single atoms that we manipulate
01:36with the laser light, and we use those atoms as qubits, building blocks, to actually implement
01:43quantum algorithm. The technology has been developed in France at Institut Optique, Plateau de
01:50Saclay, in a research lab with Alain Spey, Nobel Prize in Physics in 2022, as well as Antoine
01:57Brauess and Thierry Laet. And for the last seven years, we've developed this technology, industrialized
02:02it, and installed quantum computing machines with HPC centers all around the world. So that's
02:10what we are doing at Pascal.
02:11Yeah. Jerry, you want to go next?
02:14Great. Thanks a lot for having me as part of this panel. So my name is Jerry Chow. I'm
02:21an IBM fellow and CTO of quantum-centric supercomputing at IBM. And we've had a long history of quantum
02:37quantum information theory really came out of IBM research. But more recently, we've been
02:42really focused on building and bringing this technology to life and really making it a part
02:48of computation. It was 10 years ago that we actually put our first five-qubit, superconducting
02:55qubit quantum processor on the cloud. And then since then, we've deployed cumulatively
03:01over 90 quantum systems. We've built an entire ecosystem around an open source software stack
03:07called Qiskit. And we've been really pushing the envelope in terms of a very transparent roadmap
03:13of what we can deliver in terms of capabilities and hardware as well as software that builds
03:19on top of it. But from a fundamental point of view, we see the drive of quantum as delivering
03:26the most performant capabilities and platform, meaning hardware and software. And then on top
03:31of that, building the ecosystem to get those involved, using it as part of computational workflows
03:36and really unlocking the value towards enterprise usage.
03:41Axel?
03:42Yeah. Thank you. Likewise. Happy to be here. My name is Axel Müller-Gröling. I'm a board member
03:48of Fraunhofer Gesellschaft. Fraunhofer is the largest European applied research organization.
03:55And the mission is to benefit society and economy. And what did we do? What's our role in quantum
04:03computing? We formed six years ago a competence network with 23 of our 74 institutes on quantum computing
04:15with their specific research areas and their domain expertise, actually. And according to our mission,
04:22our aim is to see how do we get real world value out of quantum computing, right? And for that,
04:31as was just mentioned, we need the ecosystem. We need not just hardware or just software. We need also
04:38use cases. We need application. We need
04:46a lot more than just a single dimension. And to form that bridge from the potential of quantum computing
04:57to real world value, that is the idea that we try to help with.
05:05Yeah. And in that sense, we are also trying to be a technical advisor, if you wish, because
05:12just by definition and by our role and mission, we have on the one hand, deep expertise and technical
05:18issues. On the other hand, we talk to industry all day long. Yeah. And we know and think we know
05:26the
05:26needs of industry. We are talking to Airbus, to BMW, to Thales, to Infineon, to EON, to PWC, and the
05:35like of those.
05:36And that's the way how we try to push the whole topic forward.
05:44Very nice. Thanks also for the for the intros that you have provided. I said at the beginning,
05:49we're living in very interesting times, specifically with respect to the age of quantum and the age of
05:55quantum advantage. We've seen also in the latest report that we as McKinsey have issued around the
06:01state of the quantum industry itself that during the last year, we have seen an increase of 6.3x
06:08in terms of the investments. So we now see yearly investments of 12.6 billion into the quantum industry.
06:14We also see that this is not a one-off event. We see this as the new normal. And at
06:20the same time,
06:21we see this is not just pure academic research any longer. This is to a large extent also driven
06:26by private investment. And what we also see is that within the quantum computing industry,
06:32we've by now crossed also one billion revenues. And one billion revenues not just coming from,
06:38again, sort of state-owned programs or state-affiliated programs, but really also to a large
06:43extent from corporate customers who are investigating the topic, not for immediate quantum advantage,
06:49but in order to basically see what can the technology do, what would be possible IP to capture in that
06:54space. So we see that more than 300 of those companies also exist by now on a worldwide basis,
07:00and the number is further increasing time over time. And we also see very successful additional
07:07entries into the market with some of the recent IPOs and also the investment rounds that we have seen on
07:12the
07:12quantum side that essentially underpin that we're not just in a phase where
07:17the quantum computer will come in five years and in five years we'll say exactly the same,
07:21but quantum computing is a reality that will come and that companies now also need to prepare for,
07:26and not just from a security perspective, but also from a perspective of what can we actually do
07:31positively with the technology then also going forward. One of the key questions that essentially
07:37is out there is always, okay, when will we have quantum advantage? But quantum advantage is also
07:43a bit of a hype word because many people have claimed quantum advantage in the past, many people say sort
07:48of quantum advantage is something that is very hard to reach, also because the definition of the actual
07:52quantum advantage is not very clear. And maybe, Jerry, you just want to give it a first shot because I
07:57think IBM had been at the heart of some of the initial discussions around quantum advantage, so maybe you just
08:02want to contemplate a bit on
08:03what does quantum advantage actually mean for you and when do you think we will have quantum advantage?
08:08Yeah, sure. Thanks a lot for that prompt, right? I think
08:13there's really, you know, there's a couple of feelings about quantum advantage. I mean, on one front,
08:18it's a very academic pursuit, right? That there's a kind of provable, verifiable way to show that a quantum
08:27computer can have an advantage over any kind of, you know, any kind of classical solution, be it, you know,
08:33cheaper, faster, or more accurate. But a lot of that kind of pursuit really is one where it's about
08:38showing the rigor and showing the verifiability through all kinds of scientific exploration.
08:45But I almost want to say that that's one axis of advantage. And another one, there's really just
08:51bringing quantum to reality and bringing quantum towards real problems. And that we're seeing day
08:58by day, all the time these days, where we're seeing the application of using quantum for science that
09:05matters for real problems, simulating molecular structures of sizes that are beyond what you can
09:10exactly simulate, looking at things such as scientific problems, such as neutron scattering,
09:16which are actually done in real laboratories being simulated on quantum computers. We're really
09:22starting to see the way that science is being done on these, these machines in ways that hasn't
09:27been done before. And from that perspective, there's almost this advantage, which is driving the reality
09:32of using quantum where we're now beyond where we're talking about, you know, science of the devices,
09:37but we're talking about how we're doing science on the devices for advantage. Yeah, if I can, if I can
09:43maybe
09:43jump in on that particular take on quantum advantage, I would say that I fully agree that we need to
09:49focus on practical problems. Like in the past few years, people were claiming quantum advantage on
09:57maybe problems that are too academic, that people don't really care about, that would not bring value in
10:03the end. And I would then align that with a recent example. I think IBM and Pascal share the same
10:10perspective on that. It was covered in an article, a news article by Nature a few months ago, and it
10:19was about
10:19using quantum processing units, IBM one in one case, Pascal one in another case, to describe the behavior of
10:29real material. So this was a key thing that happened, that quantum simulations on a quantum computer was for the
10:37first time able to reproduce the real behavior of a real magnet, a real material that was of
10:43interest for scientists, for R&D researchers and all. So we have already entered this era where you actually
10:51gain some unique insights using a quantum processing unit compared to what you could do with the GPUs or
10:59CPUs with the best classical algorithms running on it. So in that perspective, I think we are already
11:04entering this practical advantage area where you can use quantum machines as a unique tool to study
11:12science and industry. And I think, Axel, you had already mentioned that you have numerous discussions
11:19also with some of the industry leaders around and also the application of quantum.
11:24I presume that you will have the same discussions also around the AI topic itself and therefore sort
11:30of what are the use cases. So maybe you want to elaborate a bit on what are the discussions that
11:34you have with them on quantum and how do they also relate to some of the new questions that are
11:38now coming
11:39up in the age of AI? Well, you see, in the logic of what I just said about Varnhofer,
11:46what we take is a very pragmatic definition, right? Is there a value, a relevant problem that creates value in
12:00the
12:00economic space that can be solved by quantum computing nowadays better than any classical computer could?
12:08That is more or less what matters for Varnhofer, right? And matters for our customers.
12:14They realize with interest that there is progress and that there's progress in certain scientific questions.
12:23But right now, we don't yet see a problem solved that is really of importance for a large company or
12:36a small or medium-sized company.
12:38And I think the confidence grows that that will happen one day. But I think also we don't really know
12:50how long it will still take.
12:51And I think it will not be the one moment or the one date or so. It will be a
12:57laborious path in which
13:01software, hardware, use cases, data, hybrid workflows, HPC capabilities will have to come together
13:11to really solve something in logistics, in manufacturing, in financial modeling, in chemicals.
13:18So, these are typical areas where we expect something, but we don't have it yet, you know?
13:29I mean, I might want to add to that a little bit and just, you know, kind of say that.
13:33I think, you know,
13:33one big thing, though, is like, you know, it's not the right time to necessarily sit down and wait,
13:38right, for things to happen. It's really about how to actually start to use these tools.
13:42In many ways, the situation we're in with quantum is there's a roadmap, right, towards machines that
13:51will be at a scale that we can solve problems which you can prove with pen and paper, right?
13:57At a scale with the number of qubits that we would have with the number of operations.
14:02And that requires building towards fault-tolerant, what are called error-corrected fault-tolerant machines.
14:08But what's also, you know, really interesting about the time that we're in right now and, you know,
14:12to Loic's point and what are being built and deployed are that these machines are at a scale
14:19at hundreds of qubits, running thousands of operations that are beyond what you can exactly
14:25simulate with classical computers. But what's that good for? We need to have the users to actually
14:32explore how it interfaces with HPC, how it ties in with workflows that are pushing the limits of
14:40solving problems in chemistry and optimization today to find that kind of heuristic advantage.
14:46And that's why it's really such an exciting time that we have the capability to, you know,
14:50deploy these. They have the capability to build on top of it to find, you know,
14:54the algorithms that are really defined this current phase.
14:59And maybe, Loic, since we also see that many people are currently massively interested in what
15:07are the compute capabilities of a quantum computer, but at the same time we see that
15:11we have this massive need for additional compute capacity also in order to run all of the AI models
15:16that we're currently developing. How do you see the interrelation between AI and quantum?
15:22Yeah, well, I think it's a very interesting question and we should not try to oppose AI with quantum.
15:30It's really important to have, like, to understand the complementarity between those two approaches.
15:35One example is, like, right now with the machines we have, they are very good at problems which have
15:41a high complexity, but low inputs, low outputs. So this is the kind of things that quantum machines are
15:47good for. And this is complementary to what AI can do, which is typically, like, big data sets,
15:54maybe not so much about the computational complexity of the task, of the underlying task,
15:59but really about large scale models. So we really need to think about the complementarity between the two
16:07ways of compute. And actually, when we talk about quantum advantage,
16:11this hypothetical holy grail that we are all pursuing, like, there is a high probability that when
16:17a company goes live in production, what they will implement will actually be a hybrid workflow,
16:24using CPUs, GPUs, QPUs. There is no, like, there is a relatively low probability that it will be only a
16:31QPU-only task. It will be something that will be distributed among the various sources of compute to
16:37actually try to leverage the strength of each. So that's our vision of actually how it will be
16:43deployed at scale in enterprises.
16:48Yeah, and I'd like to just, you know, add on that, right? Like, I think the wrong thing to think
16:53about is that, you know, it's AI versus quantum, right? AI, if anything, is an application of computing.
16:59It's an application of the resources that you have to help you compute. And, you know, to Loic's point,
17:04it's really about how are these different processors handling math that is useful for
17:12computation. CPUs, of course, we know they're good at basic math addition and so forth. GPUs, of course,
17:19enabled a lot of what we can do with AI today with language models. And a big part of that
17:23is
17:23they're phenomenal at doing tensor math. QPUs, this is where it comes in, in terms of providing a
17:30differentiating computational capability of computing quantum circuits. It's a different
17:37language that we can use and leverage. And so it comes down to how do you actually best use all
17:43of
17:43these. And, you know, from the AI perspective, there's a lot to be gained both from finding the
17:50right types of algorithms where all three of these can, you know, influence AI. And at the same time,
17:55there's a lot going on in terms of using AI to find new algorithms and find new ways of using
18:00quantum
18:01circuits better for solving problems.
18:06Well, I mean, on the surface, these are two different technologies, you know, on the surface,
18:12first of all. But they are competing, they are competing for funds. And right now, I'm afraid AI is in
18:19the
18:19advantage in that competition right now. Because AI delivers real world value, and everyone knows it,
18:25and everyone sees it, and everyone feels it. So we have to struggle a little here at this point.
18:32If I want to make the connection, and some of that has been mentioned already, I think one connection is
18:40sort of trivial, quote unquote, that is that AI will help to design better software and better hardware.
18:47And I'm sure it does, as we speak. That's the more trivial connection. There is one that is
18:54potentially, I think, deeper and more profound. That is, right now, AI is based on classical neural
19:01nets. And one could think of quantum neural nets, or quantum spin glasses, basically, which may have
19:11a great advantage over classical neural nets, because the information content is much, much higher that
19:18you can store in such a quantum neural net. And also, the training could be much faster and much more
19:26efficient due to things like tunneling effects and what can happen in quantum systems. So that is in
19:33the infancy, as far as I know, at least. But it is building one giant on the other, basically.
19:42Yeah, I think, so you mentioned that we mentioned the AI for quantum, designing algorithms, procedures,
19:49quantum for AI, like quantum neural nets, for example. I think there is also the area where
19:54you use data that is generated by a quantum processor to feed in a classical procedure. I think this is
20:01a
20:02very exciting emerging intersection between AI and quantum, especially in fields like material science,
20:07for example, or chemistry. Because AI struggles because it doesn't have access to the ground truth,
20:14which is inherently quantum mechanical. You have to understand quantum mechanics to actually predict
20:20the reactivity of a material, of a molecule. And this is an area where quantum computing naturally
20:27fits in. And you can use these wonderful machines, our quantum computers, to create data sets that are
20:34completely new, that you cannot create in any other way, because there are no approximations. It's really
20:41quantum mechanics, all quantum mechanics, and then use classical AI models on top to train on that quantum data.
20:49I think this is this is very exciting area of research right now and complementarity between AI and quantum.
20:56Excellent. Maybe one more question, since you already mentioned the competition for for funds between quantum and AI.
21:02One other competition that we also see is that a lot of the funding that we currently see for for
21:07quantum essentially also
21:08goes to U.S.-based companies. While sort of there is definitely a push then also on the European side for
21:15can we build sovereign players? Can we build sovereign supply chains? How do you see it in terms of what
21:21would actually be required?
21:22Do we require this to be a sovereign technology? Do we require dedicated supply chains? And how should we deal
21:28with this?
21:29Well, I would say what what we need to do or what we need to be in Europe is a
21:35competent partner
21:36in this game. And I mean both competent and partner. And and competent means that we need to have our
21:45own
21:45hardware technology, our own software. I think that is necessary and we should strive for that.
21:53But we cannot strive for autoky, right? That doesn't make much sense. It's not efficient and
22:00will not bring us forward. So partnership and and being part of the international ecosystem and have joint
22:08development projects, etc. Be open to that is likewise important. Competent partner. One could put it that
22:17way. I mean you you have to bring something to the table to be to be in the game. But
22:24you also have to be
22:25able to stand up from the negotiation table if something goes wrong. So so you have to create
22:32your options. And I think that is the the core of the sovereignty discussion, actually.
22:40And maybe Jerry, we already before we we started, we talked about your center also in
22:44Ing. And how do you see it? And how is IBM also dealing with the topic?
22:48Yeah, so I mean like we are a global company at IBM, right? And we are certainly looking to partner
22:57all around the world. And and a big part of, in fact, how we've worked to deploy these machines and
23:02and put them on the cloud is to give that access, right? And and also, in fact, we've been deploying
23:10systems all around the world in various different client locations. We have a data center that's in
23:15in again in Germany. We've deployed multiple systems in in Japan and one of them in the super
23:22compute center there at the Reichen Institute. And at the same time trying to build ecosystems around
23:29that, right? I use Japan as a real fun example because there we're deeply engaged both in driving
23:36the kind of algorithm discovery and the application work on top to to build, you know, software tools
23:42and to drive applications. And also, we've been trying to work with the local ecosystem to build
23:48hardware supply chain vendors, right, where there's really, you know, the a lot of the technologies that
23:54can be built, especially looking forward for what we need. There's a lot of opportunities there for,
24:00you know, economic development in these local areas to help bolster everything that we we want to make,
24:05right? And so, you know, there's really a truly, I say, a ecosystem to build in these various
24:11different locations, Japan being one example. And certainly, you know, we look to do similar here
24:15in Europe.
24:18Maybe like one, one other question that often comes up is you already mentioned some of the
24:24use cases that one would have for the technology specifically also in terms of actual, actually
24:29performing material research. Of course, sort of, there are many, many use cases for quantum computers
24:34that would be out there. What are like the key use cases for you? And what do you believe will
24:39be the
24:39first ones that would actually become reality?
24:42Yeah, well, that's a great question, top of mind for us, for our team, of course, because we really
24:47want to target those use cases that will come first. We do believe that the things that are related to
24:55quantum simulation materials, chemistry related, will be the first to actually be benefiting from the use
25:01of quantum computing. But beyond that, of course, we need to expand. The technology on quantum computing
25:08has a tremendous potential. There are a few use cases that we need to prove that we are, like, completely
25:16dominating other sources of compute so that we can show the value and iterate and find the new use cases,
25:21new
25:21verticals, and so on. At Pascal, we are focusing on a few key verticals to work on. First one being
25:30the financial
25:30services, financial industry, because it's an industry which is very eager to use new technology. There
25:37is a lot of things that can be tackled using quantum computing, so we are working on that one.
25:42The second one is energy and oil and gas. So we've been working closely with energy partners in France,
25:48with EDF, but also abroad, with the Aramco in Saudi Arabia, for example. We have installed a machine,
25:54quantum processing unit, in Saudi Arabia, working very closely with Aramco personnel to find new use cases
26:00for the energy industry. And we also have, of course, logistics as something that is of interest for us,
26:09for example. But those, I would say, are the main three, like quantum simulation materials,
26:14chemistry, financial services, energy, which are really the top focus for us right now.
26:23Maybe, Axel, what are you hearing from your industry conversations on the topic?
26:28Yeah. There is one class of problems which I would probably call optimization problems, and then
26:36logistics, manufacturing, energy systems would probably fall into that category. Where I'm not
26:43so sure is whether whether we really see the best of the quantum advantage in this class, actually,
26:50because the classical solutions are not so bad in that. Yeah, the difference to the theoretical optimum
26:57is small. So I'm not so sure that that this will be the breakthrough, so to say. There's another class
27:07of where one could say, okay, here we could find solutions which otherwise could not be found.
27:16In, for example, life sciences of pharma, drug discovery, for example, or the famous protein
27:23folding issue. That is really hard for classical approaches. And here, quantum could really
27:31bring a significant step forward. Materials chemistry, true. Also, financial modeling seems to be,
27:43in my view, somewhere in between. It has aspects of optimization, but it is also very hard due to
27:54the correlations. So if you would want to simulate asset portfolios or equity values or so, there are
28:02correlations and they are dynamic. The correlations are not static. So that is, in terms of physics,
28:09I'm a physicist, it's a many body problem with changing rules of how things interact. That's really
28:15complicated. So here probably is a very complicated but possible case for quantum advantage, I think.
28:27Yeah, so you know, from certainly my perspective, a big direction is, as I've already said many times,
28:35but you know, the material side and looking at chemistry, right? That's just, I think we've always
28:40heard that about being kind of, you know, Feynman's dream, right? Of using quantum computing to look
28:45at these types of highly complex quantum mechanical molecular structure. And we were actually seeing
28:51that already in many cases. For example, in our recent published work with the Cleveland Clinic
28:58and working with the Regan Institute in Japan, actually simulating proteins out to 12,000 plus atoms,
29:05right? Using a quantum centric workflow where we're able to take this large molecule, use a
29:11supercomputer to fragment it down to pieces that can be simulated on a quantum computer and then
29:17diagonalized on a HPC. So there's really, you know, the potential there for a lot of this push towards
29:23that molecular structure direction. Industries are looking at it, you know, from the perspective, for
29:28example, Boeing, one of our clients, looking at, you know, better materials that are resistant to
29:34corrosion, Moderna, looking at, you know, various types of materials that are, you know, biologically
29:39relevant and so forth. But also, you know, I think, you know, these other directions that have been
29:43mentioned, I wouldn't write them off completely, you know, in terms of optimization. I think there's a
29:49lot, in fact, there are working groups that are looking at this in terms of what are those kind of
29:54problems that we know are certainly very challenging for classical, and there are probably some classical
30:00solutions but have the right kind of structure that can work well on a quantum computer.
30:05And last year, with a large industry optimization working group, we actually published a paper of,
30:11like, what are the best kind of candidate optimization problems to look for? And, you know,
30:16it kind of is like a list to kind of, you know, start to work through and try to benchmark
30:20as we improve
30:21our quantum capabilities to work against. And then, you know, one third area I do want to point out also
30:27that I think is very exciting for me is partial differential equations, right? And so partial
30:31differential equations are behind so much of, so many problems that we have, that we rely on.
30:39Weather modeling, fluid dynamics, and things like that, which are really computationally intensive.
30:46And there's been a lot of work here in terms of algorithm discovery to really push what can be
30:51potentially solved on quantum computers. Last year, there was a paper that really, you know, brought in the
30:57idea of doing something like a 1D Navier-Stokes. But we expect to see more, right? So this is a
31:03part of the exploration phase where we're really excited to see algorithmic development. Part of that
31:08is done, you know, in a way where you can do the math and you prove it. Other ways is
31:12really by using
31:13the machines and really trying things and, you know, finding those kind of heuristic algorithms that
31:17drive advantage. Thank you. I think when I was much younger and starting physics, I had three dreams
31:28during my lifetime. To see a flight to Mars, to see the room temperature superconductor, and to see the
31:34quantum computer in actual reality. I've given up on Mars. I've not fully given up on the room temperature
31:42superconductor. I'm fairly certain that I will see a quantum computer in actual reality.
31:50But I think sort of this shows that essentially taking a bit sort of the look forward and to look
31:56sort of what the future will bring is something that is crucial also in order to see what is the
32:01potential of the underlying technology. So maybe if you would look like five years ahead, like sort of what
32:09would be the future vision and what do you think will be a reality by then?
32:14Yeah, well, thanks a lot on this. Actually, so if we, before looking in the future, if we just look
32:22at the present right now, what has changed and what it will enable for the future. I think it's the
32:27right
32:27way to look at this problem. Because in the past, like quantum computers were things that were really
32:32restricted to very few people on earth, specialists, PhDs, in physics, really running a machine, and
32:42very complex machine, which was mostly not working, and that we tried to make it work maybe one percent
32:48of the time to get one hero results. And right now what we have is a fleet of machines like
32:54IBM machines
32:55that are accessible to the cloud. They were the pioneers actually believing that by putting those machines in
33:01the hands of people, you could get something useful out of it. And right now there are all
33:06the modalities that can be accessed. If you just have an internet connection, you can use quantum
33:11computing machines right now, Pascal machines, IBM machines. And that's really, I think what we will
33:20see is that the algorithm development and use case applications will transition from like specialists
33:25like ourselves, the team at IBM, the team at Pascal in very specialized institutions to actually having
33:35something that is developed by the end user themselves. So what I expect that we'll see more
33:42and more demonstration of quantum advantage in the next few years. And in five years from now, what we will
33:47see is like every day, new advancements that are created by the end users, and something like an
33:53industry, an ecosystem that is autonomous in using those machines, very much like what we see with AI
33:59right now. Like you have the hardware, you have the infrastructure, the low-level software, and like
34:03we have this explosion of applications and companies working on top of it. And so that's what I expect
34:09to be happening with quantum in five years. Jerry, do you want to explain your future? Yeah, no. So I
34:17mean,
34:18like, I think, you know, we at IBM have a very concrete view of certainly the next few years, right?
34:24So we have
34:26targets for building a and deploying a fault-tolerant quantum computer at the scale of 200 qubits and
34:34running out to 100 million operations by 2029. And then, you know, beyond that, we have plans for in the
34:40mid-2030s to build a 2,000 logical qubit machine that can run out to billions of operations, right?
34:47You know, we're kind of pushing into that cryptographic relevance type of machine. And, you know,
34:53our roadmap is very much grounded in the technologies that we're building and the innovations that we know
35:00that we need to hit from an engineering perspective to get there. And so, you know,
35:03we're confident in that, especially as we build various devices year after year. But also, you know,
35:09I think it's really exciting to just look at this transition as we move forward that, you know,
35:17part of how we use these and how we solve problems with quantum, we hope really becomes more and more
35:24mainstream. So, as you mentioned, right, that like it becomes a tool as part of your computational
35:31arsenal, right? Not unsimilar to how GPUs went from just being, you know, gaming modules to helping you
35:40get, you know, superior graphics in your digital movies to becoming the computational tools that
35:48they are today. We see that similar evolution, hope for that similar evolution, you know, for quantum,
35:53right, that we get this algorithmic scientific push by using these as tools next to the best
36:00computational pieces available. And, you know, I certainly hope that, you know, in maybe the five
36:05year timeline, that becomes more and more relevant. And maybe you have your agentic workflows that are
36:10calling quantum computers to solve problems as part of your workflows. And then, you know,
36:15beyond that, you know, into 10 years forward, I hope that you don't even need to know that
36:18there's a quantum computer behind solving some of your problems, right? In fact, it's all hidden
36:22underneath the layers of, you know, you're just you're just being able to solve way more
36:28in terms of the problems that you're able to address.
36:33I can can only emphasize what has has been said already. If I make that comparison between AI
36:39deployment and quantum computing deployment, if I just narrow it down to just my own Gesellschaft,
36:46to Fraunhofer. At Fraunhofer, everyone knows that AI is relevant. I mean, everyone in the world
36:52probably knows. But still, we need a lot of training and education and have all sorts of formats to
37:02exchange best practices inside Fraunhofer. There's still a lot to do so that even AI can be used as an
37:12everyday instrument to its best potential, so to say. Now, quantum computing, we know quantum
37:19computing will matter. But in Fraunhofer, even, I'm not sure that too many will
37:25under sign that that this is will matter in five years. Of course, there are those that are dealing
37:32with it. So, but we will need a similar effort. And I hope in five years, we are there. That
37:42really
37:44disseminates all the technology that is necessary, all the thinking that is necessary, all the methods
37:51that are necessary to use quantum computing, which is not trivial at all. It's not just normal algorithms,
37:57as we know. We need to understand a lot more, or at least have software layers that make it easy
38:07for
38:08everyone, at least that. And then bring it to every single of our 30,000 employees in the end, because
38:18it
38:18will be one of the most powerful instruments you can imagine, right now, at least. So that is a long
38:27way
38:27still. But I hope in five years, we are more or less at the stage where we are right now
38:34with AI.
38:37Thank you. I think then, if I should summarize a bit sort of what I think sort of I have
38:45at least
38:45learned during the panel that we had today. I think sort of the first thing is, I learned that
38:50quantum advantage is not about theoretical problems, but actually about solving practical business
38:57problems and creating practical business benefit. And this may be a bit harder than doing this for
39:01theoretical problems, but it will definitely also be more rewarding than also for the industries.
39:07I was very impressed, essentially, also to see within this round, I mean, we're all coming from
39:12from different backgrounds that I think, Axel, you mentioned sort of the spirit of partnership
39:17is still there, that we should essentially try to solve the difficult both scientific as well as
39:22also engineering challenges together, and that we should really see ourselves as competent partners
39:28who would be able to then also solve the corresponding problems together. And what I feel is there is a
39:35lot to
39:35come then also, as we discussed, also in terms of the respective use cases, specifically in some of the
39:40key focus industries around chemicals, pharma, financial services. This is not the time to wait,
39:46this is the time to act. And this is a time where we now see also an inflection point in
39:52terms of the
39:52commercialization of the underlying technology. So everyone who will say the quantum computer comes
39:57in the next five years, and I can lie myself down for the next five years, will definitely do a
40:01mistake
40:02in terms of the future business performance. So I think this was a very energizing discussion, at least for
40:09me. I think what I definitely also take away from here is the best is yet to come, in terms
40:15of real quantum
40:17advantage, in terms of the usage of quantum computers. I think Jerry, you said sort of whether we actually use
40:21a quantum
40:21computer in the background or not should not be something that actually matters to the actual solution of the
40:26problems. So I'm definitely looking forward to a very interesting age where we'll be able to solve
40:32some of the problems that we cannot even imagine today. So with that, I'm very much looking forward
40:37to the future. I would like to thank you for your kind attendance during the session, and I would like
40:41to thank Axel, Jerry, and Loik then also for having this panel together with me. Thanks a lot for this.
40:49Thank you very much.