Honeywell International Inc. (HON) Earnings Call Transcript & Summary

Good morning or afternoon as the case may be, everyone, and welcome to the HCTS presentation of "A Quantum Leap or an Entangled Dream?" This is a roundtable that I'm hosting. I'm James Sanders, a cloud analyst with 451 Research. And I'm joined today by Dr. Ray Beausoleil, the Senior Fellow and Senior Vice President, Director of Large-Scale Integrated Photonics at Hewlett Packard Enterprise; Dr. Bob Sutor, Vice President for IBM Q Strategy and Ecosystem at IBM Research, and also the author of "Dancing with Qubits: How quantum computing works and how it can change the world"; and Tony Uttley, President of Honeywell, Quantum Solutions. Quantum computing is something that 451 Research has started tracking in the last few years. And given the potential effects that quantum computing can have on enterprise, it's something that requires a decent amount of attention and has a substantial learning curve with it. For a little bit of a level set, here's a video from Dr. Owen Rogers of 451 Research. [Presentation]

All right. And with that level set, let's kick off with the questions.

It is interesting to note that of the 3 companies our panel scale from, only 2 of these firms, IBM and Honeywell, are pursuing the creation of a quantum computer. Likewise, as a consumer brand, Honeywell is known more for thermostats, air purifiers and home security products. So as a small sweep to my question, Bob, why is IBM building a quantum computer?

Well, we'll start by saying IBM is a computer company. We're over 100 years old. We, for example, introduced the IBM mainframe in 1964, so we have a long history in this. But also, we have a long history in research. IBM Research started around 1945. And in fact, in 1970, so 50 years ago, one of our scientists, Charles Bennett, coined the term, quantum information science. So while that was largely theoretical, 20 years ago, we had a working qubit. 10 years ago, we had a working 1 qubit, and now we're up to 65 with plans to go much larger. We very much feel that quantum computing will likely be, together with existing and evolving classical computing, the most important computing technology of this coming century. It's important to get in now. There's a lot of innovation that has to be done, both the theory, in terms of the engineering, building out systems. So we feel that the benefits, and I'm sure we'll touch on use cases, are things that our clients will want to do. They're already exploring with us. And so we made an early start on this and have come quite a long way.

Thank you. Tony, why is Honeywell joining the field of quantum computing?

Yes. It's a great question and certainly one I get all the time. It typically starts with people not knowing 2 things versus what does Honeywell actually do. And what we do is we make some of the most sophisticated control systems that control both chemical and petrochemical plants around the world. We have building control systems in over 50 million buildings around the world. We have products that have been a part of every human space flight in the last 50 years for NASA. We are a system -- we're basically a systems-oriented company, where if you think about what it takes to go make quantum computing happen, it ultimately becomes a controls problem. And regardless of the technology that's being used to do quantum computing, some of the underlying capabilities are very common. So magnetic fields, vacuum systems, cryogenics, precision control system. These are things that Honeywell has been doing within our aerospace business, within our performance materials and technologies business, within our building controls business for decades. And what we found was that if you take and put all that together, you can build a quantum computer. So as we looked out as a company in terms of the industries that we participate in today, how profoundly they'd be impacted by the utilization of at-scale quantum computers, we realize not only do we have the technologies to go build them, but we have the kind of use cases internally where [ we own our ] biggest customers. So it really was just a match made in kind of perfect harmony to be able to bring quantum computing forward.

And Ray, why is HPE not building a quantum computer? Ray, you might be muted.

I am now unmuted. The short answer is that in this day and age, at a conference based on the cloud, we don't have to build a quantum computer because people like Bob and Tony are. So HPE is in the middle of a massive pivot to looking at supplying infrastructure, problem-solving capabilities and know-how as a service. And so what is important to us is that we can bring to customers a variety of ways of solving problems of interest to them. And we can provide those methods, those methodologies, without owning all the infrastructure that we are making available. And so what we decided to do instead was to pursue other hardware technologies that we thought would be more relevant to widespread availability of what we're calling accelerators. And we think of a quantum computer as one of a class of accelerators that is ideal for solving quantum problems, to work on interconnects, fabrics, other types of accelerators. And so it's a question of where to put your emphasis in hardware development, we decided we did not have to make qubits.

Okay. And likewise, 451 Research does consider quantum computers to be one entry in a class of compute accelerators alongside GPUs, FPGA or ASICs and in-memory processing that will be used in the post-Moore's Law world to extract further performance gains year-over-year. Not all of these technologies are necessarily relevant to all fields, however. So with that in mind, what industries will come quantum computers affect? And when will these effects be realized? And let's start with Tony on that.

Our belief is that almost any industry is going to be impacted by quantum computing and some in ways where you may not even know it's happening. Much like you get on your smartphone today, and you do a search. The [ background ] processes of what's happening when you type in that command, you may not [ realize ]. Similar, once you go down this path of quantum computing, you're going to be in the same position. You're going to have parts of jobs that are much more efficiently done by a quantum computer and other parts that may be done by classical computers or by GPUs. And so you'll see these split. In fact, you may not even see them split. They will just happen behind the scenes. But some of the industries in which we will see earlier impact are going to be areas that Honeywell is in today. Our aerospace business, our oil and gas business, our chemicals business, these are ones where -- the use cases that have some profound impacts around optimization, around machine learning, around quantum chemistry start to have such a big impact as you realize the full power and scale of quantum computing that it was worth putting investment in now. And so the time frames are something that is still we're going to see evolve. We are seeing use cases where we can take problems and shrink them down to the size that quantum computers are able to do today, and that's exceptionally important because this is an era where you can do that computation out of a quantum computer, and you can do it on a classical computer, and you can validate that you are getting the [ answer ]. We are very quickly, as an industry, crossing that boundary where you will not have that classical fact check. We'll not be able to do another process, another means of getting to that outcome except the quantum computer. So that's where we see this going.

Okay. Bob, how about you?

Well, Tony, as he started talking about how people may not realize they're using quantum computing, it reminded me of the Linux space, so when you would ask people, "Did you use Linux today?" In fact, you can still ask people, "Have you used Linux today?" And frequently, they would say, "Oh, no, I'm a Mac. I'm a Windows user," or something like this. And then you say, "Well, have you used your phone? Have you gone to a website?" because Linux is pervasive throughout many different industries in many different ways. I think there are 3 prime examples that people usually use about the first applications of quantum computing. And Tony mentioned some of them. So one is chemistry and because quantum computing is built on the ideas of quantum mechanics, and that goes way back to 1900s. And that models, we believe, what happens in very small. So here, we're talking atoms, ions, photons, many things like that. So therefore, if you're modeling physical processes which behave the laws of quantum mechanics, it makes sense to compute with quantum computing, right? So that's one area. The other area is financial services. Now is financial services really all that special as an industry for quantum? Well, the application areas are simulation of very complicated environments, such as thinking about financial instruments, stocks and bonds and derivatives, and the way they're all created in different ways. How do you possibly do risk assessment when things are happening, such as a hurricane is about to hit? Can you do these things quickly? And straight-up optimization problems. Optimization has been around for a long time, and there are some applications that has been used for this. Now financial services is very special among the industries because they invest early in every single new technology. Whenever they spend upfront, if they are successful, they will probably make it back in about 10 or 15 minutes, right? And so that makes sense. But it will -- I believe what will happen will be the algorithmic improvements, the way we actually run their use cases, will then spread to other industries, things like Monte Carlo simulation. And then finally, the other application is artificial intelligence. Now artificial intelligence is used broadly across many industries. In fact, more people should use AI, should use machine learning, deep learning. That's a fundamental message. Because quantum computing is very good at calculating certain things, technically, what we call linear algebra, we talk about matrices and vectors, that is the math down deep in AI. We believe it will improve the efficiency of what we do today for artificial intelligence as well as give us new methods. So AI is broadly -- it's across many different industries. Some will be at the forefront as we've seen chemistry-related financial services. But once others look and say, "Hey, I do that too," it will spread quite quickly.

And Ray, I know you've been on the record before as being a bit skeptical of AI in quantum computing. But I'd be interested in your insights to really all of that.

Well, I am very skeptical about applications of quantum computing in AI because the -- for example, there -- I don't think you can make an argument that you could use it to train neural network on enormously large databases. We don't really have quantum RAM yet. We don't have a quantum hard drive. And so -- and the basic algorithm that could be applied to give you statistical analyses of large databases, HHL, after the authors' initials, has a number of caveats that one has to apply to each problem separately. And so I'm not a huge booster of quantum computing for -- I mean, AI is really just machine learning. It's -- we're using the abbreviation because it's nice, but it's ML. I think that the first -- the question that you have on the screen there, I think that within 10 years, you could see a major, major impact on the quantum computers could have on chemistry, biochemistry, material science. And in fact, I -- people tend to focus on my skepticism for machine learning, but they don't focus on my belief that in 20 years, all information technology will be quantum in the sense that we will need quantum computers to design and develop state-of-the-art classical information technology that just has classical bits. And I think that we will reach a point where in 20 years, you can't be an engineer and not know something about quantum mechanics. We have been really successful for 50 years of abstracting away quantum mechanics so that engineers didn't have to pay any attention to it. Starting 10 years from now, I just do not think that will be true anymore. And one of the most exciting things about quantum computing being brought forward is not just the impact on applications but also the excitement it's going to generate in the universities. People will take more quantum mechanics classes. They'll begin to realize other places that they could apply quantum mechanics. They'll use quantum computers to design chips in 20 years. I think that I am incredibly excited about the field. I think some application areas are a bit of a distraction, though.

Okay. And just to verify, researchers would be using quantum computers to design classical chips. So for example, you wouldn't have a quantum computer in your smartphone in your pocket, but that could be used to build the chip that's in the smartphone.

Yes. Well, imagine now, if we actually go to a sub-nanometer technology node and you think about building transistors with gates, like we do now, you might have 2 or 3 silicon atoms in your gate. And it's not going to be possible to abstract away the stochastic quantum behavior of those gates. And so I think it will be a lot of fun to start thinking about how quantum computers could help us solve difficult quantum transport problems that give us a definitive -- and also, to take a page from the Honeywell book, I think that we're going to need quantum control systems to provide reliable classical behavior in what are fundamentally quantum devices, even if all they're doing is counting classical ones and zeros.

Okay. Well, let's move on to the next question. So immeasurable amounts of money have been invested in technologies that never quite made it. And this in part is a failure of business planning as not every unit of new technology could be frankly, equally applied to every business use case, despite perhaps [ overanalysis tends ] to do as we've seen with variety of blockchain implementations. Keeping that in mind, how should enterprises start evaluating if quantum computing is applicable to their business? Whoever wants to volunteer to go first on that one, please take it away.

Well, I would -- I'd modify your verb, first of all, and not say is and will say will be because the fact of the matter is there's no quantum computer on earth that will do things better than classical computers can do today. So as has been pointed out, we're talking about the future. We're talking about several years until we reach what we call quantum advantage, where quantum computers together with classical systems can do significantly better than a quantum computing system by itself. And so it's a longer-term view. One place to look is where are you doing high-performance computing today as an example. That's a good place to start. Are you in some of the areas that we've already mentioned, such as chemistry, simulation of physical systems, material sciences, financial services, things like these? These are the places to look. Now because we're talking about the future, it is critical that you think about when and how far out. As Ray pointed out, we don't yet have quantum RAM, quantum memory. But the keyword there is yet, right? We cannot access a tremendous amount of data. In fact, if you ever rethink to say humans are creating petabytes of data every day, thank goodness we have quantum computers to munch on this all. Just stop reading, just go away, that's ridiculous. So things will come in over the next few years, in fact, over the next few decades and even through the rest of the 2020s. What we think of as a quantum computer will change radically year-by-year because of integration, because of miniaturization and because of how we scale these systems. So if we have this conversation next year and 2 years, I think it will go very differently in terms of answering this type of question and how close it is to the present.

So I'll maybe go next. I have talked about quantum computing in 3 eras. There's an emergent era. The emergent era is quantum computers didn't exist, and now they exist. That's profound, right? That is a big deal, to the point where you can actually do quantum computation. I'll come back to that. But there's a third era, and I'm going to skip one, clearly, and say that's the era where you are doing classically impossible things, things that are -- where everybody flashes forward and says, 10 years from now, 20 years from now, these are the things that you couldn't operate a high-performance classical computer for 1 billion years and get there. And those are the interesting things to keep in mind and to work towards because many of them take a long time to operation. But there is this middle era, and Bob talked about it, where it's this idea of classically impractical, which means you potentially could do it using classical resources, but you don't today. And the reason you don't today is either because it takes too long. Meaning you can use a classical resource to be able to go generate the answer, but it might take you 6 straight weeks, and 6 weeks is too long. I have to have had the answer way before that, or cost too much, or in some cases, it just consumes so much power to be able to go and have that computation done that you don't. And as Bob mentioned, we will be crossing into that era of classically impractical very quickly. So why are people doing things today? They're doing things today for the same reason that if you rewind classical computing 60 years, you'd say, "Yes, you might be able to do that math in your head versus using on a classical computer". Today, you might be able to get that same answer using a classical simulator, but it expands exponentially. That's the whole beauty of quantum computing. It will be a binary switch that gets flipped, where we are doing those computations that can be validated and crosses over to you cannot. And it becomes a superpowerful tool to be able to apply in some of these early use cases. That's why people are spending energy today. That's why you have oil and gas companies and financial services companies and chemicals companies all investing now to go after it because these are also process changes that take time. For anybody who's in an industry right now where you're making an offering, you realize that your development cycle is usually measured in at least months, oftentimes years. And for developing a new molecule, it may be a decade-long process. And if you're not planning now on how you integrate quantum computing into that process, then you're going to be behind because the companies that are focused on it are starting to take advantage of, again, those smaller problems that can be done on quantum computers today.

And let me also just add, we're talking about quantum computers on real quantum hardware. That's the future. So when we do things like simulating quantum computing, when we, let me say, get inspired by quantum computing, but we still do things classically, we have limitations. We all are already at qubit counts that are greater than what you can simulate classically. So this is one way in which the world is changing, and it's why people need to get experience for these eventual use cases on real hardware.

I'll add briefly that I think the most direct answer to this question is that you can take advantage of it now. You can -- just as Bob and Tony have said, now is the time to start exploring. If you don't want to -- I mean, these things are available on the cloud. This is brilliant, right? So anybody has access to it. If you -- you can hire quantum mechanics and quantum information technologists, and you can put them to work using the gate-based libraries that all of the hardware providers are making available, or you could also take your algorithmically savvy classical computation staff and have them go interact with some of the new quantum software start-ups. The nice thing about putting these things on the cloud and encouraging people to experiment now is that over the course of time, the whole stack will get built. So there will be algorithm-based quantum software companies. There will be classical control supercomputers that can guide and direct quantum computers in action. There will be very smart tools that can be used to figure out when it makes sense to send your problem to a quantum computer instead of to, say, an existing machine solver. The more people that get involved now, the sooner that entire stack will get built up, and I consider that to be the goal. But right now, if you're quantum savvy, the libraries available to you from the hardware vendors on the cloud can get you started now. And if you are not, then one of the software start-ups is a good place to start.

Just to punctuate that, to make the point about of being on the cloud. Users routinely run over 1 billion hardware circuits a day on our machines on the cloud. Tony, you also have a cloud offering as well. This is how it's used. Quantum computing, as we know it today, is being born -- was born on the cloud. So all the advantages we've been talking about are there today. And I mentioned that number simply because if you're looking around and you're thinking, well, no one is using this, could be further from the truth. Tens of thousands of people are already experimenting with quantum on the cloud.

Okay. And so as you've mentioned, quantum computing as an industry has moved from the hallowed halls of university research labs onto various cloud platforms. And Bob, as you said, the current class of quantum computers, otherwise known as the Noisy Intermediate-Scale Quantum or NISQ systems, are useful for experimentation, but they don't provide an advantage for workloads today. Granted, like you said, we've gotten to a point where the simulators can't keep up with the qubit counts, but the resultant effect is that they're still not quite there. So what obstacles in either software or hardware must be overcome for quantum computing to become a mainstream technology? And granted, I know you just said that there's 1 million circuits run a day, but is it mainstream really?

Well, billion. Just I thought I heard million, but it's billion. Well, I also said a little bit earlier that what we call quantum computing is going to be changing radically over the next decade. This NISQ term or sometimes called approximate quantum computing, that had been going on for a very long time before that term was actually coined, right? And we can debate when that happened, but it's certainly more than 10 years ago. So the way I think about it, there are 3 main things: scalability, scalability and scalability, right? So what that means is, ultimately, we need quantum computers that have enough capacity and good-enough performance to start tackling these problems. Now we'll do it via different algorithmic approaches. But just to talk about some of what to expect is that we've published a road map. And our largest machine is 65 now. We expect -- 65 qubits. We expect 121 next year, 433 the year after, and then we'll break 1,000 in 2023. So that is one form of scalability. The other is the performance. And Tony, you talked about quantum volume and your great achievements in that regard. This is measuring really just how good your machine is, both in terms of qubit -- good qubits that you can do things with for a long-enough time. That will be improving as well. But what people need to realize is other terms like error correction are going to start coming in, partial error correction. That will change the algorithms. So for all of this to work, to see what you imagine you want to do on a quantum computer, you need great software. The reason why we can do so much on classical computers today is because we have optimizing compilers that have been developed over decades. We are also, through projects like Qiskit, developing very good software tool chains, very good compilers that can take your quantum algorithm or the way you write your circuit and optimize how it will run on whatever particular hardware you have. Now that might be completely different types of qubits, or for us, just as you're scaling this up, and you have more and more qubits, you have to lay out the circuit correctly to optimize performance. So I hope people remember this. It's not just the hardware. It's got to be the software as well can really make that hardware hum and sing and get these great results we're expecting.

I would layer in that it reminds me -- part of the question reminds me of classical computing as an analogy where somebody famously said, "Hey, I think there's only going to be maybe a handful of these computers that exist on the planet." And it's just -- frankly, just a naive view of what these tools can be used for. And first thought is I could -- if I have a hammer, I could go design an entire university campus. Like, okay, maybe given enough time. Why don't you start with what that tool is good for today? And that's where the ecosystem is rapidly expanding. And that's the -- doing algorithm, both design and iteration today. That is those end customers across industries like financial services, oil and gas, aerospace, chemicals that are iterating today because the tool becomes more and more useful, both with your own expertise and with the capability expansion. And so what Bob said, I totally agree with. These are systems that the folks who are designing hardware right now and are rolling it out have generational plans in place already where these are being scaled in multiple dimensions. They're being scaled for the number of physical qubits. They're being scaled on something called fidelity, which every [ person ] who is a technical purist will hit me later, but it's kind of an accuracy -- how accurate is the system today as well as how connected are all these qubits. When you have a computation, can you bring them all there at the same time? And these dimensions are being expanded at the same time that you're layering on these iterated algorithms and this abstraction layer of software so that you don't have to be a quantum theorist, you can be a chemist, or you can be a data scientist. And all of that is happening rapidly, and it's happening right now. And so it's almost a question of when is mainstream, mainstream? Do -- are there hundreds of organizations out there, thousands of people who are all actively involved in this? Yes. That is ongoing right now. And in fact, one of the scarcest resources that exist is just capacity, right? There are a few companies in the world who are making these systems and having that available capacity to go and iterate on them to get real answers out is one of those things that's driving this performance improvement, driving this capability improvement. So it's really a question of when will it be integrated in some of the business processes that you might expect. I think that's going to roll out in stages. When is that going to be something where it's ubiquitous? That may take a little time. There will be a quantum cloud. There's just no doubt of when there -- there will be a quantum cloud. When there will be a quantum cloud is probably the most debatable. But you will see, or again, you may not, but it'll be happening anyway behind the scenes, business processes where algorithms are just dissected, and the parts of it that can be very efficiently run on a quantum computer are going to be done there. The rest of the processing may be done on either CPUs or GPUs, depending upon what you're trying to go achieve. And you will see this hybridization of outcomes coming from this combination of quantum computing.

Thank you. Ray, do you have anything to add to that?

Yes. But not -- I'm not in the qubit-making business, and so I'm going to look at this a little bit differently. The final obstacle is a demonstration that it really matters. So in other words, imagine a day when a result is published in a scientific journal or a -- the Wall Street Journal, Harvard Business Review that is so profound, so amazing and so important that it changes the way we think about the subject matter involved. The fact that, that result was obtained using a quantum computer is actually secondary. When that happens, that's a killer app, that's the ability to say we now can do something we have never been able to do before. And you're going to get a ton of people who show up and start looking at this probably because they've been using quantum computers in the cloud for so long. People can head in that direction. You'll see a tremendous amount of innovation. And in the enterprise, you'll see opportunities to realize return on investment. And so the final obstacle is not a technical one, although it involves technology. It's the perception that all of this effort was absolutely worth it because here is this one outcome that is going to change the world. And "Oh, yes, we did it with a quantum computer."

Well said. That is pretty -- oh, here we go. Sorry about that. So there is one audience question that is pretty relevant to where we are now. And for everyone at the audience, if you have questions, please put them in, and we'll get to that at the end of the session. But I do want to just touch on this now because it's top of mind. So aside from the technical feasibility and commercial scalability of different approaches like superconductors, ion traps or atomic qubits, do these different approaches have unique applications or problems that they each address? Or are all quantum problems essentially the same? And I know, Ray, you talked about like delegating certain workloads to IC machines, and Tony, you've mentioned delegating problems to either a quantum computer or a CPU or a GPU. So I'll pose that to you and see what you're...

I don't actually -- I think this is a more profound question, and I'm eager to hear Bob and Tony answer because they're asking does a particular underlying hardware, which carries with it a particular quantum representation, native Hamiltonian, if you will. Does it make a particular hardware substrate more proficient at certain types of problems than others? The -- I think that's a fantastic question, and I'm going to get out of the way now.

Yes. I think the great thing about where we are in quantum computing is that there are things to be explored across technologies, meaning, is there going to be a winner, a single winner? There may be opportunities, and there is very likely to be opportunities where a technology could be superconducting, or it could be topological, or it could be photonic or as we're going down a path with trapped ion, do specific jobs better than others, just more efficiently. And ultimately, when it comes to some of these heavy computational problems, that's what you're looking for. You're looking for something that can be done very natively efficiently. And so some of that work is still to be done. And that's with companies like IBM and Honeywell working side by side that says, "Hey, let's take this problem. Let's chunk it down into pieces where this part of the problem is actually going to be run on a trapped ion hardware. This part of it is going to be run on the superconducting. This part of it is going to be then shuttled back out to be processed classically." And those kinds of, I'll call it, experimentation, but it's really for a future-looking true business use case, they're just getting started now. And I think they're going to need to continue to evolve to see how that goes.

So a few remarks. First is rather direct. You don't get to be a quantum computing company because your marketing department says so. It's because of what your science and your engineering departments create. That's the first thing. Second is I would divide up what people may somewhat vaguely think of as the quantum computing industry into 2 parts. What I'll call the universal gate-based quantum computers. And that's the sort of things that Tony and others do with ion traps, what we and others do with superconducting, some of the photonic approaches and things like that. And then there's another industry, which is the optimization industry. And so if you're doing simulated annealing, any which way, and that's all you can do, really, you're part of the optimization industry. So I really keep very focused on this part of this universal gate-based -- gate- and circuit-based part of the industry. And then you have a couple of fairly technical requirements. So we talked about this notion of qubits, of quantum bits, it goes back to quantum mechanics. You have to use things as we sell into the deal like superposition. And that just goes with the territory. That part is easy. But entanglement, this other physical concept, it's very weird. It's what Einstein called spooky action at a distance. This is the heart and soul. When we associate the word exponential in a good way with quantum computing, it's this entanglement concept that gives you that. And so you have to have that. You have to be able to do interesting things not just with single qubits but with multiple qubits at the same time, right? So we try to educate people over time and then perhaps all the time to get to this. We're building out a stack where the top level will be very easy to use by almost anybody, the lowest, lowest level will require experts. But there are some real technical requirements to play here. And for those of us who are exploring the different qubit technologies, we have to play by that. Generally speaking, to answer your question briefly, if you have a toolkit like Qiskit, which is open source, you can port it to run on any of the gate- and circuit-based technologies, be it superconducting and ion trap. It's been ported -- well, it runs on superconducting, but it's also been ported on to ion trap as well.

Tony or Ray, any response there?

No. That was great. Complete.

Sorry, guys.

No, no, no. It was -- I think it was a great answer to a profound question.

Well, so let's journey on and drawing on a concern frequently cited in the industry of artificial intelligence and machine learning, the idea of an AI winter, a term coined to describe periods of reduced funding and interest in artificial intelligence research, prompted by inflated expectations and eventual disillusionment when those expectations are not met. So while product road maps are forecasting encouraging advances through the next several years, the journey ahead for the quantum computing industry is long. Across the industry and inclusive of journalists and analysts, how much language and expectations be measured to avoid a quantum winter?

Okay. I'll bite. All right. So with all due respect, James, I almost only hear the term quantum winter used by analysts and the media, okay, not by the people who are actually doing quantum. Oftentimes, people talk about quantum winter specifically as it concerns venture capital investments and start-ups, are you seeing start-ups go out of business in a particular area, right, and therefore, investment dries up. Well, you do have to separate out who the players are and who's well capitalized. I mean, Honeywell, Google, Microsoft, us, Intel, look, we can play this game. We can spend the millions that's required to move this technology along. When it comes to start-ups, a large percentage of start-ups fail anyway. So whatever that percentage is, and you can find different numbers well north of 50%, but pick whatever number you like. If quantum computing start-ups are failing at the same rate as all the other start-ups, well, that's like mismanagement and bad ideas, right, just like other start-ups. So that's not a quantum winter. If quantum start-ups are failing much more frequently than regular start -- other start-ups, then maybe we can say something. I think this all comes down to, while it's fine to describe what you plan to do in the future, so road maps and things like this, you have to show real results. You have to ultimately show the machine. I don't care how many qubits you think you're going to get. I don't care how many gazillion you expect to get in quantum volume, right? Show me the machine. Write that technical paper. Have other scientists and engineers review it. That is real progress. It goes back to my point before, this is not a marketing game. Marketing supplements. Marketing educates. Marketing brings awareness. But it doesn't define it. So if you don't have the real goods, right, don't say you do. So be honest and be realistic and tell people that, look, this is going to take a few years. It makes it harder to sell. I'd rather have something to sell tomorrow, right? But this is where we really are. This is where we expect to be. Don't inflate it.

Tony?

Yes. I think I agree with a lot of what Bob said. This is idea that says this is going to evolve over time, right? So if you want to apply seasonality at the time, that's fabulous -- it could be -- it can either be the spring or summer or fall or winter of quantum. But what's happening is you're seeing across the ecosystem right now continued advancements. And it starts with, does it even work? I mean that was kind of one of the beginning parts. Can you get the system to act as a quantum computer? Can you really do that? Then can you continue to evolve its capability? And as you see that happening with additional qubits being added, with fidelities increasing, with the real entanglement that Bob said coming out and delivering results, that's where not only are the companies that are involved in the ecosystem are starting to get excited, but the end customers that we're dealing with, right? Because you can see -- you can actually plot out -- in fact, it's one of the things that we talked about is this path to value creation. That's not a euphemism. It's a plot. Imagine dots where you're saying, I am going to plot out the capability of the system right now, and I'm going to plot it out again. And what you're really doing is trying harder and harder problems on that quantum computer as you're evolving. And you can start to plot out when this is going to be impacting some of these use cases within these industries, and it's in time frames that are meaningful, so in time frames where companies are starting to not only get excited about it, but say, "Wow, I have to start taking action now if I'm going to put that into my business processes," because that takes time. And so different investment levels, particularly in kind of the -- as Bob mentioned, the VC arena. But you will continue to see these advancements coming from the companies that have put their money against the hardware and continue to evolve it.

Okay. And Ray, I want to make sure that you have a chance to respond to this partially because it's good to get the insight from someone who's not selling quantum computer. Anything you can provide us on this?

Well, with all due respect to Bob and Tony, what they've done a beautiful job of doing is attracting people to quantum computing so that they can start exploring. But the -- a problem will arise. We'll say, 3 to 5 years from now, all sorts of exploration is happening, we will reach a point where, say, 2 or 3 parties can say this algorithm run on a quantum computer that has these properties will be worth money to us. So let's assume that happens because, right now, we do not have that. We assume, and we have every reason to assume that a quantum computer is going to have enormous impact on chemistry, drug discovery, for example. Well, okay, which drug? When will it happen? The problem arises if, at the end of 3 to 5 years, the progress on the hardware platform is not matching the expectations. And so I think that one of the ways to try to adjust language and expectations, I'm trying to answer your questions, is there has to be a lot more work done on algorithms. So getting people from large enterprise companies to try their -- experimenting with baby versions of their workloads on existing quantum computing platforms is great. But it isn't going to have the same kind of impact that you'd have if there are people who are trained in developing algorithms for computers, for heuristics now for -- in place of optimization, to be able to say, "Yes, this exact problem running this exact algorithm on this kind of data is, in fact, going to be a winner." And I think we have to get to that point in 3 to 5 years and hope that Bob and Tony keep delivering hardware improvements that -- so that there is an actual road map and that people believe, these things all get tied together. When we -- when HPE through Cray sells an exascale system, the people who buy them know exactly what algorithms they're going to run, and they know exactly what impact that architecture -- the exascale architecture and the software stack will deliver on those problems. And we are not yet there in quantum. And I think over the next 3 to 5 years, there's a golden opportunity to develop that. But we're going to have to do a better job of identifying the killer apps and identifying the exact algorithms and then putting a road map up for everyone to see that says this algorithm will run on this platform in 3 years, and then you'll avoid a winter.

Okay. Well, so I do have one question left. But I think it's been, for the most part, pretty well answered by everyone so far and -- quantum computing is a specialized field that there are a few thousand researchers globally working in that field. And to draw a parallel to classical computing, it's not strictly necessary for a programmer to know the specifics of every instruction on an x86 processor to be effective in their role. But for the benefit of practitioners rather than researchers, what institutional skills or knowledge would be necessary to leverage quantum computing? And I'll toss it back to Ray real quick.

I think that in 20 years, the younger people who are working on quantum technologies now at quantum computing companies, at universities, are going to be research managers in enterprise-class companies, and they're going to have to make R&D investment decisions based on their understanding of quantum technology. And that is reason enough for any company to start working with quantum computers now because in 10 to 20 years, if I'm right, and even classical IT is -- requires a certain understanding of quantum mechanics, then the time to start building that institutional knowledge is now. And so probably the most important institutional skill or knowledge at this point is a desire to upgrade your understanding of the underlying physics so that -- I've got computer architecture books on my bookshelf here that do talk about not transistors but the limitations that hardware places on what can be done with particular instruction sets. That's all going to be different in 20 years. And the time to start building towards that is right now, which is fantastically interesting. The -- an opportunity to do something other than what we've been doing for the last 50 years, that's pretty exciting.

Bob? Tony?

Yes. Just one word. I think we as -- well, one remark. I think we as an industry are doing very well at starting to seriously educate the so-called C-suite about quantum computing. I think we're making very good progress with developers. We held a summer school, and 4,000 people showed up. 5,000 people registered, but 4,000 people showed up to learn how to code. If you were in a company right now, I will tell you that the group that really has to get on board is between the top and the developers, so the CTO, the solution architects, the application architects. This is the group that now needs to look how to translate between business and technology and technology and business as it relates to quantum computing. I think personally, while many of them have learned about this, they are lagging behind the top level executives and developers. So I think it's time for them to pick up the pace a bit.

Tony?

I agree with Bob. And usually, it's a -- I need help. I don't know what use cases make the most sense for my company. I don't know how this can be translated into a -- into Hamiltonian that can be put into a quantum algorithm. Those are common and absolutely something that people need help to try to understand, both what they should be doing in their organization, but how they should be partnering to be able to see this evolution happen. So I agree with both inputs. Now is the time to start. And that's not just a -- that's not a sales technique. That is -- this is a limited capacity capability. And those who are engaged right now, they do have an advantage. They are able to work and get the kind of attention that is both rapidly escalating their own internal capabilities and how fast they're going to be able to take advantage of these outcomes when they start being generated.

Okay. Well, so for lack of questions and given that we're just now over time, I want to say thank you to Tony Uttley, Bob Sutor and Ray Beausoleil for participating in today's conference. And thank you all for attending.

Thank you. It's a great discussion.

Thank you.

Yes, my pleasure to be here.

Hello again, everyone. Thank you very much for joining us for today's program. That's going to be a wrap on the formal content for the opening day of this year's HCTS. There's still lots more [ hopefully you ] enjoy, including our opening reception. We were honored to welcome NBA legend Kenny Smith, who will be involved in a discussion around what makes [ a champ ], definitely not to be missed, starting at 5 p.m. Eastern Time today. Please also complete the feedback survey for today. This is our first year of virtual, and we'd love to hear what you think about it. Make sure you join us again on Thursday, where we'll pivot our attention to all things data center. Have a great afternoon, and we'll see you at the reception.