Quantum at Keysight Technologies

Lon Hintze, Clayton Crocker, John Dorighi, and Philip Krantz from Keysight, are interviewed by Yuval Boger. They discuss their comprehensive support for quantum projects, the specific advantages and testing needs of various modalities, the growing significance of benchmarking and AI integration in quantum computing, and much more.

Full transcript

Yuval: Hello Lon, Clayton, Philip, and John. So nice to have you on the program.

Lon Hintze: Great to be here, Yuval. Thanks for having us.

Yuval: Who are you and what do you do?

Lon: My name is Lon Hintze. I’m based in Colorado Springs, Colorado. Been with Keysight 35 years today. Today is my work anniversary, so that’s exciting. And Keysight is very excited about quantum. I’m a quantum solutions lead for the Americas. And anything quantum, whether it’s quantum radar, quantum communications, quantum sensing, quantum materials, quantum computers, Keysight’s interested in helping our customers be successful with whatever they’re doing in quantum. So today we’ll be a little bit more focused on quantum computing, what we talk about. And what Keysight does is we help customers through their entire workflow with products to help them design electronics, especially RF electronics, into the testing phase and the characterization phase and cryogenic calibrations, ongoing installation service maintenance, the whole workflow. Keysight’s here to try and help customers be successful with their quantum projects. So that’s what Keysight’s excited about in the area of quantum. We’ve got some great experts here today that I’ll let them introduce themselves at the appropriate time.

Yuval: Wonderful. So Clayton, who are you and what do you do?

Clayton Crocker: Hi, I’m Clayton Crocker. I’m a planner and customer success engineer here at Keysight. I’ve been here for about five years. And prior to that, I was actually working at University of Maryland. Did my PhD there on trapped-ion quantum computing. So I’m here both to talk a little bit about Keysight as well as sort of represent the trapped-ion platform.

Yuval: And how about you, John?

John Dorighi: Hello. Good morning. John Dorighi and I am a photonic application engineer with Keysight. My role is to work with university research and teaching labs around photonics. So glad to be here.

Yuval: And Philipp?

Philip Krantz: Yes. Hi, my name is Philipp Krantz. Thanks for having us. I’m managing a team of quantum PhDs at Keysight working on quantum customer success. And what that means is that we engage very close with our customer use space working on quantum applications, helping them to be successful with our hardware and software. Prior to Keysight, I’ve been at Keysight for four and a half years now. I joined the company from the acquisition of Labber, which is automation software for quantum. And that was a spin out of MIT where I did my postdoc on superconducting circuits working with Professor Will Oliver, who also joined one of your episodes previously. I also did my PhD here in Sweden where I’m based also working on superconducting qubits. So that’s my background.

Yuval: Lon, congratulations on 35 years at Keysight. You don’t look a day over 40, so you must have started when you were five.

Lon: You’re very kind.

Yuval: So when you think about Keysight activities as Keysight helps vendors, do you see that the majority of work is in the design phase, in the installation, in the ongoing maintenance of a computer? Where is your focus?

Lon: We do try to help people through their entire workflow. And Keysight got started mostly on testing things once you have real prototypes, but through acquisitions and expansion over the years, a pretty significant amount of our business actually comes before having anything, software or hardware to test. And so we do a lot of work in design right now. And then we branched out from just hardware testing to we do a bunch of stuff in software. We actually have more software engineers now than we do hardware engineers inside of Keysight. And then we do have a pretty good business with ongoing services, calibration, maintenance, professional consulting. So we do span the gamut. We got started from Hewlett and Packard days in the hardware test area, but over time we’ve expanded through the entire workflow. So there’s no particular area that we’re over-peaked on. But for most people like even QuEra, we can help in the design of RF components. We can help characterize those components. We can help calibrate and integrate them into a system, do system level software testing, system level hardware testing, and then ongoing installation and deployment of the system to make sure it has a good uptime and good performance throughout its use cycle. So that’s kind of what we do. It is a bit of everything. I wish it was, in some ways I wish it was simpler. But the good news is we can help throughout the entire workflow.

Yuval: And Clayton, you’re the trapped-ion expert. What problems are, or what’s the focus of the testing? What needs to be calibrated or tested in trapped ions?

Clayton: Yeah. So there’s a number of different sides to that. Trapped ions as a platform use a bunch of different kinds of domains of electromagnetic signals. So we can both help kind of on the optics side. We do some work integrating and building the bulk optics or integrated optics for actually delivering light to the ions. Trapped ions as a platform are also pretty heavily involved in the traditional microwaves and RF side. And so in terms of building, testing, characterizing shuttling electronics, drives for acousto-optic modulators or acousto-optic deflectors, we really play kind of across the electromagnetic spectrum for trapped ions.

Yuval: Very good. And John, what do you see as the focus for photonics?

John: I think one of the really exciting aspects about photonic quantum computing is really sort of the scalability and being able to use the standard semiconductor manufacturing processes to integrate what used to be very large physics experiments on optical tables with many mirrors and different optical components have all been shrunk down on wafer. And so I think from Keysight’s perspective, it’s exciting that we can help with a lot of the classic characterization that’s required for the various components that are integrated into the photonic quantum computer.

Yuval: And Philip?

Philip: Yes. So for superconducting qubits, as many of your listeners are aware that there’s a very broad span of what’s going on now in the community of superconducting qubits. And from our perspective at Keysight, we are focusing on both the smaller qubit count systems where you develop maybe material science, you’re looking at new qubit types, that’s one topic that is stretching in a lot of different directions how you design superconducting circuits. So we help a lot with how to evaluate those, how to measure better and better gate fidelities, better readout fidelities, and more on that fundamental side. But then stretching all the way to the larger deployed systems where you have a large qubit count and your engineering aspects are more around synchronization between pulses, how to utilize the hardware in the best way, and essentially how to scale the control electronics in a favorable way. So our challenges, our customers’ challenges are very much focused in different directions depending on the focus of the research.

Yuval: At some level, customers may not care about the particular underlying technology, but they look at the performance of the overall system. Maybe they have access to a cloud service where they could benchmark different systems. So Lon or anyone else, what do you do about benchmarking, how do you help customers address the end result?

Lon: Yeah, I’ll take that one first and my colleagues can jump in too if I miss anything. In more traditional computing, Keysight’s been heavily involved in all of the standards that have merged for traditional computing. So whether it’s DDR or PCI Express or HDMI or USB or you name it, any of the traditional computing standards, Keysight’s been heavily involved in the standards bodies to create those as well as in doing the testing to make sure that everything complies and everything is compatible with each other. It’s still pretty early stages for quantum, so we don’t know exactly how that’s going to go for the quantum industry. We see some things that are emerging as kind of early, I think QEDC calls them prototype benchmarks. So whether it’s two-qubit gate fidelities or maybe it will be something like quantum volume, I think there are some emerging early benchmarks, but there’s really nothing that’s been standardized yet, nothing that’s been said, “Hey, this is the way you must test it.” But over time as the quantum industry matures, as the traditional computing industry has matured, Keysight might end up playing a role in helping to do some of the testing, whether it’s at the system level or at the component level.

Yuval: Anyone else want to augment that?

Clayton: Yeah, I can add a little bit to that actually. So a lot of our recent history in working on benchmarking quantum systems came from an acquisition we made a few years ago of a company called Quantum Benchmark that really had some of the luminaries of the field and well, benchmarking or looking at the performance of quantum systems. This is something that we have integrated into a lot of our products to help us help customers find some of the very traditional benchmarks that Lon mentioned, the quantum volume, the two-qubit gate fidelities, things like that. We found that another really key part of the puzzle is that more than just providing maybe a single number benchmark of how good your quantum system is, really providing some tools to point to where these errors are occurring. So not just a benchmark, but maybe a diagnosis of you’ve got correlated errors here, these qubits are the ones that are performing well, and these are the kinds of things that you can do maybe to mitigate errors throughout your system.

Yuval: How does AI play in this, if at all? I mean, quantum computers are complex systems. When you try to run them on an operational basis, problems may occur in different areas. You rarely want the people who built the system to be just in the data center fixing them all the time. Can you help with the ongoing operation and maintenance and troubleshooting of these systems?

Lon: Yeah, so, again, I’ll jump in first and then welcome my colleagues to help me out where and in areas that I miss. So right now we are working with a number of the prominent AI companies to do stress testing on their AI environments. So we’ll see how big of a workload they can handle. Keysight again through the acquisition of Ixia, we test the entire internet now. Anything that sits on the internet, we test it before it gets on the internet. And we do have the capability to do stress testing on AI components. We do the testing in the data centers. So the short answer, Yuval, to your question is yes, we can help with ongoing operation of data centers to make sure your uptimes are high, constant monitoring and testing of the network. And we realized that in the computing environment, it’s going to be a mix of traditional computing that’s kind of evolving for AI with some quantum computing co-processors, if you will, working side by side with the traditional computers. At least the current thinking of the industry is that’s the way it’s going to work. So again, the hope is that Keysight can help test both sides of the equation, both the AI/traditional side of the computing, as well as the quantum co-processing or augmented processing that’s going to be available through the quantum computing elements.

Yuval: John, you’re the photonic expert. What are the advantages in your mind of the photonic modality for quantum computers?

John: Yeah, I think there’s some really interesting aspects. I mean, fundamentally with photonic quantum computers, we’re encoding the quantum information on the phase and/or polarization of the light. And something that’s unique in photonic quantum computers is that we’re performing the operations on the photons as they travel through fiber or maybe travel through a waveguide on a photonic integrated circuit. So this lends itself to interesting possibilities for quantum networking. But also, if you think about the photons traveling down the waveguides, we’re maybe potentially performing operations at the speed of light. I think that’s a really exciting aspect of photonic quantum computing.

Yuval: And Philip, for superconducting, please.

Philip: Sure. So superconducting qubits have a lot of benefits. I will focus on three of them that I find extra exciting. So the first one is the gate speed of superconducting qubits. And obviously, the end goal is to have millions of qubits. And running very long quantum algorithms will require you to have very long gate sequences in order to execute those algorithms. So the gate operation speed ultimately sets some kind of a limit of the clock speed of the quantum computer. And then you might have some overhead due to error correction and various other pieces also to various degrees, depending on the topology. But at the end of the day, the gate speed, which is around maybe 10 nanoseconds for a single qubit gate on average today, is one strong side for superconducting qubits. The second point I wanted to raise is how customizable superconducting qubits and superconducting circuits in general are, that they are constructed essentially in a design environment. You design your properties of the quantum processor and you can set, for instance, qubit frequencies and coupling rates and also choose how the qubit should be constructed in itself. So it’s very much a creative process of how to design these QPUs, which sets a lot of possibilities for clever engineers and scientists to come up with new flavors of superconducting qubits. And this is something we have seen during the past two decades that this technology has gone in different directions and we managed to mitigate certain noise channels due to the designs. And it’s a very fascinating field of how you can explore these designs. So that flexibility on the qubit parameters is my second point. And the third point which might require a little bit of parenthesis around it is the scalability because that’s something that many platforms have listed as a strength. I do think that superconducting qubits are today considered to be scalable in the sense of how you can design your systems without sacrificing performance of each qubit. And by the help of, for instance, 3D integration and how to make more space efficient components inside of the cryostats, it’s getting more and more scalable, especially since industry has started to get into this game. So scalability, I think, is a third aspect.

Yuval: And Clayton, on trapped ions?

Clayton: Yeah. I’ll also talk about maybe a few advantages of the trapped-ion platform for quantum computing. First would be coherence times. This is a platform that’s really had some record-breaking coherence times, often reaching seconds or in some cases even minutes, which is to say kind of an eternity in quantum terms. This can be really liberating working with the platform, not quite being on the clock to the same degree as a lot of other systems might be. The other real advantage that kind of complements that and is related to that in many ways is the fidelities that you can get out of one-qubit gates, two-qubit gates, or even things like SPAM, the state preparation and measurement operations. These are often just becoming kind of exercises and counting nines for one-qubit gates. Probably like six nines or 99.9999% fidelity. Two-qubit gates, you can get about 99.9% fidelity. SPAM measurements also kind of looking at that three-nine level of sort of some of the world-class experiments. Really excitingly, these all seem to be ways that through hard work you can maintain even as you scale the size of these systems. And the last one I’ll talk about, it’s a little bit special to me, stuff that I did my PhD work on. There’s a really natural photonic interface to trapped-ion qubits. It’s not a unique property, obviously. Other qubit modalities can have this. But I think that also presents kind of a really natural opportunity and kind of picture for scaling by combining these maybe modules of stationary qubits of the trapped ions themselves with photons as flying qubits that can be these nice information carriers to kind of scale up the size of a system between multiple modules.

Yuval: I’m sure it takes many years to become an expert in a certain modality, and it’s sometimes difficult to switch to another one. So if you are a world-class clarinet player, it doesn’t mean that you know how to play the violin. Right? It takes time to switch that modality. So Clayton, if you couldn’t work any longer on trapped ions, which modality would you work on? And then I’ll go to John and Philip with the same question.

Clayton: That’s a really interesting question. I apologize, it might take me a moment to think about an answer for it. I hope you don’t mind if I give kind of two answers. There’s maybe the one that on the one hand, if I couldn’t work on trapped ions, I think it’s really interesting to explore some of the more similar modalities. So things like neutral atoms, they really use a lot of the same technologies as trapped ions. I think that a lot of the advantages I just mentioned for trapped ions are shared by neutral atoms. And they also have their own set of kind of really special capabilities that I’d love to be able to explore and learn a little bit more on. And at the same time, I think it’d be really exciting to, it has been really at Keysight, I’ve had a chance to work with a lot of customers across very different qubit modalities. And so there’s also something really appealing about completely flipping the script and working with something kind of completely different. I think the solid state qubits like say superconducting qubits have been a real pleasure to learn a lot about and try kind of seeing the very different ways that they, different challenges that they face and different advantages that they have.

Yuval: So John, if you couldn’t work on photonics anymore, what would you work on?

John Dorighi: Both Philip and Clayton are the quantum physicists at Keysight. And my role has really always been in the classic characterization of components. I think what’s exciting for me is the need, continued need, even in these quantum photonics circuits to do classic characterization. Whether we’re doing characterization of the ring resonator Q-factor or looking at characterizing the physical layer performance of single-photon detectors. I think that’s very interesting to me and I think that there will always be a need to do classic characterization of components in other areas as well if needed.

Yuval: So you would switch from computing to sensing or clocks or networking so you could still work on photonics.

John: That’s exactly correct.

Yuval: Understood. Philip, how about you if superconducting was no longer an option?

Philip: Yeah. No, again, I think that this is a very interesting question and it’s usually unusual to think a little bit more outside of the box. But I think I would most likely pick maybe trapped ions or I think that there’s many, I think all technologies have their advantages and their challenges. And to me, it’s fascinating to see how engineers, scientists move past these challenges and just the switching, I think, to a different frequency domain, such as working at optical frequencies. I think that that would be a good learning experience for me. So I would probably pick any of the modalities working at optical frequencies really. They’re all interesting.

Yuval: So I have a question for all four of you and maybe we do Philip first and then John, Clayton and Lon. You’ve been doing this for quite a while. What’s new? What have you learned in the last six or 12 months that you didn’t know prior?

Philip: Yeah, I can start. So one thing I really found fascinating during the last six to nine months is how the field has started to get even more multidisciplinary than it was before. So when I started working on superconducting qubits back in 2009, I thought it was multidisciplinary because I found it interesting that nanotechnology was merged with low-temperature physics and there were many different disciplines going into one community. And today the community has broadened even more, including computer scientists and integrators of classical and quantum systems, HPC, FPGA programming. There’re so many elements and so many competencies needed in the quantum ecosystem today that I found that very fascinating and inspiring for the future of quantum.

John: And so I think part of my day in and day out challenges are really driven by customer requirements and working to understand we help our customers make more effective measurements. And it seemed in the last six months, there’s been a lot of engagement helping folks try and characterize our single-photon detectors. So we’ve been looking at we help them understand what the efficiency is, so many photons in, so many electrical signals out, and also look at sort of the physical layer of performance of the jitter of the single-photon detector. So there’s been an interesting area of focus there recently that’s been exciting.

Clayton: I think it’s been really exciting in the last six months or year to see the way that the field’s kind of maturing. And by that, I mean people are really kind of thinking less about how do I do everything myself, putting systems together from nuts and bolts into, all right, I need a… I want to focus on pushing forward this particular aspect of my field, which means I need somebody else to take care of everything up to that point to kind of free me up to focus on what I’m looking to drive forward. So I think that’s been kind of from our perspective, the way we’ve seen it mature. I’ve seen it from the application layer as well, where people are still very interested in the academic side, but there’s a lot more discussion on practical use cases, on things that are outside of just the academic realm.

Lon: I think what I’m enjoying over the last six months, and I think maybe for the perpetual future is it seems like every month there’s a pretty significant milestone that’s achieved somewhere by somebody. And it can be kind of a new level of fidelity that’s been achieved in a certain qubit architecture. It can be the ability to kind of network and communicate between qubit modules. It can be new factories that are opening or expanding. I saw the QuEra announcement, that you guys are doubling your footprint in Boston to build more computers, right? And there’ve been a couple other announcements of new factories being built. New qubit thresholds are being achieved. The biggest, baddest, largest computers being achieved. New end users, new funding rounds are coming in, new applications. Seems like every month there’s at least one really significant milestone and then a bunch of obviously medium and smaller ones as well. But I think that’s what makes quantum and specifically quantum computing for today very exciting is its very regular pace of significant new milestones being achieved. That would be my favorite thing about it.

Yuval: I’m curious because you serve so many different customers in so many different modalities. To what extent is there a collaboration between the superconducting part of Keysight and say the photonics or the trapped ions or other parts? Or are you completely separate and you just show up together on podcasts like this?

Lon: There is a lot of collaboration, right? And a lot of the fundamentals can still overlap. And just as Keysight is also, we do wireless testing, we do wired networking, internet testing, we do electronic vehicles, we do medical, we do satellites. So the industries we span, basically if there’s anything with electronics in it, we probably test it or help design it or help service it. As a company, we’re very used to spanning very wide ranges. And we do have specialists as you’re seeing on the phone that are very focused in certain areas and can go very deep. But we do collaborate across teams. We love learning from each other. I think one of the reasons I’ve loved being with Keysight for 35 years is I feel like I’m constantly learning and constantly getting new challenges because we’re working in so many different areas. But in any case, that’s kind of how Keysight works. A lot of the fundamentals around RF design can be leveraged across those industries I mentioned and leveraged across quantum modalities. Characterizing components, though there are some specifics of interest within each industry, a lot of the fundamentals can be leveraged across the industries and the modalities. So in any case, that’s my response.

Philip: Maybe I can say a few words about this topic because I think it’s also a main reason why what I really like about Keysight and the kind of challenges that we are faced with is this spread of applications that we need to serve. And what’s fascinating about that is that in order for researchers, as Clayton mentioned before, what one core focus for us is to free up time for researchers to be able to focus on their core projects and not having to reinvent the wheel. And one aspect of that is for us to try to push the envelope of what’s possible to achieve with the electronics that we provide, both hardware and software, so that the researchers, academic and industries can continue to push the envelope of what’s possible by the end users. And I think that this is a strength for us to operate in many different fields. It gives us a broad idea of what kind of technology is needed for a certain application. And then we can leverage, as Lon mentioned, we can leverage expertise from different parts of the company to jump on board on projects that require some niche expertise out of our 15,000 employees. And I think that that’s an exciting part that makes us help accelerate in the fields.

Lon: Now, Yuval, I have a question for you, if you’re willing. We see some pretty amazing stuff going on in neutral atom, which we know is kind of your home base there with QuEra. And certainly we see it, but I felt like you might want to also state some things that you’re excited about with the neutral atom modality. And if not, we’re glad to fill in because we’re excited for a bunch of stuff that is going on in that modality as well.

Yuval: Well, thank you for bringing it up. Of course, my day job is chief commercial officer of QuEra, but this podcast is not about QuEra and it’s not sponsored by QuEra. It’s sort of been my personal hobby. But if you were a customer and asking about neutral atoms, I would probably say four things. One is room temperature operation, no need for cryogenic cooling, which of course is a big plus for the data center installation. The other is the all-to-all connectivity because we’re able, and neutral atom vendors are able to shuttle qubits around, then you can have more efficient algorithms. And now when we get to quantum error correction codes, you can have quantum error correction codes that take advantage of this ability to shuttle qubits that cannot be directly implemented on configurations like superconducting that have a static connectivity map. The third thing is scalability. We and others have shown thousands or even 10,000 qubits on a single device, so there’s no need in the immediate term to go to optical interconnects and to network multiple machines. And the fourth maybe should have been the first, that individual atoms are the perfect qubits. There are no manufacturing defects. You can make a billion of these qubits and they’re perfectly identical, and that of course helps reduce error rates and has several other advantages.

Lon: Those are pretty compelling, by the way. That’s why we’re as excited about neutral atom as we are about any of the modalities. Right now it’s so early, it’s hard to tell how it’s all going to evolve and which one’s going to end up performing the best. Maybe there’s multiple for different uses and applications that are all doing well. But yeah, we’re equally excited with you, Yuval, about the potential for neutral atom. The only other one that I’d add, and then my colleagues may have additional ones, but one that I’ve seen is that the density that you can achieve, how many qubits you can get in a small space, because they’re neutral, you can pack them even tighter than the ions, which have some charge to them. So the amount of qubits you can cram into a very small space and then, like you said, do the all-to-all connectivity through the array manipulation, that’s pretty exciting. So that’d be the only one that I would add in for neutral atom. In addition to your list of four, I would add in what I call density for number five. That’s pretty compelling as well.

Yuval: When you hear about a neutral atom computer, your first thought is that this is science fiction. And then you show up in a place like QuEra and say, “Oh, this is just science.”

Yuval: Anyway, I want to finish with a hypothetical to each of you, and maybe we’ll do Clayton and then John and then Lon and then Philip. So if you could have dinner with one of the quantum greats, dead or alive, who would that person be?

Clayton: It’s a great question, Yuval. That is a lot of fun. I’ll admit, one thing I often think about or fantasize about is what if you could talk with some of the real luminaries of the early quantum field today? What kinds of things would really surprise them? So that’s who I’d probably pick, somebody like Schrodinger, Einstein, somebody who has really started this field. It’d be really exciting to hear their perspective on it. It’d be really exciting to hear what they would think about all of the developments that have happened today, things that they could never really have imagined.

John: Yeah. And I think I would even maybe go back a little bit before quantum. I think it’d be fascinating to have dinner with Marconi. I can’t imagine the challenges of trying to transmit radio waves across the Atlantic Ocean. And any time I’m having a hard day, I can imagine my challenges much less than his were. So I think that’s fascinating. And I think that’d be my dinner.

Lon: Those are good answers. I’m going to go a little maybe out of bounds or out there a little bit. I’m reading a lot in the industry literature about how quantum computing kind of emulates the way nature works, the way physics works. And we see many of the articles published in Nature magazine. So if I could talk with anybody, I think I’d want to talk to God because God invented nature. To go, “God, what were you thinking about? How this was going to work? And how did you put it?” And I view God as the ultimate engineer who designed all of this stuff. So I would love to tap into his brain on what he was thinking because I think that’s what we’re doing with quantum is bordering on entering into the divine about how God created nature and life to work. So I realize that’s a very big answer and maybe a little bit out of bounds, but that’s the one I’m going with.

Philip: I agree that this is a very interesting question indeed. And I think that there’s a handful of, if not more people that I would pick, but if I would pick one, I think it would be Nikola Tesla because I think, just telling him everything that his inventions have led up to. And when you have an idea and an invention that leads up to other inventions and how things evolve scientifically, I think it’s fascinating. And I think we will go through many of those steps also through quantum computing and quantum technology as a whole during the upcoming decade or two. So I would pick him.

Lon: Yuval, you’ve got to tell us yours!

Yuval: So I hope I don’t get you in trouble. I would love to have dinner with Bill Hewlett and David Packard. And I’m surprised that none of you said that. I admire their work on multiple levels and I think that would be a fantastically interesting dinner to have.

Yuval: Well, gentlemen, Clayton, John, Philip, and Lon, thank you so much for spending time with me today.

Lon: Thanks for having us so much, Yuval. We really appreciate you doing this.

John: It’s been a pleasure.