Entangled Things
Entangled Things
Episode 148: Gustavo Cancelo on Bridging Classical and Quantum Computing
Use Left/Right to seek, Home/End to jump to start or end. Hold shift to jump forward or backward.
In Episode 148, Gustavo Cancelo, Senior Scientist at Fermilab near Chicago, joins Patrick and Ciprian to talk about QICK, the Quantum Instrumentation Control Kit, an open source platform developed at Fermilab that has become a widely adopted tool for quantum control and readout across four continents. Gustavo walks through what quantum control actually involves, why calibration and characterization are as important as the qubits themselves, and how QICK's MIT license and affordability have lowered the barrier to entry for universities and startups worldwide. The conversation covers QICK's reach into quantum sensing and networking, including dark matter detection experiments, before closing on QICK 2 — a next-generation version with onboard AI engines and low-latency inter-box communication designed with quantum error correction at scale in mind. Large industry partners including Microsoft and HRL have recently adopted QICK, and QBlocks has acquired commercial distribution rights ahead of the IEEE QSC conference in Toronto in September (https://ieeecsc.org/event/conference/2026-ieee-quantum-week)
Hey Tiffrin, how are you doing?
SPEAKER_01Hey, Patrick. Doing well, looking forward for another great episode of Entangle Things.
SPEAKER_04As usual, we're not going to be disappointed. We're joined by Gustavo. Gustavo, do you mind introducing yourself to our audience? I think they'll be intrigued by what you're up to.
SPEAKER_02Yeah, sure. My name is Gustavo Cancelo, and I work at Fermi National Accelerator Laboratory or Fermi Lab near Chicago in Illinois. I've been at the lab for 36 years. So this is an accelerator lab. One of the two biggest in the world. We led the field of the energy field for many years, and now that is at CERN in Switzerland. But we are aiming for the most powerful intensity beam. So Fermi Lab is still the most important high-energy physics, or let's say practical physics accelerator in the country. Okay. So I've been in my 36-year tenure here at Fermi, I've been doing a lot of things, including accelerators and collider experiments, astrophysics, and most recently quantum.
SPEAKER_04Yeah. We're very excited to have you. So I mean, quantum must be, I'd hope, near and dear to your heart as a physicist, because it's letting us get where classical computer has kind of failed us in physics.
SPEAKER_02I like to say that it's not um failing, it's just founding some limits. And quantum computers uh may overcome that problem and will be the machine to be used for certain problems, not for every problem. So classical computer will continue to exist for, I don't know, 99% of the users. And some computers will will uh use quantum a lot.
SPEAKER_04So no quantum spreadsheets or video games, probably any time.
SPEAKER_02Probably not.
SPEAKER_04So um you guys have been up to things that you're trying to enable the environment to develop quantum quicker, as I understand it. Is that true?
SPEAKER_02Well, in in quantum, um, like in most fields, you have a range of areas uh that you need to push for and advance, uh, starting from theory to the applications. Okay. So uh, for example, there are a large community, there is a large community in developing the qubits, the devices that will do the quantum part of the computation. And those uh devices may be of different kinds, uh, could be made out of superconducting materials, uh atoms, um spins, etc. And to in order in order to interact with the classical environment, because we think and live um in a classical way, right? Even when we are made out of quantum material. Uh so that interface that allows the classical world of humans to interact with the quantum devices is what we do. Is what is called uh quantum control and readout.
SPEAKER_04And and you guys have built something that other companies who are trying to bootstrap their way into this field, um, you I as I understand, you're doing the blocking and tackling for them with uh is it called KIC? Q I C K?
SPEAKER_02Yes, uh stands for uh Quantum Instrumentation Control Kit. That's the acronym, KIK. And uh there are there are several companies and um academic labs who have um done similar control systems or or different but uh all intended to be used in Quantum. Um I think that we were lucky in the sense that a lot of the community adopted KIK found uh interesting properties in what we do. So things have evolved very rapidly, and we have now a large community of people using our system in four different continents, not only here in the US and uh North America, but also in in Europe, in Latin America, in uh Asia.
SPEAKER_04Is is it modality independent? Does it matter whether it's superconducting bits or trapped ions or neutral atoms or does that matter really?
SPEAKER_02Yes, it it matters because um the the the way you control superconducting qubits is not exactly the same uh that you control um neutral atoms or spin qubits, for example, or trapped ions. But there are similarities. So you don't need to build one per every kind of um qubit technology. You modify what you have, but you can reuse probably 90% of what you have already done. So uh that that's a good point because uh it requires work to go from one technology to another technology, but it doesn't require to start from the beginning. And that's one of the things that have been attractive uh to people about uh KIC. That's great.
SPEAKER_01And uh people talk a lot about like quantum error correction and about gates and everything, but I think people uh a lot of the time uh maybe don't understand the importance, right? Uh and also the challenges of essentially uh the the controlling part. So uh can you please just for our uh audience like do like a quick overview of of what are the challenges when it comes to to quantum controls, right? Because people understand that the environment in which uh or the environments in which qubits operate are sometimes radically different from the environments that you want to basically use the results of the calculations. There's a difference there, right? And quantum control has to bridge that. So what what would be like or what are the the core challenges here um that one needs to overcome?
SPEAKER_02Yeah, that that's a very good question. So um obviously quantum mechanics has certain rules, right, that um we need to be aware of when we try to control a system, right? Um so typically our quantum control um refers to the preparation of the quantum machine and then reading out the results. Now, when you are developing new qubits, you typically need a lot more from the controls. You need continuous control and operations to characterize how good your qubits are, whether what you build is is successful or not. And characterization and calibration is extremely important and relies a lot on your controls. So, what exactly do I mean with that? Um well, for instance, you have a certain parameters that will give you a performance measurement, a metrics of how good your qubits are. Like, for example, how long do they live? Do they have a certain type of error? Is it prone to certain types of noise? And these are quality measurements, right? That you need to repeat over and over and over again because of the probabilistic nature of the quantum system. And for that, you need to prepare the qubit and do a rotation in the block sphere or uh the Hilbert space. If you have more than uh one qubit, then you may have uh entanglement measurements and and so forth. So for all that, you need controls. And and and what what are those controls? Well, let's focus on superconducting qubits, for example, that is represent 70% of our community. Okay, superconducting qubits uh use RF signals, and those radio frequency signals are typically in the range of 4 to 8 gigahertz of frequencies. And if you want to make a change um in the state of the qubit, you probably do that with an energy pulse. And that energy pulse has a certain frequency and a certain amplitude, and you can um tune up you by only by measuring your your qubit, you know what is the right frequency and what's the right amplitude of that pulse. Okay, so you now you create an experiment in which you um sweep your frequency, sweep your pulse amplitude or time to find the sweet spot in which you calibrate your qubit to get the best result. And then you do a measurement like, for example, T1 or T2, okay, which are um lifetime measurements, right? Um but during the process, you're not only controlling the qubit with that particular pulse, you need to read out the outcome of your experiment. So you send another pulse typically through a different line in which you send another RF frequency and you read out the state of your qubit, which is a projection of the quantum state of the qubit. Okay, so those two are fundamental measurements and sorry, fundamental um actions that you perform in control. Um sometimes these pulses are not radio frequency pulses, sometimes they are just pulses, what we call baseband pulses in DC. Sometimes they are very short, like a few nanoseconds short. And all these operations are sequenced and very time controlled. So you put that into a memory or a processor that sequence an experiment. A typical case is a benchmark test that is called randomized benchmarking. So one or two qubit randomized benchmarking are thousands of operations in a row in a chain. Okay. Now controls may get more sophisticated as well, because you may say, Oh, okay, I want to take um action based upon a readout. So you have a feedback mechanism in your control that is waiting for a readout outcome to take the next action.
SPEAKER_04So this seems like it's a big, big boost-up for universities doing research, trying to spin out a company, startups and entrepreneurs who are trying to do things, countries that haven't gotten to this point. Was that the intent is to be an enabler of the quantum wave that's that's crossing the planet right now?
SPEAKER_02Yes, it has become something like that. Um, but at the at the beginning, we were not thinking that big. Okay. We were thinking um helping few people in one university, one lab. Um the the the the the first uh user for which for whom we developed uh a lot of the things uh is uh Professor uh David Schuster, uh formerly at the University of Chicago and now at Stanford, who's a good friend now, uh, but I didn't know him at that time. And uh he guided us in what he needed to control his qubits. Okay. So uh in those days when we started, uh you can see the pictures seven years ago or so. Uh there were racks of equipment just to control one or few qubits. And you could spend maybe on the order of a million dollars in equipment, maybe more, um, just to control a couple of qubits or maybe one qubit. So we we we we changed that, and now it like you said, I mean, uh kick is affordable for most of the labs around the world. So if you if you have a refrigerator and you have qubits, or you have a um a different type of qubit, uh they're all there, the setups are all always expensive, right? Yeah, uh so the controls have become uh non expensive, not the most expensive part of the of the of your con of your qubit system. And even more, we have versions of Kick for a particular board that we call the 4x2 can be acquired in the same way you can acquire kickboxes. And uh that board is $2,500 for universities, and you can fully control one or two qubits.
SPEAKER_03That's crazy. That's great.
SPEAKER_02Yeah. By the way, Kik is open source under the MIT license. Uh, in particular, I think that is one of the features that has made uh Kik um broadly accepted because other labs are just using it, modifying it, and building on top of it.
SPEAKER_04That's right. That's awesome. Is is there um are there other development projects that you could mention that might be targeting the same kind of enablement, or is this one uh uh just a happy coincidence that you built this thing and it turned out to be is this a part of the a new mission for Fermi Labs or just a side tour?
SPEAKER_02Yeah, but by by no means is uh restricted to qubits and quantum control. Uh in fact, the reason why we started with Kick is because we were using, let's say, not I don't want to say similar, but in some way, similar um controls and readout for detectors. Okay. Um we have done controls and readouts in in my department for accelerators, for um astrophysics experiments and and and many other things. And for several years before KIK, we were working in controlling and reading out superconducting detectors. So that's why there was a small jump into superconducting qubits. Of course, the readout of superconducting detectors is different, but it's also radio frequency, and it's also uh that you read out uh multiple detectors on a same uh radio frequency line. And these particular detectors called uh M kit, microwave kinetic inductance detectors. And uh with the help of Fermilab and University of Santa Barbara, we built two cameras with 20,000 pixels to study uh exoplanets and things like that. One is still operating at Mao Nakea uh in Hawaii in an eight-meter telescope. So, yeah, I mean um then more recently, the instrumentation department in the accelerator at Fermilab uh is using KIK for instrumentation purposes. So, yeah, I mean it's a very general general uh platform.
SPEAKER_04Very versatile.
SPEAKER_01Yeah.
unknownYeah.
SPEAKER_01It looks like it's it's spanning into also like quantum sensing and other stuff, not necessarily only computing.
SPEAKER_02Oh, yeah. And we we use uh Kik now for uh quantum sensing and for quantum networking. There is a quantum network group at Fermilab uh in collaboration with uh Caltech, uh University of Illinois at Wana Champaign, and and other people, and they use Kik for um for that.
SPEAKER_04I had no idea. We've had a few guests talk about it. Is there is there any area of it that you think is uh particularly promising, or are you just supporting it in general?
SPEAKER_02Um yes, I I think that quantum sensing is a whole new chapter. So I don't I don't think I understand the reach of quantum sensing yet, but I can give you uh one example that is uh near to my heart because I've done uh experiments in dark matter detection for many years. So searching for dark matter is is one of the fields quantum sensing can be can be useful.
SPEAKER_04Oh, I hadn't heard that. That's interesting. I guess it would be quantum if you're gonna sense dark matter.
SPEAKER_01Yeah. No, but what's interesting, Patrick, right, is that a lot of people believe that quantum computing actually invented uh a lot of these modalities, right? My my example is always like trapped ions. Like we were using trapped ions in atomic clocks, like way, way before uh we were even thinking about building computers and things like that, right? So there's there's a lot of things that these modalities, or some of those modalities, have been already used, like for I would even dare to say decades. Right? We're just trying to also repurpose some.
SPEAKER_04Well, history repeats itself. If you if you worked with the mainframes back in the early days, then you saw them invent new things that had already been on the mainframe. You know, old schools, new, right? Yeah, of course.
SPEAKER_01So I I'm I'm curious, uh Gustavo, at a like higher or broader level, um speaking about Quantum in general, like computing and networking and sensing. How do you see the development? Are we uh accelerating? Are we stuck? Uh uh is there like, uh especially for computing, like, is there reason for optimism? Uh, because I distinctly remember when Patrick and I started the podcast, which was like more than five years ago, there was still a question of if we can build quantum devices that would do computing and and and other things.
SPEAKER_02Yes. That's an excellent question. So um I'm I'm going to say there is a reason to be optimistic. Um, but at the same time, I was optimistic that I was going to find dark matter one year after I started uh uh with the dark matter experiment. Right? So excellent point. So I'm I'm very optimistic, I I think um many of the things um are leading us to think that a quantum computer is within reach in our lifetime. Now I mean if you if you if you talk to some people like IBM people, they will say we this is already a reality, we are already um outperforming uh classical computing for certain tasks. Okay. Um now there is there is a broad spectrum of optimism. Some people are very optimistic, some are more carefully optimistic, but the the reality is that everything points to a place where we will have a quantum computer, and the question is when and what technology that computer will have. But at the same time, I I I I want to make a parallel with uh what happened with uh uh standard computers or cell phones. Okay, uh it doesn't necessarily mean that the first quantum computer will be based on the same qubits that will um be the optimum qubits 20 years later. So um there is there is a reason to think that all these avenues are possible ways of creating a quantum computer, and that's why people are working in these all different technologies and avenues. And some will get there sooner and some will get there later, but nobody knows which one. Even if you come later, you can you can prevail.
SPEAKER_04And we've said on this podcast might quite a number of times that it's possible there won't be a single modality to rule them all. It'll be in networking, it'll be photonics, in this it'll be trapped ions, over here it'll be superconducting, and they'll be suited for purposes that each one has its own strengths and weaknesses.
SPEAKER_02I I'm not an expert to to to confirm that, but it's it's very possible.
SPEAKER_04Yeah. Yeah. I mean, let me go change a vacuum tube in my computer real quick. I mean, we've we've we've done we've only gone through a couple of modalities in classical computing, but that doesn't mean we're not going to see a lot more variation in in this new thing. Um very, very cool stuff. Is anything else that you want to highlight or that that that we should know about this? Uh I mean, I would almost imagine this might be used in small modular reactors and and and fields that we hadn't even talked about. Is it is does it does it scale to far, far afield, or um is it more in the spaces you've already talked about?
SPEAKER_02I I don't want to speculate what uh applications um these could be used for. I I'm not the person to be an expert in uh the I I listen to talks and I listen to scientists talk about many things, but um we'll see. We I'm I'm more focused. I I can tell you a little bit about the the the roadmap for kick if you are interested. Very interesting. Um okay. So we we are in the in the in the process in the middle of building uh a new kick. Okay, and this is gonna be called kick two. And that version of kick is meant to be thinking about scaling up. Okay, so um you guys mentioned quantum error correction, for example, and there is a there is a new family of FPGAs that we use for kick that allows you, for example, they have internal fabric that's built in for AI engines. Okay, so everybody's now using AI, and you can use AI on top of KIK, the existing KIK, KIK 1. Okay, and people have done that. Uh there are universities and and and companies working putting AI on top of KIK for calibration and optimization, uh optimal control and stuff like that. Well, uh uh KIK2 is going to have AI engines inside the FPGA. So you can you can merge uh AI and uh quantum control into the same FPGA, maybe uh moving some of the AI functions into the chair, into the FPGA chair. Um and this will also allow you to uh have a stack of boxes that communicated with a special communication system that we did for KIK that we call the XCOM. And the XCOM system is very fast, deterministic low latency communication among all the boxes. Within a hundred nanoseconds, you can send a message to another box. So quantum error correction, for example, and even calibration systems require low latency communication. Okay, so that those those two things. I mean, XCOM is available in for KIG 1 as well. Um, but for for KIG 2, I mean if you if you are thinking big, if you're thinking about um having hundreds or a couple thousand channels, um doing quantum error correction algorithms with many qubits, then you may be the person who wants to do it in Kik2.
SPEAKER_04I think that's awesome. I think that's just in time. I mean, we're a year and a half ago, we were talking about the Harvard team with 48 logical qubits in a system through with QRA, and now IBM has just talked about 70 logical qubits. And it's that the goal's moving, but but I think one of the limitations is exactly what you're focused on. So I think that's awesome.
SPEAKER_01And I I I also think this is a brilliant example of what we used to call the the hybrid computing, right? Where it essentially you bring the quantum part to the classical part closer and closer and closer. And what you're describing is actually you are bridging the gap between the two using hardware and FPGAs, which which I think is a brilliant example of how how things might look in in the future, where a classical approach like AI, right, helps uh uh address a quantum-related approach, which is basically quantum control.
SPEAKER_02That that is already happening, right? Um the hybrid computers are already uh working with control systems like Kik to do hybrid uh computing, but it will be even easier with the with the KIK2 version. Uh of course, we are in a very close collaboration with AMD. We could not have done any KIK without uh the help of AMD and those FPGAs. But now the the AMD team is integrated in the development of KIK. I want to say this first because AMD is very important for us. And and they have a competitor which is Nvidia, uh, but uh Nvidia is also uh linking to KIK with the NVQ link. Um so um there are experiments in which AI will be running on uh both the um kick system and outside the kick system to to perform calibration, quantum error correction, and and other things.
unknownYeah.
SPEAKER_02Cool.
SPEAKER_04Well, you need those partners, they're very important, and you know, Fermi's been Fermi and the other US labs have been very good at bringing together the capabilities and the skills to do amazing things.
unknownYeah.
SPEAKER_02Yes. The the other point of interest maybe uh for for you guys is that um now we we are seeing in in the last year, we we are seeing that um larger companies or or large companies, I would say, um are using kick for for their systems. Uh one of them is uh the first one may be uh HRL, Cube's Research Lab. They developed uh uh spin-off of Kik that they call SpinKick because they they use Spin Qubits, right? And also Microsoft uh recently um has adopted Kik for their uh higher and chips. Oh wow. Yeah that that yeah you were a kid when you said big companies. Very closely with um with them. Uh now it turns out that HRL was recently or is still in the process um being purchased by IBM. And IBM has a totally different approach. I mean it the the um we are we are not kick is not um a control system um that will be suited for their amazing uh qubits and and and electronics they have, right? Um but it's interesting now that um HRL is using KIK and now has been purchased by IBM. Interesting. Maybe maybe they will throw it away.
SPEAKER_04Well, hopefully, hopefully they'll adapt it. So so Cyprian and I's background is in the software development space, and I I can speak for him, I think, safely in saying we've built things where people use them in ways we never imagined, and I'm assuming that's that's happening to you and your team as well, that that customers are taking this thing and and applying it in places that you you hadn't thought about. And so you're solving problems you never can countenance. Um we're we're kind of running low on time, we're not out of time yet. But is there anything else you wanted to discuss or highlight or bring up that people should check out? Um, because this is a fascinating topic.
SPEAKER_02Yeah, I mean, uh you are exactly true, I mean I I we we have uh GitHub, Open Quantum Hardware Kick, and you can follow the links from there, and there is a link to read the docs, and then another link from their papers. And there are the last time I counted, there were like 85 papers or 83 papers, and the diversity of uses that people have made um out of using KIK for their explain experiments is quite amazing, uh, in including biological qubits based on proteins and and things like that.
SPEAKER_04Really? Oh, that's fascinating. We need a different podcast for that, I think.
SPEAKER_02Well, I can tell you the the expert uh on that subject who did the the work. So yeah, I mean it's uh it's a good place to go and and and spend uh a few hours looking into the information.
SPEAKER_04So Gustavo, uh you is there a place or an event that you're gonna be speaking at anytime soon that people could see you at?
SPEAKER_02Yes. Um there is the IEEE uh QCE 26 uh in Toronto. Um that's in September, I believe, right? Yeah, starting September 12th or 13th. And so I'm gonna be one of the keynote speakers in the QBlocks um workshop. Uh Q QBlocks has acquired the right to commercialize and distribute Kik. And we have finalized an international creator with the company, so they they will uh be helping um enlarging the the number of users, the user base for for for kick. So I'm participating of that, and then we will have our annual um tutorial uh led by Sarah Sussman, who's a postdoc uh at Fermilab, uh Show Wemura, and uh Horacio Arnaldi, part of the KIK team, and I will collaborate with that. So yeah, there are plenty of kick activities in in QC26.
SPEAKER_04Awesome. We'll put that in the notes as well for uh for people to see. Hopefully, they get a chance to see you.
SPEAKER_02Thank you. Yeah.
SPEAKER_04Um well uh we I think we'll leave it on that. That's a great note. We really do appreciate your time. We appreciate Fermi Labs for allowing us to come and talk to us, and we hope uh we can talk to you guys again because you're doing amazing stuff. Uh thanks again, and we'll see everybody next time.
SPEAKER_02Thanks, everyone. Goodbye. Bye. Thank you. Bye-bye.
SPEAKER_00Cybercrime is one of the biggest threats to businesses of all sizes and industries. With almost half a million open cyber positions, the problem is compounded by the lack of available talent in the marketplace. At Pulsar Security, our elite team of highly credentialed experts collaborate with you to assess your current defenses and develop solutions tailored to your specific needs. With services ranging from cybersecurity education to advanced penetration testing and red teaming, you can start reducing your risks today. Visit PulsarSecurity.com and let's secure your digital future together.