Why IBM quantum computers are colder than space

Welcome to season 2 of the Coherence Times, where we bring coherence to the entangled world of quantum computing. I'm your host, Ryan Mandelbaum. Every other week, I'll bring you stories about how scientists, developers, and businesses are making quantum computing a reality. We'll explore new research and development, highlight the latest advances in the field, and see how users are trying to extend quantum to real-world problems. Superconducting quantum processors are incredibly sensitive. They need to be protected from outside interference and cooled to temperatures near absolute zero, while still being connected to a bunch of control electronics. So, as they continue to scale, the systems that support them have to scale, too. That's why IBM is developing modular cryogenic platforms designed to connect, operate, and scale quantum computing systems. These will be required for the future of fault-tolerant quantum computing, where modular processors will work together to run larger quantum circuits. So, [music] I've invited two guests to help me understand the importance of cryogenics for quantum and how IBM is preparing quantum hardware for this sort of scaling. First, I'm joined by Matt Hollister, IBM Quantum's head of cryogenic systems engineering. Matt has helped to shape the infrastructure that supports increasingly [music] large and capable quantum computing systems, including our new modular platform. [music] Also with us today is Ali Lindler, a cryogenic assembly engineer who works on the design, integration, [music] and operation of IBM Quantum's cryogenic systems. Matt and Ali, thanks for joining [music] me. All right. So, uh Matt, let's start with the basics. For our listeners who maybe don't know that much about superconducting quantum computers or about cryogenics, uh why don't you tell us about why these processors need to be held at such cold temperatures? >> Okay. So, uh superconductivity is inherently a low-temperature phenomena. It uh occurs in a number of different materials, some of which are elemental metals, some compounds. There are some exotic materials that display superconducting properties at relatively high temperatures. There's a class of materials called high-temperature superconductors, but even there it's high temperature in this context is still cryogenic. The devices and the materials that are generally used in superconducting quantum devices are more of the very low-temperature variety. So, we're talking about phenomena that only really appear when you're operating down in the the sort of 4 Kelvin temperature range, which is at around minus 269° C, which is 269° Out with the superconducting quantum devices specifically, quantum phenomena in general is something that's inherently quite unstable. So, to maintain the quantum phenomena that we're actually using in these computing devices, it helps to operate at very low temperatures in order to reduce the phenomena of noise coming in from the environment. Any heat that enters these devices is enough to destroy the the quantum phenomena that we're actually relying on in the processes. That's something though which is true beyond superconducting quantum computing. So, even non-superconducting devices will often benefit from operation at cryogenic temperatures, although maybe not quite such deep cryo as as we generally work with. >> -269° C is that sort of 4 Kelvin range. This is very cold. Can you give me a sense of how cold this really is? >> The typical temperature that people refer to is deep space, which is sort of slightly below that temperature. It's kind of hard to put it into a physical context in terms of what we would experience in in everyday life. Um so that is a very difficult one to answer. However, the the 4 Kelvin temperature where a lot of these phenomena start to turn on, that actually is the temperature of liquid helium. So this is not something that you could have carried around in a constant or something because it is so cold it evaporate very very easily. Um people are probably used to seeing liquid nitrogen as a cryogenic fluid as the sort of the the typical incident demonstrations of people plunging items into into nitrogen and freezing them and causing them to shatter. That is in terms of the temperature scale, that's about 20 times warmer than than the 4 Kelvin temperatures that we're talking about for helium. >> I want to just make sure that the sort of listeners understand there's like two temperatures that we're kind of talking about here, right? We need to first get to the temperature that things are superconducting at all, which is already a very cool temperature. But then we get to sort of this quantum you know, we have to push it even cooler to the sort of millikelvin scale essentially so that not only do we have access to these quantum phenomena, the superconductivity, but also access to actually be able to control it without all of the noise that might ruin the quantum computation, right? >> Yeah, that's that's actually a good way to think of it Ryan. In reality, the refrigeration systems that we tend to use are all really multiple refrigerators working in in in unison. And you can't go from room temperature to these very cold near absolute zero temperatures in a single refrigeration system. So the actual technology we use really is a cascade of several refrigeration technologies to access these very deep cryo temperatures. >> So before we get into the construction of the actual hardware, maybe we can talk a bit more about like what it is that we're trying to protect against in this quantum realm, Ali. Like can you tell us about what it's like to be a quantum processor at these cold temperatures? >> Yeah. Um so to be a quantum processor in the very bottom of your system, you think about an OVC initially. So it's a dark space that your cascading cryostat sits inside. Um the processor is primarily trying to be shielded from things like um thermal radiation and electromagnetic radiation. Um So the ways that we kind of do this are we shield around the processor because electromagnetic and thermal radiation can actually act like light and it can kind of bend around crevices um inside of the computer. And if there's something like a room temperature photon that could kind of travel from room temperature um all the way down to the cold stage where your processor sits. Um there was a colleague that actually described it to me like a lightning strike on your payload, which I felt was pretty neat. And it's very amazing that something um as something as simple as a photon could strike that payload and cause a decoherence of sorts um given that the pulses that are that go inside of the computer are um so much smaller than room temperature photons. >> I see. So if So even if we have this chip held at this negative you know, 273° C basically near absolute zero Kelvin a room temperature photon sneaking in it it really can do some damage to your computation at the very least. >> That's why we're very precise in how we machine things like shields and stuff like that to try and keep those photons out. >> So let's work together to understand what these things are and how they actually work. I think Matt you got us started. Can you tell me about this like cooling in successive stages? What what is this and why does it work the way that it does? >> So, the the story here really starts in the early 1900s. Um around that time there was a lot of interest in driving to lower and lower temperatures. Uh and there really was a race of sorts amongst a number of scientists in Europe and North America to uh liquefy uh gases at progressively lower temperatures. So, things really start to get interesting in about 1908 when helium was first liquefied by uh by Onnes working uh working in Europe. And this was the first time that that that sort of low temperature boundary, the sort of 4 Kelvin boundary, had really been been reached because every other uh liquefied gas uh exists at much higher temperature than that. This was also intimately just uh connected with the discovery of superconductivity itself. Um it turns out that uh mercury superconducts in that temperature range. Uh so, being able to achieve that 4 Kelvin temperature limit also allowed for the discovery of the superconducting phenomenon in the first place using using elemental mercury. Following from that, there were further experiments that sort of pushed the low temperature limit down to about 1 Kelvin. And that's still a temperature that you can achieve using regular off-the-shelf uh helium 4, which is the common isotope of helium. And at about that time, there was some discussion of is it possible to sort of achieve even lower temperatures? It turns out you can't just by using simple liquefied gases. Um but in the early 1900s, around uh the mid-1920s, there was a new class of refrigeration technology was invented which uh relied on um ordering and disordering of magnetic uh domains in uh uh salt materials. And that allowed a a push down a little lower in temperature. And that That the first time that it was really possible to access these millikelvin temperature regimes. The problem with that technology is that it's inherently discontinuous. So, it's great, but you can't operate something for a long period, uh which makes it challenging to do um any sort of computational process or similar where you where you really want to maintain that temperature for very long periods. Um so, it wasn't until uh helium-3, which is a the rarer, lighter isotope of helium was first uh identified in 1939, that it really became possible to achieve these millikelvin temperatures continuously. The base technology that we use is a type of refrigerator called a dilution refrigerator. Uh and that relies on a mixture of the two heat the two primary helium isotopes, helium-3 and helium-4. But, it what it does allow you to do is reach these millikelvin temperatures in a continuous process. Uh so, that was a real breakthrough in in the early to mid-1960s, where you were suddenly for the first time were you able to achieve these millikelvin temperatures continuously. Now, fast forward a little way, uh the dilution refrigerator technology has been used for a lot of different applications in fundamental physics. Uh a lot of them connected to both studying superconductivity, um but also the development of superconducting devices such as the quantum processes that that we use in that we use in our our quantum computing systems. The way that the refrigeration process actually works, um turns out that if you cool a mixture of the two helium isotopes to sufficiently low temperature, uh because of the different quantum mechanical properties of the two molecules the two atoms, um they actually spontaneously uh form uh separate phases. Uh so, a little bit like oil separating from water. Um but an important difference is that unlike oil and water separation, you don't get a complete separation of the two helium isotopes. What you actually have is one phase which is essentially pure helium 3. But then the other phase is mostly helium 4 with a very small amount of helium 3 dissolved in it. Uh and it turns out that if you encourage helium 3 to move across that phase boundary, it produces a cooling effect. And that's what the dilution refrigerator technology is reliant on. Now, as I'd mentioned a moment ago, these systems really are several cooling stages cascaded together. Uh the dilution refrigerator part really only operates below a temperature of around about 1 Kelvin. So, you have to provide some sort of cooling between room temperature and that sort of liquid helium temperature range at a sort of the midpoints of the refrigeration system. Traditionally, this had been provided just with a bath of liquid helium. Um but an important innovation in the late 1990s into the early 2000s um made use of a mechanical refrigeration technology or should say several types of mechanical refrigeration technology that allowed you to reach temperatures around 4 Kelvin without needing that bath of liquid helium. So, really the refrigeration technology that we use now as our workhorse is is a 4 Kelvin refrigerator which is pre-cooling that dilution refrigerator to reach down into the millikelvin regime. >> Got it. And that makes sense cuz it does a couple things, right? Obviously, the most important is it reduces you can't have these lightning bolts photons that I described shooting in there. So, we cool in successive stages. I know there's thermal stresses, things like that. And then I actually wanted to bring back something from the previous season that I thought really described the evaporative cooling here really well, which is it's almost like you're have like the sweating, right? That you're It's literally like a little evaporative process of taking heat from one helium isotope, the helium 3, into the other helium isotope, the helium 4. >> Yeah, it's exactly analogous to what evaporation process. >> And then Ali, can you tell me about the I I I don't think the reader listeners will be able to hear this, but Ali's sitting in a lab with all of these refrigerators right now. And I I think it's really amazing and I want to actually hear from you like walk us through what it looks like, um what it sounds like, and maybe how these fridges are actually built, right? I mean, they're gorgeous chandeliers. Can you get us through a bit of the construction piece? >> Um so, in terms of walking into the data center, you kind of walk in and you hear these pulse tubes. So, these pulse tubes sit on top of your cryostats. Um so, you kind of just walk in, you see a lot of cabinets. Um we have different cabinets for telemetry. Um we have cabinets for gas handling. So, telemetry kind of puts our systems onto network to where clients can access and use them. Um we have cabinets for our room temperature electronics. So, we have cabinets that go um that give communication from your telemetry to your RTE. Um we also have cabinets, like I said, that are for the gas handling system. So, the way that you can operate and maintain a cryostat is through this cabinet that's known as the GHS. Um So, what this has is essentially a bunch of turbos. Um it has a bunch of pumps. Um it has a PLC interface to where you can access valves, pumps, your turbos, um everything else. Um and that GHS has a line, two lines actually. Um you have a still line and a condensing line that run kind of over the ceiling and they kind of go um and attach to the chandelier itself. So, that chandelier hangs in our data center. Um it looks like a cylindrical can, which is your outer vacuum can. Um and the chandelier sits inside. >> Okay, so we have this this chandelier, we have this fridge, this cooling system. And now, Matt, I know that the we need it to be bigger than just one, right? One of these chandeliers, maybe they call the chip that's got like 100 to 200 cubits or so. Um but I think one thing that's really amazing is this discovery of this modularity that we can start to do. Can you tell me a bit about that? Like how what Why do we need to build modular quantum computers and why we need to How do we build a modular fridge? >> So, the the real motivation for going to a modular architecture is we we know we have to build systems with higher cubic counts than we're able to do at the moment. Um so, depending on who you ask, there's numbers of anywhere between a few thousand up to a few million uh individual cubits uh in a useful quantum computer. Um to fit that into uh a single cryostat is is very challenging. Um the while it's possible to build large cryostats, and and here we're talking much larger than the systems that we've historically used up until this point for our deployed quantum systems. Um it doesn't make a lot of sense to try to build a single big system to support uh computers at the scale of what we're talking. It's also very inflexible to take that sort of approach because you you have to design for a particular size system, and if we decided we want to make the system 50% larger, there's a lot of inefficiency in then having to re-engineer everything from from square one in order to accommodate that. So, our approach for these very large future systems is really to adopt modularity at all levels of the stack. The technology processes associated with the production of the quantum chip, of the electronics, as well as the infrastructure such as the cryogenics, naturally lends itself to a to a particular maximum size, and that does not necessarily match to to where one would want to be if you were to build try to build this to the monolithic system. So, really at all levels, everything in uh our scaling plans point towards making the system modular. Um in the cryogenics space, uh in least from the point of view of systems operating at millikelvin temperatures, that's not something that's really been done before. Uh certainly people have experimented with uh with large monolithic cryostats in this sort of temperature range, usually for a specific application such as a specific one-off experiment. Uh but this field where we're talking about wanting to build many copies of our systems to to to couple them together to make this this expandable modular system, this actually is quite a unique and new uh technology space that we're operating in. >> And Ali, how do you actually make a fridge modular? >> Yeah, so we have in the lab right now, um the way that we make these guys modular is we have uh chambers. So, these chambers are kind of customizable to however you want to orient um whatever system or position that you're wanting to create. So, the way that they work is they almost look like a bank vault to where they have doors on four sides, um and then you can take off doors or add as you please or as you continue to build more modular systems, um putting more cryostats inside. So, one cryostat per one chamber. Um You look inside, if you were to open the doors with it fully assembled, you would actually just see a very um shiny MLI blanket and shields. So, MLI actually protects um your processor. Uh reflects a lot of that initial infrared and electromagnetic radiation and things like that coming inside. Um but these modular systems actually have panels. So, for the shields, you have front panels to where if you wanted to do a servicing of sorts, or you wanted to access the wiring inside of the fridge, you can actually just take the front panels off simply and have access to your processor. You have access to your wiring. Um you have access to essentially everything. Um which is much different than your standard cryostat to kind of where it hangs in a frame and it's a cylindrical can. And if you wanted to, you know, access the system, um you'd have to take the OVC cans off and you'd have to take all the cans off to kind of access. Um But the way that it's truly modular is that we have the ability to have as many chambers um that we need for system or how many qubits that we want or that we're looking for. Um so, the systems actually have bridge shields. So, the way that you can kind of connect these systems together is by both connecting them under vacuum. So, both um sides of the chamber have either an O-ring groove or they have an O-ring receiving surface to where you can almost um winch or push them together and apart. Um And then from there, you have bridge shields that allow for um essentially a connection for your processors to be run through and in between. Um the shields actually have a small gap in between, so the shields from one cryostat to the next do not actually touch. Um but they are coated in a dark material that allows for light tightness and having the tunnel, so to say, or bridge um to be able to connect multiple payloads together in multiple uh cryostats, which is thus making it uh modular. >> So, I like to imagine this because I have limited mental capacity. I imagine kitchen refrigerators. And I imagine you have a big one, and then you get increasingly small, and you have a very small kitchen refrigerator in the middle. And that one can miraculously cool your lunch to basically absolute zero. Um so, are you essentially saying that and if it's better to imagine that you were going from top to bottom, and so the innermost one is both in the middle and also at the lowest of all these refrigerators. The freezer's on the bottom, like many for one for kitchen freezers you buy. And so, essentially there's like tunnels that can get connect in these little you can imagine like a whole rows and columns of these stacked or maybe just rows and aisles of these refrigerators in the way you might imagine a superconducting or a supercomputing center with lots of racks. And so, basically you're saying that there's like tunnels between those innermost refrigerators that we have to do a lot of hard stuff in order to get them all connected. >> Yes. Uh well, I wouldn't say hard stuff. Um I would say you have your cryostats assembled, um and it's actually the modularity of it makes it not too difficult to actually put multiple systems together uh with the bridge shields. So, a good way to kind of think about it is, like you said, you have the freezer like as the smallest and the most cold. Um but you have that bridge you have a bridge for each thermal um shield stage. So, you have a bridge at your still, your 4K, and your 50K. And same for the modular cryostat for the system that sits right next to it and next to it and next to it. So, they kind of continue on in a line, um and then you have indoors or end caps on both ends, and you can cap off or close your shields on your ends. Um And if you ever wanted to change that, you could just simply, like a refrigerator, open the door um and make your changes and connect them all together. >> Amazing. It's like going to the quantum grocery store and >> [clears throat] >> break the fridge. >> It really is like a beautiful piece of engineering. It I mean, it's it's incredible. >> Uh and I know that this is not easy. I think when I talk about it in these analogies, it doesn't sound as hard as it as it is. So, I maybe want to talk to Matt about the like the project plan that actually got us here. How do you go about sort of going from building what is Maybe people don't understand that dilution refrigeration has actually been around for so long that we almost think of it as like borderline off-the-shelf technology to now what we're doing is building this big modular crazy fridge designed just for modular quantum computing. So, I'd love to hear a bit about that, Matt. >> Yeah, it's it's possibly a sort of oscillation in the way the technology has developed. Um so, uh as we discussed before, the um dilution refrigeration was first demonstrated in the 1960s. Um but for a very, very long time, these refrigerators only existed in university labs essentially for for fundamental research. Uh part of the issue there was that there wasn't a a major commercial application for the systems. A lot of people would build and operate uh their own their own fridges, and there was a lot of a lot of sort of esoteric knowledge uh that only existed in these uh in in these university labs. Um quantum computing, I think, was the first major technology driver that actually brought the fridges out of the lab and more into the hands of companies that have the impetus to develop a a lot more robust and more industrial type products. And as you say, a lot of that was driven by companies like IBM that were pushing to develop the the superconducting and other quantum computing technologies. Where we are now and traditionally what we've done with our systems today is that we've taken a relative these small off-the-shelf cryostat, which is the sort of classic cylindrical chandelier that that appears in any number of published photographs. And fridges of that sort of scale are capable of, depending on the exact architecture, they're capable of supporting anywhere up to low thousands of qubits. IBM itself has demonstrated over 1,000 qubits operational in one of those stand-alone fridges. The issue then though is to go larger, you have to have a way by which you can couple multiple fridges together. So we we have separately developed the capability to connect the individual quantum processors together. But you still have to get signals from each quantum processor out to room temperature. So that's where a lot of the chandelier hardware that you see really, that's all wiring to control the processor and read out signals. So that becomes the the driver really of how dense you can make these systems. And it turns out that a a good size module, at least for our readout architecture, is slightly larger than the the sort of commercial off-the-shelf fridges that we've been using. Up until now, these systems actually we still we still use. So we're really developing the next technology here. And versus making some single large system. Um we we had experiments a little bit with making larger systems um in the early 2020s, we had a a project called Goldeneye, uh which aims to build a a large monolithic refrigeration platform. Um and while that project was um it was largely successful uh and spurred of development in the uh in industry to to to make larger fridges. Ultimately, we uh determined that that system um the the building block was really too big. Uh there was some ability to do modularity, but but really the individual the individual block was too large. Uh so, following the Goldeneye project, we then started to explore some other options for smaller, higher-density modular uh cryostat configurations. Uh and that um morphed into what became known as the Union project, which is the the modular cryogenic system that we're we're in the process of of developing uh in our lab. Um uh really what Union represents is a uh a building block which is suitable for installation in uh data centers. So, it it's about the size of a large server rack. Um but it has the ability to be coupled in such a way that you're also able to make these these connections at the level of the quantum processor, so that you can string one processor to the next to the next to make these ever-larger systems. >> So, it is going to kind of be like a supermarket of quantum processors, if you imagine the freezer aisle. >> I I I I prefer to think of it as a server rack, but yes. >> [laughter] >> A server rack of quantum processors. I love Anyway. So, Ali, this is a hard project, I assume. And I would love to hear about some of the challenges like when you first were brought into this project and had to see what we were doing and how we're moving from these single fridges and monolithic fridges into this you know, server rack of fridges. What what were what were some of the first challenges that >> you saw? >> Yeah. Um so, we kind of started I joined pretty much when this project um the parts started actually forming and started coming to IBM. Um so, starting out it was initially a lot of site preparation. So, making sure um even little things like thinking about how are we going to take um this cryostat and put it inside of this modular chamber. Like we we need the infrastructure to be able to do things like this. So, it was a matter of getting the infrastructure here on site to be able to do that. Um and then going from actually receiving the first ever um there's a lot of good people and a lot I've had a lot of help here in Poughkeepsie um and in Yorktown as well. Um it's really been a team effort in terms of getting everything here and then from getting everything here to reviewing documentation and kind of figuring out exactly what goes together and how. And not only reviewing documentation to see the assembly process, kind of going through the motions yourself and kind of seeing like hey, maybe we would like to do this this way instead of another way. Um I think um a decent example you're not only kind of taking new parts and building them onto a cryostat, you're also having to um deploy, set up, and maintain an actual physical cryostat that IBM is more used to. Um in a in a certain way, the cryostats are are similar in terms of configuration. Um the modular cryostat has different plate shapes. Um but it's a similar process. Um but it was it was a lot of trial and error. Um and there were sometimes where I kind of had to take what I have learned in terms of thinking about all of these processes and things that can affect the fridge, things that can affect the processor. Um I kind of had to take those into consideration and kind of practically apply them to what I was doing here. And there were times where I would get stuck and having brilliant and smart people, uh great leaders like Matt, who I could feel like, "Hey Matt, um I would like to have your opinion on this." And kind of working with the designers, working with the people that kind of move this project forward, and getting the parts here, and actually giving that final push to have a fully functioning system. Um And having, you know, having to troubleshoot things sometimes because you can have one thing on paper, but it can get here, um and it can be something entirely different, and you have to kind of take the science um fundamentals that you know and kind of practically apply them to build a system that is functional. >> Can you give me an example of that, like the zip ties and dental floss for this project? >> So, not I wouldn't say dental floss here, not yet. Uh I would think a good example would be the shields. So, kind of how I touched on, you know, infrared, um electromagnetic, you know, light tightness in shields. Um there were sometimes where, you know, tolerances aren't always perfect, and vendors, you know, don't always send you the perfect um board through hole or put the perfect pin down in the perfect spot. So, there were some instances kind of where um you would kind of put the shields on and you'd have a problem of hey, there's a gap here and it's not light tight. Or you'd have another issue of, you know, vendor making a part and its protrusion was more than you expected and it's trying to short out to a different stage. Um which that would cause the cryostat. You would most likely not be able to condense the system or cool it down to its base temperature. Um so, things like that we kind of had to, you know, you talk to the designers, talk to Matt, um talk to everybody and kind of see, hey look, I let's have a conversation about this or let's try and figure out a solution. So, like for the shields, um copper tape is a has been a great has been a very good friend of mine um in terms of filling those gaps. So, they're not in places where you can't necessarily make up for them um on a prototype system, um copper tape has been a great resource. Um our Poughkeepsie model shop has also been a wonderful source in terms of we want to modify or cut off this lip here so it doesn't short out to a stage adjacent. So, like the still stage uh shorting out to the 4K or 4K shorting 50 um etc. So, we've kind of had to take some innovative ideas and try them out and, you know, um it's a prototype. So, there's been a lot of learning involved, but there's been a a lot of smart people on the project um that have kind of worked through these problems and tried to improve the system as we go. >> And then maybe Matt, from your point of view looking at this at the project level, I mean, were there any surprises, like lessons learned or things that while you were watching the team put these things together, you maybe had to make some pivoting or or do things differently than you expected? >> There probably were a few instances where uh we we had to adjust what we were doing, but a lot of that was in response to changes in other parts of the technology. Um one of the goals for wanting to do to bring the design of the the modular system actually within IBM rather than outsourcing it to a vendor was that it allows us to execute much faster um because the the field is developing very rapidly. Uh there's a lot of developments that are happening in in other parts of the system such as the processor such as the readout wiring. So having control of the design uh ourselves and being able to respond to the changes in other parts of the system uh really has allowed us to uh to to to have a have a have a system which is much more closely tailored to I think to what we to what we would look for going forward. Um that approach was relatively new for uh IBM Quantum as an entity because previously uh with the use of the off-the-shelf type cryostats, uh you were really designing other components to fit the fit the existing cryostat. Whereas with this project, we've we've taken a much more holistic approach to the overall design uh and the cryostat appro the cryostat design really has been very closely in conjunction with with other parts of the system. >> Maybe to follow up on that, um that kind of brings me to that good point that we're in this era right now where quantum computing is not really a lab thing anymore, right? Is you actual clients and partners are using them to do real things. And so with that being said, how are you designing a dilution refrigerator to kind of meet that needs, right? to be this sort of commercial exterior to the commercial product that lives inside? >> We've taken the approach of um trying to to uh maintain tight control over technologies that are relatively straightforward, but more specialized parts such as the actual cooling system, that we leave with the specialists. So, the core cooling uh system that we use in the uh in these cryostats very much looks like the commercial dilution refrigerators that we've used in previous systems. But, really what we're doing is breaking up the different elements of the cryostat design and optimizing each one separately. So, uh early on in this process, we we went out to several dilution refrigerator manufacturers with a fairly broad technical specification, uh which laid out the key features that we needed in terms of the mechanical footprint, in terms of the refrigeration performance. Uh all of which was based on what we understood of the rest of the the hardware stack at the time. Uh and then we're taking that piece from this vendor and another piece from another vendor and a vacuum chamber for another vendor, and we're integrating everything into a finished product. Uh in the same way that we we would take an off-the-shelf cryogenics system and then integrate our processor and our wiring into that, we've we've kind of uh expanded the scope of what we're integrating in-house. Um we have a lot of experience with uh making these systems robust because, as you said, this isn't something that lives in the lab anymore. Uh but, we uh are aware of the criticality of keeping these systems operating. So, there's a lot of work has gone into and and is still going into making these previously lab-based systems into a much more robust industrial type product Uh that we can deploy either in our own data centers or to customer sites. >> So, I'm hiding this question at the end of this episode and I don't know if I prepared you all for this one, but I know it's something that everybody actually wonders and so I need to know the answer to it and either of you are welcome to answer it, which is that I have I have brought my lunch, which is a frozen a a frozen a hot a frozen burrito or something and I would like to put it into the the quantum dilution refrigerator. What would happen and if it's not possible, why is it not possible? >> Um, I I think in principle it's possible. One one place you might struggle is I'm not sure how compatible with high vacuum systems frozen burritos are. >> That's exactly what I was going to say. I was I was a little bit nervous about the vacuum more than I was the cooling. >> [laughter] >> So, like basically what would happen is as you make it a vacuum, it would explode, right? That's just what would happen? >> Uh, it would yeah, it would it would give off a lot of gas, I think. >> It would really ruin the expensive investment that you've made. >> think it might. I mean, I occasionally you find people find people keeping soda cans in cryostats, but that's generally frowned upon. >> Ah, all right, good to know. Uh, and I think that maybe I'll I'll I'll let's just do the outlook then. I am interested um, more than what the refrigerator will do to my lunch is what the refrigerator will sort of do to the future of quantum computing overall. Um, so maybe we start with Ali and just ask um, what is um, where do you see where does it go from here? I mean, what are you excited about and what do you sort of foresee as somebody working in this space? >> For me, um, I'm a little bit biased towards the chemistry field. Um, I do have a a little background in chemistry. Um, it's my degree in. So, I I look forward to I read a lot of things that IBM has done um in partnerships with different national labs, um their different customers. I think about specifically for me, um being able to model a molecule. Um so, when I say model a molecule, like you could take, let's say you had um X molecule that you needed um to formulate a drug molecule to defend against um just things like that, things that kind of benefit and help us in the world overall. Um kind of seeing quantum computers um almost perform outperform classical entirely. Um And in terms of the hardware, um I think it's going to be really amazing to see um right now looking at, you know, two modular cells and thinking in the next couple years, um that number will multiply greatly. And thinking about, you know, the workforce that IBM will bring in, having a computer with multiple logical qubits is is something that is incredible to think about. Um those are things that, you know, I read about when I was in school. And seeing it kind of come to life while I'm sitting here is is I mean, it's incredible. >> Awesome. And then maybe Matt, can you tell me a bit about the trajectory in terms of the cryogenics? Like, what are we going to have to do in order to realize the actual hardware that cools these fault-tolerant systems? >> I I think where where we are at the moment really is only the first step towards being able to scale to very large systems. Um we've we've demonstrated that the fundamental architecture works. Uh we'll continue to grow uh the the availability and the size of these systems in the near term. Um I think the major next challenge uh partly is related to the uh refrigeration in general, uh because in terms of the overall power consumption of the system, at least at the moment, the cryogenics is is kind of a large contributor to the overall power consumption. Um and there's a lot of interest in driving uh the the wall power of the system down. Um so that will uh that pushes to both increase the efficiency of the the cooling, possibly by adopting different technologies and and pushing the development uh in the in the private sector uh space towards more efficient uh components in the refrigeration stack, uh but also uh it drives us to increase the the density of our systems. Uh the more qubits that we can pack into a single module also will reduce the overall power consumption of the system. Um Out out with that, uh I think there may be a fundamental sea change going forward. Um what we what we're building now is is sort of our our our best based on the technology that we have. Um but we're always looking for developments uh both in fundamental engineering, uh but also in uh the the product space, the availability of different technologies from different vendors. Uh so I think we'll be continuously innovating uh as we push towards the default tolerant systems that are that are on our road map in the next 5 to 10 years. >> Awesome. Well, cool. Thank you for taking us into this the cool world of cryogenics. Uh this was super fun for me. I am uh I I was delighted that we got to experience Ali's sort of in the actual the in the actual space and bring us into the world, even if we can't really see it. Uh and Matt, thank you so much for the history. I I appreciate this one. This was super fun. So uh yeah, thanks again, guys. >> Uh thanks very much for having us on. >> Thank you again. That's it for this episode of the Coherence Times. If you enjoyed the conversation, please be sure to subscribe wherever you get your podcasts, comment in the comments section, and share it with somebody who's curious about quantum. You can find us on Spotify, Apple Podcasts, and YouTube via the IBM Research channel. And for more episodes, resources, and deep dives, please visit us at ibm.com/think/podcasts. I'm Ryan Mandelbaum. Thanks for tuning in, and remember the quantum future isn't just coming, we're building it right now.