Sebastian Hassinger: 00:31

Welcome back to the new quantum era. I'm your host, Sebastian Hassinger. This is the podcast where we interview people working in the field of quantum technologies, quantum computing, quantum sensing, quantum networking, which are very much in the earliest stages of their development. And that's really one of the things that I find the most fascinating and compelling about the topic is the early stage that it's in. I like to say that, you know, it's not like we've reset the clock to the beginning of the Internet era in the nineties, the beginning of the PC era in the eighties, but really, we've reset the clock to something like the late forties or fifties when there was still a lot of uncertainty about how to build computers and how to even build the basic unit of computation, computation, the the the transistor or before that, obviously, the vacuum tube.

Sebastian Hassinger: 01:21

And we have defined a qubit at this point, in quantum computing as a two level system that's controllable and measurable, and we can perform logical operations on it, but not necessarily come to a consensus on how to build a qubit. The ones that dominate the news these days, obviously, are superconducting or atom based qubits, neutral atoms, or ions from various companies. But there are other modalities that have been tried, like NMR, that are still in the process of being developed, like photonics and, spin qubits. And spin qubits is what we're gonna talk about today. These are systems that are built using the traditional fabrication methods from microprocessors, from from classical silicon.

Sebastian Hassinger: 02:12

And they hold the promise for the future of quantum computing in many ways because of the advantages of of leveraging those CMOS and other technologies. So today, the last interview that I conducted at APS, was with Andrew Zurak, who is the founder and CEO of Dirak, a quantum computer company based in Australia that is experimenting with quantum dots. And I learned a ton from Andrew. It was really interesting conversation. So I hope you find it, interesting and educational as well.

Sebastian Hassinger: 03:22

Thank you very much for joining me, Andrew. I'm really looking forward to the conversation.

Andrew Dzurak: 03:26

Me too, Sebastian. Very much looking forward to it.

Sebastian Hassinger: 03:29

So let's start with what's special about Diraq. I mean, you are building qubits, and they're a different type of modality of qubits than I think people are probably more familiar with who have been tracking the industry, right? Superconducting, trapped ions, neutral atoms more recently. There's been rumors of photonic qubits for years, but you're building spin qubits or silicon spin qubits, right?

Andrew Dzurak: 03:55

That's right. Our our technology, the basic concepts go back, I mean, over twenty years. But the first demonstrations of spin qubits are really only about ten to fifteen years ago. And the particular flavor of what we do, which is we modify standard silicon transistors. And inside the transistor structure, we confine or trap single electrons in those transistor devices.

Andrew Dzurak: 04:25

And then we use the property of the electron known as its spin. It's like a tiny magnetic moment. Think of it as a, you know, a compass needle pointing north or south. And we encode the quantum information, the ones and zeros on the direction of that spin. And yeah.

Andrew Dzurak: 04:41

So this technology, you know, we we we've been working on it for for nearly twenty years. So, you know, it's kinda like a overnight success twenty years in the making sort of thing.

Sebastian Hassinger: 04:50

Yeah. Exactly. I think I mean, quantum dots go back to even, like, early eighties, I think, originally. Right?

Andrew Dzurak: 04:57

That's right. I mean, the very first studies of behavior in in you know, at the time, they were tended to be called semiconductor nanostructures. Now these days, all transistors are nanostructures. Right? Right.

Andrew Dzurak: 05:12

Effectively, the term quantum dot is actually a a broader term than the type of we we use a particular type of quantum dot device. Right. The general term quantum dot means any structure that is small enough that the behavior of the electrons inside that structure show quantum like properties in it Right. Due to the confinement. When you make something very small, quantum properties become more apparent.

Andrew Dzurak: 05:40

At At a microscopic level, they're very hard to see. And so you people are probably much more familiar with things like quantum dot technology used in, let's say, display screens and so on. Right? That's that's a different type. So that's where you have a a tiny nanocrystal where, again, the electrons have these quantum properties.

Andrew Dzurak: 06:01

But in that case, you're using the specific energy levels of those states to emit light at particular frequencies or wavelength. That is a optical optically active quantum dot. The type but we we don't use that type of structure. In our case, the quantum dot is just formed by the very small size of the electrode much like the electrode in a transistor. And because that's only a couple of tens of nanometers across, when you have an electron sitting underneath that, you get these very strong quantum properties.

Andrew Dzurak: 06:39

The energy levels are quantized and importantly, the spin is quantized and accessible.

Sebastian Hassinger: 06:46

Right. Right. So essentially, as you said before, it's a single electron transistor, essentially.

Andrew Dzurak: 06:52

Yeah. It it kind of is. Gotta be a little careful with that terminology, though, because there is actually a specific device called a single electron transistor, which annoyingly does not have single electron in it. It actually has can have many electrons, but the but the operational property of that device relies on the tunneling quantum tunneling of single electrons through the device. Got it.

Sebastian Hassinger: 07:19

Got it.

Andrew Dzurak: 07:20

And and to be even more confusing, we use single electron transistors, not as the quantum bits, but as the readout sensors that are observing our quantum dot qubits. So there you go.

Sebastian Hassinger: 07:33

And and the the single electron transistors, I think they're used in scanning electron microscopes. Is that right?

Andrew Dzurak: 07:40

They can be, yes. There are certain types of scanning probe microscopes that do use single electron transistors. And the reason for it, much like the reason we use them, is it turns out that they are the world's most sensitive electrometers. An electrometer is a a device that can measure a very small change in local electric potential.

Sebastian Hassinger: 08:04

Right.

Andrew Dzurak: 08:04

And and a single electron transistor is the most sensitive electron. In fact, it can measure changes as small as a millionth of one electron charge in a device. And and we use that sensor, in fact, in the readout of our single electron quantum bits.

Sebastian Hassinger: 08:21

Do you have you have a practical application of quantum sensing inside your quantum computer?

Andrew Dzurak: 08:26

We do. We do we do billions. We do billions of quantum sensing experiments every day Amazing. On our devices, yes.

Sebastian Hassinger: 08:35

Amazing. So obviously, there's two things that sort of jump out from your introductory explanation. One is that you're clearly you're operating at very, very small scales. These qubits are very small devices, each one individually. Whereas something like a superconducting qubit is using electrons en masse in a relatively large device to be a quasi particle, you're actually acting on the electron spin directly.

Sebastian Hassinger: 09:05

Correct. And the other thing is you're in a position to leverage silicon and leverage CMOS technologies to fabricate these devices. Right?

Andrew Dzurak: 09:17

Correct. In fact, that really is the key reason. I mean, that that's the underlying concept of our company, Dirac. Right. It is to use standard manufacturing both because it is convenient to do manufacturing and it's proven to prove to produce high yield electronic devices.

Andrew Dzurak: 09:39

You can make many millions of those devices reliably. But the other aspect is that they are already very small. So you mentioned size comparison.

Sebastian Hassinger: 09:49

Right.

Andrew Dzurak: 09:49

So we are able to fit millions of these silicon quantum bit devices on a single silicon chip, similar size to the silicon chip in your phone or laptop. Whereas if you take, for example, a superconducting qubit, it tends to be at best tens and typically hundreds of microns across, whereas we are tens of nanometers. And so when you convert that length scale to an a surface area, that's a factor of millions. Right.

Sebastian Hassinger: 10:20

Right.

Andrew Dzurak: 10:20

And so

Sebastian Hassinger: 10:21

But so that's clearly I mean, and as you said, the if you can tap into sixty, seventy years of CMOS technology for fabrication and design for that matter, you clearly can that'll give you a head start in scale and reliability and manufacturability and all of the things that a technology requires to really hit that inflection point and really take off for mass deployment, mass adoption. So you're sort of looking ahead to these devices proving some kind of quantum advantage that the market wants, and you'll be able to supply that demand when that when that time comes.

Andrew Dzurak: 11:00

That's right. I mean, so the bottom bottom line and the reason that we've taken this approach is because while, you know, there's some fantastic efforts out there in trying to find nearer term uses of smaller scale quantum processing units with maybe thousands of qubits, It is generally accepted in in the community that are doing seriously looking at quantum algorithms, you know, real use cases that you need. You're gonna need systems with thousands of logical qubits. Right. At a minimum, hundreds of logical qubits, and that pretty much means you're looking at millions of physical qubits.

Andrew Dzurak: 11:38

Right. And so and then if you need millions of physical qubits, you have to work out how to make them reliably. And you also want to make them small enough that your overall system is economical to build. Right. And so we we've designed now a system that will go to many millions of cubits that can sit inside one single refrigeration unit, pretty much the size of a rack in a data center.

Andrew Dzurak: 12:04

So for just a couple of square meters, we're able to build something that it will be an error corrected, fault tolerant quantum computer with, you know, tens of millions of cubits. So that's really the heart of it. And, you know, the big step for us, because you were talking about the manu CMOS manufacturing, we we had just over ten years ago, did the first demonstration of one of these quantum bits based on a MOSFET transistor, showed it worked, showed we could do one qubit and two qubit logic. But it was all made in a university clean room facility, which is very much you know, it's like a very, well, experimental. And so while, you know, we we made devices there for a decade and it produced, you know, a whole string of papers in nature journals where we showed all of the basic operations and concepts, that is not a scalable pathway.

Andrew Dzurak: 12:57

And so the big step for us over the past two years when we so we only founded the company three years ago and immediately formed very close relationships with foundries both in The U so Global Foundries in The US and IMEC in Europe. And literally just in the last six months have now shown that on full scale 300 millimeter wafer manufacture, which is the standard manufacture in chip foundries, we can make qubits just as well and in fact, better than we were able to make in our experimental environment. And these and these qubits show very high fidelities, all fidelity metrics above 99%, which is a crucial metric for quantum computing.

Sebastian Hassinger: 13:38

That's amazing. You mentioned error correction and fault tolerance. Is there the layout of the qubits? Is this something that you're going to be able to sort of do a native implementation of surface code or are you going to have to come up with your own approach or some sort of modified hybrid approach?

Andrew Dzurak: 13:56

Yep. It's very good question. So I do I mean, this the nice thing about the surface code is that it it works on a two dimensional grid, a two dimensional layout. One of the things that's the strength of our qubits is also one of the challenges because they are so small. Getting in a dense two dimensional grid is is challenging.

Andrew Dzurak: 14:18

And so what we've designed are, let's say, smaller modules. Still, you know, we can still fit many, many millions on a chip, but these are not necessarily fully they're not like a thousand by a thousand. I I can't go into the exact layout because that's proprietary. You can understand. But but to give you a general idea, it's not like a full dense two dimensional grid, but there are, let's say, regions of qubits.

Andrew Dzurak: 14:45

And we actually move electrons around along along these pathways.

Sebastian Hassinger: 14:50

On your website a few days ago. Yep. That how how I mean, when you think of silicon transistors, those are fixed elements on the chip. Yep. How are you moving qubits around on on the chip surface?

Andrew Dzurak: 15:03

Yeah. So the so just to step back and explain how a MOSFET transistor works in a traditional microprocessor. On the top is a piece of metal, so it's called the gate electrode. Underneath it, you have an insulator, which is typically silicon dioxide or another good insulator. And then underneath that is the silicon.

Andrew Dzurak: 15:25

And it's in the silicon where the electrons sit in the transistor. You put a positive voltage on that metal electrode, and underneath the insulator, it draws electrons in the silicon. And that and those electrons carry a current which which is what the transistor you know, it's the current is either turned on or off in the transistor, a one or a zero. In our case, it's we we have our electrons sitting underneath that gate just like in a transistor, but we just have one electron. Or sometimes we group in odd numbers like three.

Andrew Dzurak: 15:54

Right. And then and so in order to move the qubit, we actually just shuttle the electron along the surface of the the lower surface, and we have the gates on the surface that kind of by changing the voltages on those gates Right. They kind of move the along. That's right. Exactly.

Andrew Dzurak: 16:14

They're drag along in something called a bucket brigade type process. And and this this technique is not a new technique. So the idea of moving electrons around in these sort of silicon CMOS devices is used already in what are called charge coupled devices. Right. So it's a standard thing in imaging technology and so on.

Andrew Dzurak: 16:34

So the only thing that we're doing that's a little bit more advanced is when we're moving sing not only we're moving just single electrons, but we're carrying their spin information with them. Right. And and we're carrying that quant quantum encoded information on that spin which moves around the the processor.

Sebastian Hassinger: 16:53

That's really fascinating because it sounds like, you know, one of the for example, neutral atoms as a platform, one of their strengths is that they can with those tweezers, they can entangle qubits and then move them around

Andrew Dzurak: 17:05

That's right. To get

Sebastian Hassinger: 17:06

all to all connectivity. Very plastic environment. Can reconfigure your operating environment according to the next step you need to take. That I didn't realize you could actually do that in in silicon It in qubits as well. That's fantastic.

Andrew Dzurak: 17:19

Absolutely. And in fact, look, there's many similarities. You know, all the different modalities are trying to solve the same problem and we're just using slightly different physical ways to do it. So in the case of both ion traps or neutral atoms, they move either the ion or the atom around, you know, the array in order to bring one cubic closer to the other. We're doing exactly the same way except we move it inside the silicon and we use electrostatic voltages on the gate electrodes to to shuttle those electrons.

Andrew Dzurak: 17:49

And so Right. But the now, we all use slight we might use slightly different architectures. We might be using slightly different error correction protocols. But in the at the end of the day, we're all doing the same thing. Okay?

Sebastian Hassinger: 18:02

Yeah.

Andrew Dzurak: 18:02

And so the differentiation really comes down to are you able to do it in a conveniently manufacturable way and can you get a lot of them in one small chip?

Sebastian Hassinger: 18:15

Right. So okay. So let me take I've got two questions. One is just staying on that shuttling and also including gate operations. What's the speed of those operations like?

Sebastian Hassinger: 18:26

How quickly can you move a qubit from one location to another?

Andrew Dzurak: 18:30

Yeah. Well, actually moving the qubits is one of the fastest things we can do. We we can yeah. So we can move an electron over a significant number of qubit locations in in nanoseconds time scales. Yeah.

Andrew Dzurak: 18:44

So a a few nanoseconds. The the the gate time operation is typically somewhere between a hundred nanoseconds and one microsecond. And so, you know, we we quickly move them around much faster than we need to on the time of doing individual operations. And one one other thing that's important to note and one of the reasons why, you know, we we believe that the silicon integrated approach is so attractive is that everything is essentially done just by pulsing on and off voltages on these electrodes. So it's it's it's extremely analogous to the way that modern microprocessors work.

Andrew Dzurak: 19:26

And Right. And, actually, I do wanna come back to something you mentioned right near the start of our discussion, which was not only can we are we able to leverage the manufacturing technologies from CMOS, but we can also leverage the design technologies. So we've we've just, in the last year, designed a chip that's been made for us by GlobalFoundries. And we designed that chip, has classical transistors monolithically integrated on the same chip with quantum dot qubit devices. Right.

Andrew Dzurak: 20:00

And the whole chip was designed using a standard electronic design application software, Cadence. Cadence, you know, Cadence and Synopsys are the two main chip designs. Yeah. And we we happen to use Cadence for this one. And and that's an enormous advantage because having all of the tools that are already there, not having to reinvent the wheel is incredibly valuable.

Sebastian Hassinger: 20:24

Well, not only that, the the skills to to use that application are out in the marketplace. So you're not starting from somebody who's maybe a PhD in quantum physics, but hasn't actually designed silicon before or used the Cadence tool, so that's great. And then also the the the applications are aligned with the operation of the the the whole tool chain. So you're you're sort of dropping in with a design that I I would assume is largely already compatible with the way the fab is operating that

Andrew Dzurak: 20:55

That's right. That's right.

Sebastian Hassinger: 20:57

Incredible.

Andrew Dzurak: 20:58

And actually, you know, it's interesting you well, significant you mentioned training and experience of the team because one of the things that, you know, we our team has an amazing array of quantum engineers and scientists who are familiar with operating quantum devices. But when we had to devot design these chips, we ended up hiring people from the conventional CMOS chip design industry to help do that design who'd never didn't even know what a qubit was. Right. And so, you know, they're working alongside our qubit team and in in, you know, a tight integration loop. So, yeah, I mean, the the fact is then we're able to leverage not just the toolkit, but also the experience of people who are very, very experienced chip design engineers.

Andrew Dzurak: 21:41

Because many of the ultimate problems of scaling something up to the million qubit level, a lot of it is classical microelectronics. Because a big part of a quantum computer is the classical control and readout circuitry that is required. In fact, it's it's it's it's an equally hard problem to developing the qubit components is developing all of the classical circuitry to control that well. And when you can do that in a tightly integrated way with CMOS, it's a huge advantage.

Sebastian Hassinger: 22:14

I was going say, you were mentioning all of the classical components and elements within the chip architecture. So that's all sitting in the fridge.

Andrew Dzurak: 22:24

Correct.

Sebastian Hassinger: 22:25

You're not using superconducting, so I assume the fridge is there for essentially for reducing environmental noise and isolating the the the cubits from the from the environment.

Andrew Dzurak: 22:35

That that that's right. Technically, the reason for the low temperature is that when you when you've got single electrons in a in a very sort of tightly controlled state, what you don't want is thermal noise that kicks them around and jostles them somewhere they don't wanna be. That's why we gotta cool down. But we superconducting qubits operate at the millikelvin regime, a thousand few thousands of a degree above absolute zero. We operate at the Kelvin regime, so one Kelvin, which, you know, to to kind of human beings who are used to living at 300 Kelvin, the difference the difference of one degree Kelvin doesn't seem very significant.

Andrew Dzurak: 23:24

But if you're an electron, the difference between 20 millikelvin where superconducting qubits are and one Kelvin is a factor of 20. And if you translate that to human terms, the surface of the Earth is 300 Kelvin. If you go 20 times hotter than 300 Kelvin and you're up at 6,000 Kelvin, then you're kind of on the surface of the sun. So actually, this is a huge difference at the level of quantum operation. And so what it means is also at that temperature, we can run conventional CMOS electronics.

Andrew Dzurak: 24:03

Right. And we've actually recently done this with a team that had been working at Microsoft Quantum in Sydney. Right. They've recently a t the team demerged from Microsoft. They basically become their own independent company.

Andrew Dzurak: 24:21

I think they're gonna make an announcement about it soon. They did a bit of a they did a bit of a soft launch at the APS conference this week. But that team developed a chip that can be used to control qubits and we've now used that to control our direct qubits all in one module sitting at the same temperature stage. That's right.

Sebastian Hassinger: 24:44

So you won't have the sort of large room temp electronics racks that that you see around a superconducting fridge.

Andrew Dzurak: 24:51

That's correct. That's correct. Yeah. You know, to be fair, you know, it's important to note that the superconducting community and others are also developing Yeah. Cryogenic electronics because they wanna put some of that stuff at cold temperatures.

Andrew Dzurak: 25:04

But unfortunately, you've got to get from those chips that are sitting at a few Kelvin to millikelvin. That is a real challenge. You then have to have all the wires and you have to be able to thermally isolate. And so, you know, when you see these pictures of the superconducting qubit fridges, looks like a chandelier. Yeah.

Andrew Dzurak: 25:26

All of those cables are to take signals from the milli Kelvin up to a higher temperature stage. Don't you don't escape from that. Right. If in that case but

Sebastian Hassinger: 25:36

So so Yeah. Oh, go ahead.

Andrew Dzurak: 25:38

But but because our qubits can sit at a very similar temperature to the classical electronics, it all can just sit in one module. So

Sebastian Hassinger: 25:46

That's fantastic.

Andrew Dzurak: 25:47

Yeah. So I mean, think about, you know, if you got, you know, your your mobile phone, you break it open, you know, when you after you've it was gone out of gone out of use and you have a look at the board on there. It's got so many chips integrated into a module with many high speed links. Basically, we're looking at our our quantum processor will look very similar to that. It'll be one module with many chips, some quantum, many of them classical, all sitting inside just one refrigeration unit.

Sebastian Hassinger: 26:16

Incredible. So, okay, there's one you mentioned that, you know, you can't lay out the qubits in a 1,000 by 1,000 grid. These smaller cells within the architecture because of the challenges of addressing all of those electrons in such a dense so is that still the connectivity of the small groups of electrons or single the qubits to the control electronics. Is that still sort of a a critical challenge ahead of you? Or or is that sort of a solved problem?

Andrew Dzurak: 26:51

It's it's solved in terms of design. So we have all of the designs. And what we're doing at the moment now with our fabrication partners is doing a series of demonstrator prototypes, gradually increasing the integration and the connectivity. Okay. So there has already been some work done at the R and D level of doing this type of integration at cryogenic temperatures.

Andrew Dzurak: 27:19

We're working with IMEC and others in doing our first demonstrations of this just in the next year and a half.

Sebastian Hassinger: 27:29

Wow. That's fantastic. That so that brings me to my my last topic, which is, of course

Andrew Dzurak: 27:34

Timescale.

Sebastian Hassinger: 27:35

Yeah.

Andrew Dzurak: 27:36

When when when can you when can you buy it and what color will it be? Yeah.

Sebastian Hassinger: 27:39

Or when can I play with it at least?

Andrew Dzurak: 27:41

Yeah. Yeah. So, look, Dirac has taken an approach, actually, which has got some similarities with, say, companies like CyQuantum. You know, one thing that we agree with when it with CyQuantum is that in order to be truly commercially useful, you you need to have system at at serious scale. And and, you know, so we we set that scale at least at of order 100,000 cubits or and preferably in the millions.

Andrew Dzurak: 28:10

So we we decided not to release systems at, let's say, the 100 cubit level, a couple of 100 cubit level. Yeah. We we will make those available, let's say, to specific close partners who who want to gain some experience on them. We might do some bespoke project work for them or or for governments, etcetera. But as a first product release, we're targeting a system with many thousands of cubits.

Andrew Dzurak: 28:39

We haven't we're not publicly stating the number yet. We we we may well maybe I don't know. An year or so might say what that number is. We don't wanna give too much away Right. To our competitors.

Andrew Dzurak: 28:48

But it but our aim is that it will have sufficient number of thousands of qubits to have enough logical qubits to do useful useful problems that are beyond the scale of existing classical supercompute. And we're targeting that in the 2029. So that will be a first product, and that will be a fully integrated product in a single refrigeration unit and with with the error correction controllers. I I should mention actually that the error correction controllers, and this is true for for any quantum computing system, are gonna have to be very high throughput, high speed Yeah. Units.

Andrew Dzurak: 29:30

And those those won't operate at cryogenic temperature. Those will sit in a rack next to the refrigeration unit. But, you know, we believe we can get everything into a single rack, so keep the overall thing quite common compact.

Sebastian Hassinger: 29:43

And that the that external air correction computation, will that be CPUs, GPUs, ASICs, FPGAs? What's the That's what's the horsepower like?

Andrew Dzurak: 29:52

That's a very good question. I can't reveal our current strategy for that because you appreciate that it's a little bit dependent on which particular partners we work with on it. But but what I what I am happy to say is that, you know, we believe that there are a variety of different pathways that that can go. I I using standard GPUs, we think is possible. Custom ASICs, definitely also possible.

Andrew Dzurak: 30:20

The exact blend really is a matter of, let's call it, supply chain development time and cost.

Sebastian Hassinger: 30:26

Of course. Yeah.

Andrew Dzurak: 30:27

But but but but all of the above are possible, and and we're currently speaking with partners on each of those solutions.

Sebastian Hassinger: 30:35

Well, Andrew, this is super interesting. It's really fascinating to me how you've listening to your clearly thinking about industrialization and commercialization. And that's such a clear contrast to sort of the earliest days of qubits, when it was clearly just a science experiment. But we've been sort of gradually moving through this foundational science, basic science, and then sort of getting more and more capability. And listening to you, it's really clear that your bet is on the threshold of industrial value, commercial value, and industrial scale being in the very near future, which I think is super exciting.

Andrew Dzurak: 31:21

Well, Sebastian, I just want to, you know, really just as a final point, emphasize that because our view is that, you know, you're not just gonna have one or two of these gigantic energy hungry quantum computers. That is not a commercially viable pathway for this technology. If we want quantum computing to be ubiquitous and used widely, we actually there are gonna need to be thousands of quantum computers, separate quantum processes that are integrated with high performance computing GPUs and so on. I've actually just I I in fact, I'm sitting at the moment at NVIDIA's g t GTC day. They've just had a quantum day, and a lot of the discussion has been about the integration of AI with quantum using quantum computers to provide essentially the high quality data that goes into the models.

Andrew Dzurak: 32:17

So for example, for drug design and so on. But but for that to work, we need to have a data center that's gonna have thousands of quantum computers.

Sebastian Hassinger: 32:24

That's right.

Andrew Dzurak: 32:25

And it's only gonna be possible if we can make them affordable and compact. And that's and that's been the mission of DIRECT from day one.

Sebastian Hassinger: 32:31

Right. Well, fantastic. I look forward to hearing more as you progress. And thank you so much for joining us. It's been really, really interesting.

Sebastian Hassinger: 32:40

Thank you, Andrew.

Andrew Dzurak: 32:41

Really enjoyed the discussion, Sebastian. Thank you. Bye bye.

Sebastian Hassinger: 32:54

Thank you for listening to another episode of the podcast, a production of the New Quantum Era, hosted by me, Sebastian Hassinger, with theme music by OCH. You can find past episodes on www.newquantumera.com or blue sky at new quantum era dot com. Thanks again to the support from APS for this episode. If you enjoy the podcast, please subscribe and tell your quantum curious friends to give it a listen.