Rigetti’s Vision for the Future of Quantum Computing: What Makes This Company So Special?

WE NOTE THAT THE FOLLOWING TRANSCRIPT WAS CREATED BY A ROBOT SO PLEASE FORGIVE ANY TYPOS.

[Tom Zuber]
Tom. Hello everyone. I’m Tom Zuber. I’m the managing partner of Zuber Lawler. We’re a law firm headquartered in Los Angeles with offices in Chicago and New York and various locations in between, it’s a pleasure to speak with you. I’m here today with Subodh Kulkarni, the president, the chief executive officer of Rigetti Computing. Welcome Subodh. It’s a pleasure to speak with you. And do you want to tell us a little bit about yourself and your company before we dive into this exciting conversation?

[Dr. Subodh Kulkarni]
Sure. Thank you. Tom excited to be here, and nice to meet all of you. So I my, I’m support Kulkarni. I joined brigati as a CEO a couple of years ago. Before that, for almost three decades, I’ve been in the semiconductor industry, in either the manufacturing process side of semiconductors or semi cap equipment for semiconductor industries. Before that, did my PhD in semiconductor. So basically, I’m a semiconductor guy who has moved to quantum computing a couple years ago. Regulating computing is at an exciting point in quantum computing. I mean, first of all, quantum computing itself is in very, very exciting phase right now. It’s going to go through an explosive growth potential in the next decade. I mean, for the last couple of decades, quantum computing was still very much in the academic research domain and proving the concepts. We are finally at a point where we are beginning to see the real value of quantum computing and what it can enable, and it’s mind blowing. I mean, the kind of problems we can solve, the computational power benefits we can get, and also energy savings we can potentially get with quantum computing is just incredibly explosive, and that excites us tremendously, given where we are and what we could do in the next decade or so. So, rigidity is one of the leaders in quantum computing. We play in what we call super conducting quantum computing. That’s a particular way of doing quantum computing we along with some other larger companies, like IBM, Google, Amazon, but there’s also smaller companies like us, like iqm quantware and so on. Very exciting space. We are building, essentially a quantum chip, if you will, in our fab, and then cooling it down at really cool temperatures, like 10 milli Kelvin and creating the quantum effect. And that’s how we are generating the computation rigidity has been around for about 10 years, started by Chad Righetti, the founder. He left the company about two or three years ago. That’s when I came in as a CEO. But we will continue to name and we continue the journey of improving our technology to the point where we believe we believe we can start demonstrating real value with quantum computing in a couple years. So right now our focus is basically improve the technology to demonstrate practical value of quantum computing the next two or three years.

[Tom Zuber]
Well, you said a couple of exciting things there, and I want to touch on them. One the energy savings. I haven’t thought about that. What do you? Can you elaborate on that? That’s something I haven’t focused on. Yeah, absolutely.

[Dr. Subodh Kulkarni]
I think people forget that. I mean, everyone gets excited about with quantum computing because of the exponential power of qubits to entangle and the computational benefits we get. And that’s absolutely true, and we should never ignore that. That is one of the main reasons. But what people forget is that the same exponential power that gives us computational power benefits also helps us reduce the energy consumption. So if you think about it, a supercomputer has like 10 million or 20 million CMOS chips. We can do the job of 10 20 million CMOS chips with a single quantum chip, so you consume a lot less power. And if you think about what’s going on right now with GPUs, that’s a big headache right now. I mean, GPUs consume a lot of energy, and you literally are talking about small nuclear reactors next to data centers now, and that’s not a sustainable equation. So I think one of the real benefits of quantum computing is energy savings as much as computational benefits.

[Tom Zuber]
Well, in retrospect, that seems quite obvious, doesn’t it? That really is, yeah, that really is potentially a sensational benefit of quantum computing. And we’re going to get to the commercialization of quantum computers, I heard you say two years, and I want to explore that, because that’s a very exciting timeline. Let’s go ahead and start with some context. Though, if we can, rigidity has been at the forefront of quantum computing innovation. Can you share your vision for the future of quantum computing and the role Righetti plays in shaping that future, just to set a general tone for the for the rest of the conversation. So first,

[Dr. Subodh Kulkarni]
quantum computing as a whole area. I mean, we are literally at the just at the Infancy point here, with the potential explosive growth everyone. I mean, there are many reports that talk about the size of the quantum computing market and in general, the numbers are in several 100 billion dollars roughly a decade from now. But it’s never a linear line. When you talk about businesses emerging businesses like quantum computing, there’s a few inflection points that drive that growth. The first inflection point, we believe, is in about two to three years. We call it narrow quantum advantage, which is when. You can practically demonstrate practical applications and demonstrate that quantum computing is superior in performance compared to classical computing. The next inflection point we’d see is five to six years from now, which is broad quantum advantage, basically the same thing, but with a lot more applications. And then you can start talking not only performance, but can you, can you repeat that word you said, which is, what, in four to six years, broad quantum advantage. Broad quantum advantage. Some people just call it quantum advantage, so that that’s when you can really show that quantum computing is superior than classical computing for most practical applications. And you can take problems that kind of impossible to solve in classical computing. And that really is where the energy discussion we just had over here. That’s when you’ll be able to show that quantum computing actually consumes less energy than classical computing for many, many applications. So two big inflection points, two to three years from now, narrow quantum advantage, and five to six years from now, quantum advantage. Assuming those happen, and we strongly feel they will happen, then the business will indeed be several $100,000,000,000.10 12 years from now. So that’s what the whole quantum computing industry landscape looks like now, within that rigid EV are focused on, as I said, super conducting quantum computing, which is a subset of quantum computing. We believe super conducting is the leading, one of the leading modalities, if not the leading modality. And we’ll get into that, I’m sure, on the pros and cons of different modalities. But within that, then we believe reality is very well positioned compared to the competitors, I mean. And we realize we have some really big companies as competitors, including IBM and Google, and frankly, government of China is investing very heavily in the superconducting area. But we have a lot of IP in this area. We have more than 200 patents. We have been, as I said, we’ve been in this business for about almost a decade now. So a lot of the foundational IP belongs to reg 80. We right now we have a 84 qubit quantum computer that is really good, that is sitting on AWS and Microsoft Azure that anyone can use. So it’s a competitive race. It’s exciting competition going on between us, IBM, Google, we think we are the three leaders in this space. We don’t know exactly where government of China is, but we think they will be in the same league where we are. So we continue to stay focused on being a technology leader. So our goal is to continue to increase the qubit count, increase the fidelity, increase make the gates faster and so on. And that’s the way to get to that nqa milestone took three years from now. Very

[Tom Zuber]
exciting. One of the milestones, of course, to be achieved before broad commercialization of quantum computers is Scaling. Scaling quantum computers remains a major challenge. How is rigetti addressing the technical and practical hurdles associated with increasing quantum cubic count and at the same time maintaining the coherence that makes this power possible?

[Dr. Subodh Kulkarni]
Scaling is one of the areas where we probably are the most differentiated in the superconducting modality side. So if you look at what everyone else is doing, IBM, Google and other companies, basically they’re making the chip larger to increase the qubit comment. So they’re just going for a while, we were doing the same too. But about two or three years ago, we started working on a completely different approach, leveraging the chiplet approach from the CMOS industry, the conventional semiconductor industry. So instead of just making the chip larger, what we are doing now is taking a small chip, or nine cubit chip, and perfecting it. And we are tiring the nine cubit chips like chiplets in the semiconductor industry, and we and then we are using an interposer to tie those chips, and we can create, we can have the entanglement effect across the different chips. We have proven that it works just fine. So our approach in scaling the qubit count is basically tiling more and more of nine cubits. Now we may choose a different number right now, we are staying with nine qubit because we feel it’s a good number to scale up, but it could be a different number. It could be 36 or it could be 100 qubit or something with that. But we firmly believe that chiplet approach is the way to scale up the qubit count without sacrificing performance, and in that sense, we are different than other companies in superconducting and certainly other modalities. So IBM Google, these other companies are not using this idea of chips to achieve scaling of cubic count. They are not using chiplets as of today, now, IBM has made some statements that they are planning on doing that, but as far as we know, they haven’t demonstrated chiplet approach. As of today, we are the only ones who are going this route and demonstrated it. So far. Can you give

[Tom Zuber]
us a size of the scale here? I mean literal size. What how big is a chiplet, and what is that like? So

[Dr. Subodh Kulkarni]
our nine qubit chiplet right now is six millimeter by six millimeter. So if you do the math and say that you want, let’s say fault or quantum computer that we are all looking at about seven, eight years from now, we are talking roughly half a million to a million qubits at that point. And even with the current job. Inventory we have, which is nine qubits in six millimeter by six millimeter, maybe a factor of two reduction, which is not that difficult to do in semiconductor industry. In terms of the density or area density, we can fit in roughly half a million to a million qubits on a panel about a meter square. And that’s like your large panel TV, basically, and we can fit in half a million to a million qubits, and that’s really our vision of how we are going to get to a $4 quantum computer seven or eight years from now. What

[Tom Zuber]
does the rest of a quantum computer like that look like today? And what does it look like in eight years? Not, not the half a million qubits. I’m talking just, just something with nine qubits. What’s the rest of the computer look like. In other words, when we, when we have visions in our head of quantum computers, we picture these huge hydrogens, right? So what does this look like today, a nine qubit, let’s say a nine qubit capacity. And what does the whole computer look like in seven in six or seven years?

[Dr. Subodh Kulkarni]
So if you go to our website, or, frankly, anyone’s websites, IBM or Google’s too. You will see in the superconducting quantum computing something which looks like a cylinder, when it’s actually working. It’s a dilution refrigerator with a control system sitting next to it. So control system is basically CPU, GPU, servers feeding the information and taking the information from the quantum computer. The physical quantum computer is a dilution the refrigerator which has and when you open it, normally it’s in vacuum and very cold, but when you open it, you can see what’s actually going inside. It looks like a chandelier. It’s actually called a chandelier, and it’s gold plated tubes and pipes and wires and essentially there most of it is for cooling the chip down to 10 millikelvin. 10 millikelvin. Okay, that’s really, really cold, and we are pulling it. So most of it is the dilution refrigeration technology to cool the chip down, and then this, of course, there signaling going to the chip and signals coming out. So you will see that when you open the dilution refrigerator. So it looks very esthetically. It looks very, very pleasing actually. And even some Hollywood movies have been made, uh, using a quantum computer as a villain, you know, like to show that quantum computer takes over the world and and what we would do, and that kind of stuff. So it looks very fascinating, whether when you open it and take a look at it, but in real life, it’s sitting in a vacuum. We cool the chip down. So right now, the nine cubit chip, as I said, is just six millimeter by six millimeter. So the whole thing, the cylinder, is roughly the size of your kitchen refrigerator. Okay, so it’s not that huge. And then this control system stack, which is a standard server stack, so it’s, it’s the size of a dining table, effective the whole quantum computer right now, now going forward, when we envision this one meter by one meter to fit in a half a million qubits, or 1 million qubits. Obviously the dilution refrigeration has to be a lot bigger than what it is, so at least the diameter has to be more than your size of the panel. So we are talking about roughly a diameter of 1.5 meters, or maybe your two meters to fit in a one meter by one meter parallel. So it’s going to look a lot bigger cylinder than what it is today, but the control system stack should look comparable. I mean, that standard CPU, GPU stuff to drive it, so we don’t expect that to change that much. So essentially, right now, it looks like a skinny cylinder. It will become a fatter cylinder, but everything else we think is going to be more or less the same.

[Tom Zuber]
You had mentioned superconducting qubits as being as a differentiating factor between rigetti and your competitors. Let’s talk about superconducting qubits. What do they offer in terms of advantages to other technologies like trapped ions or photonics? So

[Dr. Subodh Kulkarni]
sure. So we believe superconducting is the leading modality, primarily because of two factors. One, scaling, because we are using semiconductor chip technology, we can take advantage of the 50 years of semiconductor industry know how, and can increase the qubit count, like the stuff we just talked about. We are leveraging the chiplet approach. We are increasing the density, the aerial density, the way the classic semiconductor industry does. So we have all the benefits that semiconductor industry has learned over the last five decades, and that really helps us to scale up the qubit count. But probably the more important advantage we have in superconducting is speed. We are dealing with electrons just like CPUs and GPUs are today, so we are relying on electron speeds, and so our gate speed effectively is in the 10s of nanoseconds. Okay. When you go to other modalities like trapped ion and pure atoms, you are dealing physically with physically moving ions and atoms, which by definition, are significantly larger than electrons, so the speed that they are at is like hundreds of microseconds. So when you say 10s of nanoseconds to hundreds of microseconds. So basically, we are 10,000 to 50,000 times faster in gait speed than a trapped ion or a pure atom type modality, and that that creates a massive challenge for them when they are trying to do interfacing with CPU and GPU. I mean, the. Speeds are so different between a CPU GPU and a trapped ion or a pure atom modality that they have to create all kinds of buffers and stuff like that. We don’t have that challenge. So that’s why we can talk about hybrid computing, where we can coexist with CPU and GPU, with superconducting QP use, I believe trapped ion and pure atom type modalities will have a tough, really tough time doing hybrid computing because of the speed differential. So two factors, superconducting gives us a scaling advantage because of semiconductor industry leverage, but also the import more important for this speed advantage, having said that, The reason there are companies still investing in trap and other modalities is because they have the fidelity advantage we are dealing with engineered devices. So intrinsically, we have more errors today than their trapped ion or pure atom type modalities are. So right now, our our one qubit gate fidelity is fine. That’s 99.9% already for most of us, us, IBM, Google and so on, but our two qubit gate fidelity, which is where we entangle the qubits and see how accurately they entangle. Collectively, the superconducting camp is between 99 to 99.5% right now, whereas trapped ion pure atoms, they are between 99.5 to 99.8% so they are better in two qubit gate fidelity collectively, than where we are now. That gap is closing. We are already talking like 99 to 99.5 vs 99.5 to 99.8 9.5 to 99.85 years ago, that gap was a lot wider. It’s already closed. Within two years, three years, I believe we will start overlapping with trapped iron and pure atom facilities. All of us will obviously continue to improve our facilities, so we will nudge all of our facilities in superconducting camp. 99 699. Seven. Google just recently published a paper. It’s the first time I saw super conducting modality showing 99.7 now it was a small system and but it’s the first time I have seen 99.7% two qubit gate fidelity in the superconducting side. So I think we are beginning to catch up to draft. I hope you recommend other modalities infinity, but the benefits we have definitely are scaling at speed,

[Tom Zuber]
very good. Let’s talk about quantum cloud services. You’ve been using cloud services. We’re getting cloud services to make quantum computing more accessible. What are some of the use cases, the key use cases applications that you’ve seen from your user base, and how do you envision expanding this approach, this platform? So,

[Dr. Subodh Kulkarni]
good question. I mean, so we have offered our quantum computers on AWS and Azure, both as well as our own cloud. We do see several 1000 customers using our quantum computer routinely on those cloud service platforms. But the usage is not that high. So so the if you look at actual usage of our 84 qubit quantum computer, even including both AWS, Azure, plus our own, we are still less than half utilized. So people are and from what we can see, most of the users that they come from various different vertical markets, financials, education, defense, so they have the customer bases across the entire spectrum. What is, what seems common is they are mostly research oriented customers. They are still primarily quantum physicists or algorithm developers who are trying to understand how to develop algorithms with quantum computer so they understand that these are what we call noisy era quantum computers. We call them NISC, noisy intermediate scale quantum computing in isq. So they understand that so no one’s bringing production workload type applications and testing as compared to a GPU or something like that, and because they know that they will not look good when we do the comparisons. So most of it is research applications, where they are fundamentally trying to understand how a quantum computer works, how they have to define their algorithms, and they are preparing for the future, when we get better fidelity and qubit count, very

[Tom Zuber]
good. And so let’s talk a bit about it’s a related topic. But let’s talk about the hybrid quantum classical systems. I know that rigetti has been a pioneer in this area. It makes a whole lot of sense. I’d like to explain for the audience here. How do you see these systems unlocking near term value for industries before we achieve fault tolerant quantum

[Dr. Subodh Kulkarni]
so I mean, our vision clearly is that hybrid quantum computing is going to happen before full fresh water and quantum computing. We are in the noisy era right now. We are still dealing with fidelity issues and stuff like that, but there are some applications where quantum computing can do a better job than what a CP or GPU can, specifically applications that are probabilistic in nature. So when you’re asking questions like, What are the chances of something happening? What are the chances of an economic recession? A question like Moody’s would ask, what are the chances of that a particular transaction is fraudulent, like a bank, big bank would ask, what are the chances that the weather is going to do this? The Meteorological kind of applications of what. Are the chances that a protein would fold this way or that way? Those would be pharmaceutical type of implications, so almost any probabilistic type application. Fundamentally, quantum computing seems more suitable than classical computing because of the way our qubits work. They entangle. I say that it’s more like a human brain the way your quantum computer works. So just think of how our human brain works with a bunch of neurons that are entangled, and we get analog input, a few neurons act on it, and we get analog output. Quantum computer is very similar to what your human brain is doing. So things that are probabilistic become very difficult for deterministic, things like CPU and GPU. So that’s where, when you look at the algorithms that you have to do with CPU and GPU for these probabilistic applications, they become increasingly complex and very, very intensive. Whereas with quantum computers, with our qubits and simultaneously be able to tackle the problem, those problems become valuable. So our view is that a qpu Quantum processing unit will start coexisting with the CPU and GPU in data centers and some of the probabilistic type applications will drift more to qpu. The standard computational stuff may will continue with CPU, the standard parallel stuff will continue with GPU. That’s the way we believe the ecosystem will evolve, even while we are still dealing with noisy era quantum computing. So we are talking roughly three to five years from now. Okay, that’s what our vision is, that this coexistence will happen in data centers long term, once we reach this fault or in quantum computing, there’s the VC that quantum computers could be standalone systems that are taking the full workload, production workload in a data center. But even then, we never think that a quantum computer will be sitting in an isolation, and there will not be a CPU or GPU in the vicinity. So just like if you look at how the data center has evolved in the last 20 years, I mean, 20 years ago, it was all CPU, now it’s CPU and GPU. Obviously GPUs, the volume, volume of GPUs is still small. It’s just that the value of GPU is much higher. We think qpus will be similar. Volume will be smaller than CPU and GPU, but the value will be much higher. That’s our view of how the coexistence will evolve over the next decade. How

[Tom Zuber]
long until we have those Fuller, not quite standalone, but essentially standalone quantum computers.

[Dr. Subodh Kulkarni]
How long before that happens? Is that a question? Yes, we think it’s at least seven, eight years from now where a fault or a quantum computer with a half a million qubits, that is, that will sit alone. We are dealing with that. And as I said, even with that timeline, our users will always be still in coexistence with some CPU and GPU.

[Tom Zuber]
What we had talked a bit about, accessing quantum computing power today in practical terms limited users, but still in practical terms in a cloud context, and also how these hybrid classical systems can facilitate usage and commercialization today, or at least practical solutions today. You had mentioned some of the areas, the obvious areas, defense and education, where you see your users starting to show up. Can you give us a specific example of a problem that somebody has come to solve, or at least the type of problem that somebody would come to solve, like, what are you what are they asking your quantum computer today to do, or to solve? What problem to solve?

[Dr. Subodh Kulkarni]
Sure. So one example I can pick, and we have published a few papers in this area, is with Moody’s. Moody’s is obviously a big financial organization, and of central banks of the world are asking Moody’s all the time, what are what is the likelihood of an economic recession happening, because that’s such an important input for determining the interest rates and stuff like that. And that’s what Moody’s does for last 100 years. They have a lot of data, classical data, with classical algorithms, and they’re trying to predict economic recession. And the best they can do right now, with all of their prior history is about 77% so they can predict economic recession accurately. Only 77% that means 23% of the times they are wrong, which is a huge problem, obviously. So we have been working with them for two or three years now, and we showed them that you can take their data their classical algorithms, you are just mapping their data with a quantum computer and feeding it back to their algorithms, and we can improve the because even as simple as a nine qubit quantum computer with a simulator, you can improve the number from 77% to 86% okay, that’s huge. And obviously that’s just a nine qubit one. And then we are doing both simulation and actual work. The actual is a little lower, because simulation is perfect by definition. So the actual is at 83% but still 77 goes to 86 with simulation. And the current noisier quantum computer is at 83 so our job as a quantum computing development company is to close that gap between 83 and 86 but the bigger challenge. For us at Moody’s is when we get 100 qubit and 1000 qubit quantum computers with high fidelities. Can we truly bump up that number 86% to 95 or 99% that would be a huge application. I mean, if we can go back to the federal banks of the world, right, the central banks the world, and say, We have improved prediction of economic recession that, I mean, you realize what, what kind of value is associated with those decisions. So that’s just one. Then obviously the whole pharmaceutical area is an obvious one, right? I mean, there are millions and millions of variants of molecule drug molecules, because of the multiple proteins and how they fold and unfold. And right now, all that work is done on classical computing. And practically, what they do is that they narrow the number of potential molecules down to 100 or so, because that’s all they can experiment with and and so they are really taking some they’re taking some guesses as to which direction to go with a quantum computer. You don’t need to potentially worry about narrowing the number down to 100. You can let the simulations run and and let it run for 1000s, if not millions, of variants, and say, this is the way, and that, I think is the most likely chance of success, success. So there are many practical applications where we see classical computing already struggling, where quantum computing showing potential to improve. We are not quite there yet. I’m not ready to say, declare victory and say we can predict decolonization better than what Moody’s can be classical. But we need to continue to work on in these kinds of areas. But I see a lot of potential in these areas, very, very promising.

[Tom Zuber]
Let’s talk about forest. Regetti has contributed substantially to quantum computing programming tools like Forest, an open source tool. How have these tools helped shape the broader quantum development community, and what, what advancements can we expect to tools like this in the future? And also, I think, more to the point, from a commercial standpoint, how does the development of this open source project serve or get a shareholders?

[Dr. Subodh Kulkarni]
All good questions. I mean, fundamentally, we believe we are in the R and D mode right now. We are still very much in developing technology. Yeah, when you’re developing technology, you owe it to the community at large that you share your some of your thoughts on it, as we all understand we are in the capitalistic system. We all want to survive and make money for our shareholders and all that good stuff. It’s all there. But when you are in R and D, first, you have to worry about growing the pie before worrying about how, what part of pie you are going to get. Right now, the pie, by definition, is zero, because we are still in R and D mode. So it’s the onus is on us and other companies like IBM and Google to share things so that we can collectively grow quantum computing community. Once we start growing, then we can fight and compete with each other. But right now we are still in the early stages, so we are doing our part with things like Forest. IBM is doing its part with things like GIS kit and so on. And we all share information in conferences and proper venues. There are things you could legitimately learn from each other, and we try to do that to help the community at large. So yeah, sometimes it becomes contradictory. I mean, sometimes shareholders have a right to challenge us and say, Why are we sharing these things? But I think the my answer is, if the size of the business is 05 years from now, who’s going to win that business? Does it matter? So let’s make sure that there is a viable multi 100 billion dollar business 10 years from now, and do our part to make enable that, and then we’ll try to go a big chunk of that.

[Tom Zuber]
I want to end with a somewhat personal question. I’ll confess you had mentioned the word simulation, and I am a big, nerdy simulation theory guy, and so I’d like to ask you, Are we living in a simulation?

[Dr. Subodh Kulkarni]
This question keeps coming up. I mean, there are some really freaky things that go on with quantum computing that start overlapping with human brain and stuff like that. And you start wondering where we are headed here and where we have come from. And sometimes you do become philosophical about what is our role going on right now? And is this? Is it really? Are we part of some kind of a gigantic simulation or not? But then they get so complex, I finally come back to reality and say, Let’s just focus on the next year and now not take not to ask too many difficult questions that become philosophical in nature. 

[Tom Zuber]
Very good. Subodh, It was an absolute pleasure speaking with you. Thanks for taking the time. Very exciting work at Regetti, and very excited to see where we’re getting goes and where this quantum conversation takes us. Thank you very much for the time. 

[Dr. Subodh Kulkarni]
Well, thank you for your interest. Look forward to hearing from you.