speaker-0: Yeah. Yeah. Quantum computing is gonna be a part of supercomputing. Eventually it's gonna be a new accelerator. We've hit a very exciting inflection point in the last year in quantum computing. We don't build our own quantum computer. Just like we're not building our own robot, we don't build our own self-driving car, but we work with every company that does. Over 300 partners total. 48 out of the 50 largest quantum startups. So there's quantum for AI and there's AI for quantum. AI for quantum is already happening right now everywhere. I think we're just scratching the surface.

speaker-1: Is quantum actually overh

speaker-0: Genuine excitement about technology that hasn't realized its potential yet. I have a lot of faith that ⁓ that quantum is gonna reach its potential.

speaker-1: I'm Kim Eisenberg, Superintelligence Editor-in-Chief, and today I'm joined by Sam Sandwig. It's a pleasure to have you here, Sam. Sam leads Nvidia's quantum computing product team where he focuses on how Nvidia's accelerated computing, GPUs, AI, and software tools can help move quantum computing from research toward practical applications, right? Sam, you've built quantum hardware at Regeti.

speaker-0: Pleasure to be here. Thanks, Kim.

speaker-1: worked on control system at Keyside and now lead the quantum product team at Nvidia. How did that path shape the way you think about where Quantum is actually headed?

speaker-0: Great, great question, Kim. Thanks. Thanks for having me again. ⁓ so at at Reggetti, they build the full stack for superconnecting qubits, ⁓ all the way from fabricating the chip to the cloud services. And so for me, coming in early at a a startup like that was really great to get deep in in quantum technology. I was fortunate enough to play a few different roles at Reggetti, ⁓ from Fab to quantum engineering. And so to see everything that goes into building a quantum computer ⁓ and and offering it to to users so so a lot of depth. ⁓ and then Key site was almost the opposite. Key site builds it's a horizontal layer, right? There are control systems that many different types of quantum computers can can use. And you know KeySite's in a position to to work with many companies. And so that was my first, you know exposure to breadth and seeing all the different approaches ⁓ in the in the quantum ecosystem. And you know I I realized that That it takes both. And you know, now I'm at NVIDIA where we're in the fortunate position to get to have both depth and and breadth. So obviously at NVIDIA we build a lot of our own technology and are very deep in the details, but we're not building our own quantum computer. And you know, we're in a position and we do work with ⁓ just about every company that that does. ⁓ so I I'm really grateful for both the depth and and the breadth. ⁓ the other thing I I learned is at Regetium Keysite is. Quantum computing is gonna be a part of supercomputing eventually. It's gonna be a a new accelerator and it needs lots of accelerated computing, a lot of AI, and NVIDIA is in a position to help make companies like Regetti and Key Site and others like them very successful. And so it's a it's a great place to be.

speaker-1: I mean you already obviously mentioned NVIDIA, but let's ⁓ stick to the role NVIDIA has in quantum computing. What is the role Nvidia is like having? And to add the second question right on top, why is Nvidia so involved in quantum computing actually?

speaker-0: NVIDIA is an accelerated computing platform company. And people associate NVIDIA with AI and GPUs and used to associate NVIDIA more with gaming, but the truth is bigger than that. NVIDIA is always investing in new domains of accelerated computing, building the platforms and and tools needed to make them successful, partnering with those ecosystems to help these technologies realize their potential. And so did it with gaming, did it with scientific computing. AI, generative AI, now physical AI and agentic AI and and all these other other flavors. And quantum computing is a part of that journey. And so it it's in NVIDIA's DNA to invest deeply in quantum computing and do it in a way where we're creating value for everybody else in the ecosystem and helping everyone else be successful. So just like we're not building our own robot, but we work with every company that does, just like we don't build our own self driving car, but we work with every company that does, we don't build our own quantum computer, but we Work with every company that does to help make them more successful.

speaker-1: Yeah, exactly. You already answered my following question. You're not building a quantum computer yourself. So but you already mentioned you're helping the industry, so to say, right? So ⁓ in what way are you building to advance the industry? Do you have like some examples?

speaker-0: Yeah. and you know, we're lucky in that quantum in some ways is on the journey that GPU computing went on twenty twenty years ago. Of course there are technical differences, but fundamentally this is a new accelerated computing technology that's not going to replace what came before it. It's going to augment and complement ⁓ CPUs and GPUs, just like GPUs didn't replace CPUs. Right. And you know, what do you need to make that successful? You need a CUDA. You need a programming model that to make that new accelerated computing available broadly to domain scientists. And so we've built a quantum version of that, which we call CUDA Q, which is open source, hardware agnostic, and and now supports the the largest ecosystem of any quantum software. ⁓ you also need pieces around bringing accelerated computing and AI to to help quantum computing succeed, ⁓ both in terms of research. And in terms of of tightly integrating AI. So we're gonna have AI do error correction, we're gonna have AI calibrate our chips. And we're u using lots of accelerated computing and AI to simulate and design our our chips and algorithms before quantum computers are are built. So just about everything except for the the hardware, there's a role for accelerated computing and AI, and so it is a role for us to help.

speaker-1: I mean you already mentioned the ecosystem, right? ⁓ could you like ⁓ explain a little bit more briefly how does NVIDIA work across the quantum ecosystem, like yeah, in the more practical

speaker-0: Absolutely. So we have over 300 partners total, including I think 48 out of the 50 largest quantum startups. Just about every quantum company deploying quantum hardware is integrating it with Q to Q. Just about every simulator is accelerated by our libraries for simulation called Ku Quantum. And we're a platform. And so that means we don't build ⁓ the full stack solution. We build ⁓ pieces and and primitives that accelerate the work everyone else is doing and help them be more successful and and go to market together.

speaker-1: Interesting. Going a little bit more deeper into like the technical ⁓ stuff, what challenges in quantum computing are you working with companies, research centers and others to solve actually? So where is where are you heading? What are the problems you are facing? What are you trying to achieve?

speaker-0: Two big ones are quantum error correction and calibration. ⁓ quantum error correction is how we take ⁓ noisy qubits and turn them into perfect noiseless qubits. It must be done for quantum computing to be useful. And ⁓ calibration is how we tune up a quantum computer. So this is somewhere between a computer and a physics experiment, and so it needs careful tuning. And as quantum computers get get larger, these both become. Very complicated problems and they both become AI problems. And ⁓ what we've done recently is release open models, open AI models to solve both. ⁓ this is a family of models called Icing. There's one for QEC called ISIN decoding. There's one for calibration called called ISIN calibration. And these are integrating into our partners' hardware and and into their solutions. And they are taking these models and fine-tuning them and building agents out of them. ⁓ to help solve the problems of of calibration and and and decoding. Okay.

speaker-1: Okay. ⁓ just l follow up question because I'm super ⁓ interested and and excited in in this topic and just curious, ⁓ what I from your perspective, like the biggest achievements we could face, what are like the advancement, what can we expect with the help of quantum computing in the very near future? So I've heard about material science was some very big topic that is coming up, right? So yeah, just

speaker-0: ⁓ material science, definitely. ⁓ drug discovery. I think applications where you are simulating chemistry, simulating biology, simulating nature, there's potential for quantum computers to accelerate very important parts of those problems and complement GPUs and and AI. I'd also add new types of training data for for AI. And you know I think we've seen with especially physical AI, you know, if we go from Training on the internet to build a chat bot to training on the natural world, we can build kind of AI that practically can do a lot more different things. Right. Now if we train on the quantum world, we train on the world of electrons and photons and atoms and and how they're all how they're all interacting. ⁓ we're excited about if we can build AI to solve problems in that space.

speaker-1: Yeah. Fantastic. Absolute ⁓ excited. So if someone wanted to start exploring quantum computing, how would that work through Nvidia's tour now?

speaker-0: We have ⁓ our platform, CUDAQ, we have a set of educational tools that come along with it called CUDAQ Academic. ⁓ and these are very easy to to get started, easy onboarding, easy to deploy ⁓ notebooks and courses from the basics in quantum computing to including advanced topics like error correction and and quantum algorithms. And so highly recommend checking out CUDAQ Academic. I think more and more people are learning through. AI and and agentic platforms. And so we've we've built skills into our products so that you know if you ask Claude Claude Code or ⁓ or Codex to ⁓ teach you teach you quantum computing and code up quantum algorithms and ask them to do it in QDQ, yeah, you get excellent performance. And that's actually how how I do my learning these days. So ⁓ I should recommend that as well.

speaker-1: Great. I it's it's a missing side of the time we're real living in right now. You can teach yourself so much new stuff with all these tools, right? But ⁓ now let's say you're a quantum computing company. What can you do now, especially with the tools of NVIDIA? So

speaker-0: If you're building quantum hardware, ⁓ I'd say you want and and need GPUs and and AI tightly integrated with that quantum hardware to solve problems in error correction, error mitigation, ⁓ calibration and control. And you know, we're at ISC right now, which is ⁓ you know, there's a huge focus on how to integrate quantum hardware with supercomputing, both from the QPU builder side and and from the supercomputing center side. ⁓ and CUDAQ is is excellent for that and most of the companies here are are already doing that. Yeah. Okay.

speaker-1: And well, obviously NVIDIA is known for AI, especially recently, right? accelerated computing, obviously powerful GPUs, probably the most famous for. How is AI connected to quantum computing? I think you already mentioned it a little bit, right? But especially does one advance the other? I think you already ⁓ agreed on this thesis, so to say, right?

speaker-0: Ab absolutely, but but happy to drill down a bit. ⁓ you know, A AI, so there's quantum for AI and there's AI for quantum, right? And and we're excited about both. You know, quantum for AI is more of a forward looking ⁓ research endeavor where there's a lot of great work on how we can build hybrid algorithms that use a little bit of quantum in addition to AI and and like I mentioned, really excited about using quantum to generate new types of training data to build better AI. So quantum, quantum for AI. AI for quantum is already happening right now everywhere. And I think we're just scratching the surface. And this is at every layer of quantum computing, from how do you design a million qubit quantum computer for the first time? You can't you can't do it by hand anymore. ⁓ fabrication, projecting fabrication to how it's going to look like measured, calibrating, correcting its errors. ⁓ and then a lot of the promising advances in algorithm applications are AI plus quantum algorithms. ⁓ so this is a This is a huge space and I actually think it's it's just getting started. We're just scratching the surface.

speaker-1: I mean you just said we were just getting started. Where is quantum computing today? How do you see its trajectory as a technology maybe in the near future and a little bit more advanced in the further future?

speaker-0: Think we've we've hit a very exciting inflection point in the last year in quantum computing. And that's the transition from physical qubits to logical qubits. And so, you know, we've we've made incredible progress to build systems with a hundred or two hundred physical qubits, ⁓ which are are good for experimentation, but we know for useful problems we need to do lots of error correction and build logical qubits. And what we've seen from many groups, ⁓ large companies and startups, different modalities is they're now doing that beyond breakeven. So they're running error correction and the quantum computer is getting better and and not worse. And so now we're on the path of scaling that. And now you know our our goal is can we take the million X improvement that's come to AI in the past few years and bring that to to a million X improvement in quantum.

speaker-1: That would be fantastic. So probably also accelerating the development of quantum computing. I saw that Nvidia's planning on opening a quantum computing center in the Boston area, right? What is the plan for that in terms of who you'll be working with and the type of research that will be done there?

speaker-0: Right. So the the quantum center is is going to be a center for our research and a blueprint for how to build a quantum supercomputer. And so we're we'll have a GPU supercomputer, ⁓ we'll have different types of of quantum processors, ⁓ and we'll be doing research to inform all of the problems we just talked about. How do we ⁓ do very precise simulations, build digital twins of quantum processors, how do we train AI for for quantum computing? What's the right way to deploy these systems together and and to physically integrate integrate with them? And we'll be doing that, of course, with our partners.

speaker-1: Mm, okay. And you already mentioned that AI, so AI is like a big term, obviously, but so let's stick with the term AI is already helping ⁓ accelerating quantum computing. But what can we learn from previous computing revolutions more generally to accelerate this quantum computing development?

speaker-0: I think it's the one thing we can we can learn going back is they all build on each other. Right. So generative AI could not have happened without building on top of great progress in AI, which could not have happened without building on great progress and and GPU supercomputing, integrating GPUs, which couldn't have happened without many computing revolutions before that. And so when we think about how to make quantum as successful as possible, as as quickly as possible. There's a lot of, you know, how can we build on AI and and GPU supercomputing? And that's the path to to making quantum as as successful as possible.

speaker-1: I mean we already covered several topics, many questions were very greatly answered, but I have to ask this question on top, is quantum actually overhyped? I mean ⁓ when I ask you like one of the leading experts, probably not from your perspective, but straightforward, is it overhyped or not?

speaker-0: I wanna distinguish between hype as statements that are not true and hype as genuine excitement about technology that hasn't realized its potential yet. I think there's lots and lots and lots of gen of genuine excitement about technology that hasn't reached its potential yet. And that's mostly what you're seeing. ⁓ and that's sincere. And are there hard problems to solve? Yes. Are there big open questions? Yes. ⁓ but they won't be solved without a little bit of a belief, ⁓ even if even if irrational and a little bit of you know hard work. And you know, I I take optimism from the f from how far we've come. Right. Right. You know, ten years ago we were talking about one or two qubits in academic labs. Now we have hundred qubit systems with 10 to the minus three error rates deployed in the cloud for anyone to use. And you know the teams that delivered these things have already solved lots of hard problems, have already delivered lots of lots of miracles. And so I have a lot of faith that ⁓ that quantum is gonna reach its potential.

speaker-1: Everyone is nowadays talking about AI, but I would even say the quantum is still a little bit niche, so to say, right? So what should people be paying more attention to in quantum? Why should be they excited about, curious about

speaker-0: Yeah. I think there's been a natural focus in quantum computing on very deep technical details of the hardware. Right. That's mostly what we've talked about in in in this interview. And as quantum computers become valuable, that's gonna invert. Right? When was the last time you heard, ⁓ you know, eight-bit or sixteen bit or you know that like programming and assembly code? ⁓ no, like it's about applications. And I think it's it's as the technology is maturing, it's really important to get more concrete on the applications where quantum is going to add value and how those architectures need to be built to enable those applications. What GPUs do you need? What CPUs do you need? What specs do you need from from the QPU? And use that to start driving roadmaps a little bit. I think that's a really important part of the transition from kind of a Sci a scientific endeavor to a computing and technology endeavor.

speaker-1: Perfect. So we actually already covered all my question. Is there something left you wanna ⁓ let the community know? Something you you wanna highlight, something you wanna emphasize on, something that is maybe even really important for you, something like to to to finalize this interview.

speaker-0: Well, we we didn't talk about Europe in specific at all. And and we're we're at ISC, which is taking place here in in Hamburg, Germany. And ⁓ you know, I I just want to highlight there's been incre Europe has been historical leaders in quantum computing, and there's been incredible progress over the last few years on everything we just talked about. So ⁓ European companies are integrating our our products, QDQ, NVQ Link, ISING they're they're integrating with supercomputers ⁓ here in ⁓ here in Europe. ⁓ and they're they're leading the way a bit. ⁓ so you know, we just had a a US executive order on on quantum computing and and American leadership in in quantum computing. And ⁓ I think we're seeing a great ⁓ you know, investment and desire to work together across America, Europe, ⁓ Asia to help build this technology.

speaker-1: Fantastic. So obviously I'm going to link all the blog posts, all the announcements, all the informations ⁓ you cover from your website ⁓ to this video, to this post. Sam, thank you so much for talking to me. It was such a pleasure. It's so interesting. And have fun. Good luck with the ISC and meeting so many people. So thank you so much, Sam. Such a pleasure. Bye.

speaker-0: Thank you, Kim. It was a pleasure.