Quantum Leap The Next Frontier of AI Computing | theCUBE + NYSE Wired | DigiCert CEO Amit Sinha

Welcome back onto the cube here in the Palto studios. I'm John Furry, host of the cube with Dave Vol and our entire cube team. Special Sunday edition of pregame GTC preview. We had six expert panels. We're on our sixth. Going to wrap up then head to San Jose to kick off GTC. This panel is the quantum leap, the next frontier of AI computing. We have a lot of cube alumni here been on before. Adam, head of engineering at Sandbox AQ. You your first time, but we had you Jack on Prenav, CTO, co-founder inflection, Anuj with C chief product GTM of Foranix and Amit CEO of DigiCert, many time cube alumni. Gentlemen, quantum it is the hottest topic. I wouldn't say that the mainstream is talking about, but you're starting to see mainstream media pick up quantum. Certainly a lot of the heavyweight tech vendors are starting to flex quantum in their messaging. At Mobile World Congress or MWC, it probably was the biggest quantum washing I've seen people putting quantum on their booth but weren't showing any quantum, right? But certainly AI was hot in the telco era. So quantum seems to be breaking through this year. I mean we've you've been on two years in a row you had prep quantum preparedness and then quantum readiness. Yeah? Where are we? Is it breaking through finally to the mainstream?

Well, we're certainly very appreciative of that because you're the one who kind of tipped our hand to start digging into the quantum. And I think now what gets me excited is you're hearing it on the main stage in Davos, for instance. Quantum was huge. You're starting to see real engineering. Certainly you highlighted the alpha geeks getting in the networks like Cloudflare of the world because they know the concept. But now you start to see it more of a product strategy value creation piece of it. Chips being made, systems being worked on, Nvidia endorsing it in a meaningful way, not in kind of a haymaker way. CES, Jensen you guys saw Jensen kind of made a comment off the cuff about stock prices, the whole market crash because Jensen said 'why is the stock so high?' But then he actually intentionally walked that back in a proactive way and said, 'No, what I meant was the stock prices because we're using quantum.' Then he unveiled that Nvidia is aggressively behind quantum because the GPUs can be used on things like error checking, things that are helpful for engineers talking about this. So you're seeing the engineering piece on the product development side, the infrastructure there. Where are we on the product? So who wants to take a shot? Let's go there first.

I'll be happy to take that on. And I mean it's been an amazing transition in the last 5 years in quantum being kind of the back shelf future technology to now being very much on the show floor for GTC, for Mobile World Congress, for Davos, etc. I would be remiss not to show maybe a little prop, show the audience what the heart of a quantum computer can look like.

That's cool.

Indeed. This is cool. And this is Inflection's neutral atom core. So we use this technology to build quantum computers and quantum sensors. Inside of this core, we get individual atoms to be extremely still and those become cubits or quantum bits. We have the world record for in a commercial system number of cubits: 1600. And what is fun is that tomorrow, this is the night before GTC starts, we will have our quantum computer on site. We've announced this at the Nvidia booth. And it goes to show how Nvidia is thinking about GPUs and QPUs playing together to unlock new frontiers of compute. One of those frontiers maybe especially relevant to this audience here is quantum computers do have an implication on security. We have shown on a system recently the first demonstration of decryption using a quantum computer with error detected logical cubits. This is the real frontier for how eventually quantum computers will render existing crypto systems broken. But there's also this frontier of how quantum computers help in terms of accelerating AI, the same way that GPU stood on top of CPU to create AI as we know it today. We think there's a frontier of physical AI and world models that are going to incorporate quantum mechanics in their building blocks, and also ways that quantum computers can extend the power of existing models like large language models with better memory, better context windows, and overcome some of the existing scaling laws. So I think it's an important transition.

Well, there's two things I want to just double click on. One is the NVIDIA booth situation. Then we'll come back to the frontier. What are you guys showing in the booth? What's the use case? What's the feature?

Sure. So, we're showing our quantum computer connected to Nvidia's latest GPUs. It's the GB200 series and so it's right at the NVIDIA booth. And as you pointed out, GPUs help push forward quantum computing. So, we have shown now in a publication just a couple months ago solving a material science problem for design of better batteries, better solar cells with our quantum computer in tandem with Nvidia's GPU. And so that's one of the main use cases in the near term where we expect QPUs to be co-processors accelerating GPUs and vice versa.

Do I get it right if I say classic computing was data processing and then material science is a whole another world that quantum is more on that side? Is that how should people think about? I'm trying to figure out a way to explain it to someone who wants to try to understand. I don't get the quantum piece. Is it like an Nvidia GPU? Well, not really, but kind of. But like, so explain how to position quantum and then specifically how's Nvidia using it to help the engineers and help people build.

Yeah, sure. So, there's a great analogy actually just to GPU. GPU, we sometimes forget the G stands for graphics. So, the very first use cases of GPUs were for video games, for rendering your computer screen. For the same reason, the QPU is first going to be used for quantum applications that are manifestly quantum when you zoom in. Material science being one of them. When you really pull in, you find electrons interacting with electrons as quantum. But just like GPUs today, now they've been used for cryptocurrency mining, for AI, things that have nothing to do with graphics, there's this next frontier of quantum applications that have nothing to do with quantum. It's things like factoring very large numbers that matters to encryption. It's things like extending the context window of AI. So, it's designed for solving quantum problems, but it unlocks entirely new worlds out there of breakthrough applications just broader in compute.

Talk about security. You guys in security, that's a big piece of it too.

It is, it is indeed. I think we're passing through the noisy era of quantum computing and going into the real era of quantum computing. And with that, the challenge comes around harvest now decrypt later, which is what bad actors do when they steal that current data which is secured with some sort of a cryptography. And if it is asymmetric, it's prone to be broken down by the quantum computers. And that's where we are being more advancing towards in terms of providing quantum safe algorithms. And NIST has done a fantastic job in terms of releasing all the quantum safe algorithm logics. So that's what we are integrating and we are providing support to our customers. But also supporting through the journey like Amit talked about, right? Majority of the enterprises are confused, what state they are in. These are the data centers which are built like 50, 60 years ago and have evolved over the period of time, but they don't know where the cryptography stands. They don't know where they're using what kind of algorithm, what service is protecting and what not. So we help with the entire transition to the post-quantum cryptography, and that is important.

And you guys have some news that this week. You can't really reveal it because you'll get in trouble, but...

Adam, talk about Sandbox AQ. Jack was on, very interesting story. Probably the biggest startup that's ever been incubated at Alphabet.

Gosh, I don't know about the size. Thanks, John.

It's like the Y combinator of Alphabets that comes out. No, but I mean it's a lot of hard problems you guys are attacking.

Yeah. So, we're doing some, and thanks for having us on the cube, by the way. It's kind of our voidal prism. Anyway, we have a division working on cryptography. Started off with a post-quantum cryptography angle, is now really providing cryptographic agility, as I'm sure will come a bit more later. We have some quantum navigation and hardware aspects. And then we also have a business unit really focused on drug and material science problems both from an AI perspective and from a physics-based modeling perspective. And I think we are starting to see some very exciting applications of quantum to those fields. The case for the modeling of drug and materials, it's kind of subtle actually. You really need to find a problem where a classical computer doesn't do a good job already of finding a good material or a drug. It was looking for a little while there like maybe quantum computers could just be like big prime number factoring machines. I don't think that's the case. There really are applications coming right now where I think we're going to see these things providing some good calculations of things like transition states, predicting drug molecule binding not binding affinities, but they're called off times, which is sort of a superior proxy for for photosy and some other material science. Talk about some of the hard problems you work on for people that haven't seen Jack's video, but it's basically my lame attempt to explain a frontier model for science.

That's way is that a how should people think about the problem that you guys or the models you're building?

Yeah, we like to call them LQM as kind of an intentional positioning against the natural language space. It's not so much like we have one big model called an LQM that does everything you would want. But it's more like we really see that just throwing more cat videos and text into a big box is not really going to lead to the emergence of AI that can treat hard scientific problems where you really need to get the structure of the molecule, the exact quantum behavior of a system, really how this thing is going to work in the lab. It's not like things like Gemini won't be useful for that. Of course, they will be, but they're not going to scale up to that on their own. That's kind of our core core thesis. To do that, you really need sophisticated algorithms and ground truth data that's specific to the problems at hand. And one major application of quantum is in computing those ground state systems without needing a lab in the first place, we would say.

So there's two things going on here that you guys are all first of all, great work that you guys are all doing. It's amazing.