Does Traditional Encryption End in 2027?

Key Takeaways

If I just think of like all the infrastructure that's been deployed and TCP IP and everything that we use on the internet today and the companies rely on, you said this is much bigger than chat GPT moment and it sounds like it's crazy big. And when you do have access to a quantum node that is large enough, you'll be able to compromise all of that traffic starting now, T0. And so everything that we think is confidential today becomes plain text at that point. The problem is every website, every network connection, every VPN is going to be affected by this, right? >> Yes. 100 cubits, you can get 100,000 cubits in the next 2 years. And so by doing that, now you can run an algorithm that requires 100,000 cubits on a distributed quantum node or a distributed quantum data center that could be available way faster than you ever thought, which means that your encryption, my friend, is not safe. Hey everyone, it's David Bombal back with the amazing V Joy. V Joy, great to have you on the show again. Always a pleasure. So you are the man I talk to you when it comes to quantum computing. It seems like things are rapidly progressing. Can you give us an update about what's happening? >> If you think about quantum computing, I mean we've all seen the transformer or the chat GPT moment where there was a step function change in capabilities available to us for computing and it changed the game. I mean it changed the way humans do work and that was a step function, that was a paradigm shift. Quantum is going to be like 100x bigger than that because the entirety of computer science theory would have to be rewritten for that to happen. And it's imminent. That paradigm shift is imminent because I recently drew a graph and we plotted the number of announcements of significance in quantum computing over the past few decades and the graph is just going exponentially high, which tells you that something is just about to crack open and just make our lives incredibly different. And so that's what's happening. >> When we last spoke, sorry, you were showing me a new chip that Cisco had developed and you said all these changes, a lot of announcements happening. So give us some updates since we last spoke and then, you know, how long do you think it's going to be before all encryption is broken? >> Well, if you have a timeline in mind, Yep. I would actually reduce that by a fifth. Oh wow. At least. Oh wow. And the reason for that is, I mean we at Cisco, we are a distributed computing company. Yep. So we are in the business of building a network that connects servers in the classical world. So think about all of these cloud data centers. Yep. We're in the business of building a network of fabric that connects all of these servers to enable distributed quantum computing, I'm sorry, distributed computing whereby you can run much larger workloads than you would be able to run on a singular machine. Yep. And that's the whole thesis behind scale out. And with quantum, we're trying to approach the same design pattern. So we're saying, yes, whereas all of these computing companies have a roadmap that talks about 100 cubits today, maybe 1,000 cubits in a couple of years, 100,000 cubits in maybe 10 years for any practical application like even compromising security, Yep. we are looking for 5 to 10 years as a horizon if you're using a singular quantum machine. Okay. But we just came in and changed the game. We said, we're going to connect all of these machines. So instead of 100 cubits, you can get 100,000 cubits in the next 2 years. And so by doing that, now you can run an algorithm that requires 100,000 cubits on a distributed quantum node or a distributed quantum data center that could be available way faster than you ever thought, which means that your encryption, my friend, is not safe. Wow. Okay, so you see you said 5 to 10 years, but reduce that by a fifth. So perhaps in 2 and 1/2 years, 3 years, we might have that. I I I believe that in the next 2 to 3 years, we will see distributed quantum computing take shape where we are taking a bunch of these quantum nodes from various vendors, all based on different technologies. They don't They don't have to be the same kind of technology. And we'll be able to connect that through a fabric and run algorithms and workloads in a distributed manner within a data center. I believe that that is going to happen in the next 3 years for sure. Wow. The problem is every website, every network connection, every VPN is going to be affected by this, right? >> Yes. And so encryption mechanisms, Diffie-Hellman or RSA, that's the key exchange mechanism that we use today primarily, that is susceptible for compromise if and when these nodes come online. And even as we wait for those, there's this whole class of attacks called store now and harvest later attacks. Yeah, a lot of governments doing that, right? So explain that, sorry, for people who don't know. Yeah, so the the the simple idea is that I tap into your network either physically or through a route hijack Yep. and I store all your traffic Yep. today. And maybe I have access to quantum nodes today, maybe I don't. Some nation states, we feel, do have access today. Oh wow. >> But let's say they don't. Yep. Even in that case, you would store all of this traffic which is highly classified, confidential. And when you do have access to a quantum node that is large enough, you'll be able to compromise all of that traffic starting now, T0. And so everything that we think is confidential today becomes plain text at that point. Which to you and me, if you're buying socks on Amazon, it's not a big deal. Yeah, exactly. Uh and it's also temporal, I don't care. Yep. But for governments, for federal agencies, for financial agencies, for health care, for retail, for power grids, I mean this thing becomes really, really critical. So that's what we're facing today. And so there is a stopgap measure that we can deploy today Okay. things to to to help us buy time. Yep. Literally. that is post-quantum cryptography. And so this is a software change. So think of it as a new version of uh TLS, Okay. a new version of SSL and and and uh HTTPS. And it's based on newer encryption algorithms that we feel, and I say feel, we feel are resistant to quantum attacks. And the reason I say we feel is because one of these standards actually got compromised a while ago, 2 2 years ago. Got compromised on a laptop within a week. Always. So it's very hard to prove and say that this is provably quantum safe. Yep. That's where real quantum mechanics comes in, quantum computing comes in, and quantum networking comes in. Well, I mean it sounds, if I just think of like all the infrastructure that's been deployed and TCP IP and everything that we use on the internet today and the companies rely on, you said this is much bigger than chat GPT moment and it sounds like it's crazy big. Yeah, so I think there there are multiple dimensions to this. So one is the encryption problem that I'll get into a little bit more in a second, but the other dimension to look at is why even bother with quantum computing, forget the networking part of it. >> Okay. So why even bother with quantum computing? And there's a Feynman quote and I'm going to butcher it, but I can tell you the the the TLDR of the quote is And and he uses a bunch of curse words in there, so you should check it out. I mean it is amazing. But he was He's credited with inventing quantum computing. And his quote is that nature is quantum. And so you cannot simulate nature using classical computing Yep. because you're trying to force fit something that it's not. And so when you think about nature-oriented problems, so drug discovery or materials or just newer laws of physics and the way we understand the universe, I mean these things today we're trying to force fit it into classical computing that is deterministic. I mean you hear all of these LLMs and foundation models going after this problem, but in the end you're force fitting a nature simulation problem into classical compute. >> Yep. And that's why you're running into these huge clusters and even those are not enough and I need more and more and more. Yep. If you take the same problems and simulate it using quantum computing, you get resolution in much smaller footprints. Okay. So it's it's sustainable, it's small, it's quick. You can explore a much larger space more quickly than you can with classical compute. So there are these whole classes of problems like even logistics and and supply chain problems that traditionally you think of them as hard problems in computer science, so these are NP-hard problems. And NP-hard problems and hard problems actually become easier to solve for because they're not exponential anymore >> Yep. when you deal with it in terms of cubits and the complex complex other the exponential spaces that they can explore through quantum computing. Wow. So the question is, you know, how do we get there? Yeah, so we were talking about the sizes of quantum nodes that exist today. They're at 100 cubits, 150 usable cubits. But we've seen this problem and design pattern in the past where you keep scaling up. Yep. >> So don't stop at 100. Yes, go to 1,000, go to 100,000 and so on. But then you also scale out. Yep. And a combination of those two will get you what you need in terms of scale. And Cisco and the fact that we want to get into quantum networking, we are getting into quantum networking, is going to enable that scale out piece of this equation. Yep. So as compute vendors are scaling up, we're in the business of connecting all of these compute vendors together in a fabric, in a network to enable distributed quantum computing. And so the chip, excuse me, that we talked about May of this year, is the first building block in hardware to enable quantum networking. And it generated 200 million entangled photons a second. And quantum networking actually works on the principles of entanglement, principles of teleportation where instead of sending packets from point A to point B like in classical networking, we distribute entangled pairs of photons to every member in that network and then that's it. And then one when one node changes state, It's amazing. the other node gets that state change instantaneously. Yep. This is the spooky action over a distance that Einstein used to talk about, it's happening on the network today. And it's happening at room temperature, it's happening on telecom frequencies at existing on existing fiber, Yep. no ripping and replacing your networking infrastructure, which is a big deal. Yep. So we announced that a while ago and then just about a month ago, we announced a software stack Okay. to tickle all of this and turn it on. And so we looked at new protocols. You talked about TCP IP, this is a whole new class of protocols. Entanglement distribution, teleportation detection. These are things that seem like voodoo and black magic, but they are pieces of software, classical software, that make quantum network work. And then we announced a few applications above it. So, the first one is network aware compiler. Okay. >> So, that takes a quantum workload, a quantum algorithm, splits it, and pushes it, schedules it across a network of quantum elements. So, this is the first step towards doing distributed quantum computing, but it also allows you to plan for your environment, to work in a brownfield environment, because we tackle heterogeneous nodes in the environment. So, ion traps, photonic, etc., etc. So, you can actually plan, schedule, upgrade, do all of those system-like things that you would do in a classical environment. You can do all of those things in a quantum network as well. So, we announced that, and then we announced two special use cases for classical use cases. >> Okay. And that is the most exciting, David, because yes, all of this is great. VJ, you are saying 3 years, maybe. I was skeptical, I'll think about it as 5 years. Exactly, yeah. But there are use cases of the quantum network, entanglement-based quantum network, in classical computing today. Oh, wow. That is the most exciting to us. As we're getting a whole bunch of these design partners come to us, and these are classical use cases today. So, the two use cases and the two software and demos that we released, one is called quantum sync. >> Okay. The other is called quantum [snorts] alert. And if you want, I can get into the details of how they work. >> Yeah, go for it. Go for it. >> So, so, quantum sync, very straightforward, is like I take a photon entangled pair. Yep. I send you one of those, I have one of the other ones. And let's say we want to use this entangled pair as a starting gun, or as a decision coordination mechanism. Yep. >> Where we both want to act on certain piece of data, or you want to run an algorithm, but we want to do it instantaneously. >> Yep. A use case here is high-frequency trading. Yep. And so, we all have the exact same data, but for fairness, I want to make sure that you sitting in Tokyo, and me sitting in, I don't know, France, Paris. I just love those two cities, but but that's [laughter] okay. We act on those pieces of data, or run that algorithm at the same time. >> Yep. You cannot do that through a No. standard classical network. But through entanglement-based teleportation, I change state, and I say, "I'm ready to go." And you instantaneously get that information saying, "Yep, we're ready to go as well." So, the two of us end up running that algorithm at the same time. And that exists today? So, the demo exists today. The stack, the software stack, exists today. The building blocks more or less exist today. The place where we are actually saying I'm hesitating is we are not shipping it as product at scale today. So, you can't buy hundreds of these. >> But it exists? But it exists. And so, we are working with design partners to actually build the solutions out for some of these end use cases like HFT. And once we are comfortable with those, we go into product at scale. Wow. Because I mean, it feels like quantum, quantum, quantum, it's like it's coming, it's coming, it's coming, right? But the fact that you're saying something already exists today changes everything. >> Changes everything. And so, the other one that we just touched upon is quantum alert. Okay. And you talked about security, and when we are there to when are we going to break encryption as it exists today? So, of course, you can go PQC. That's step one. I mean, everybody should do it. It's like hygiene. Yep. Let's do PQC. >> Yeah. I think all of those devices will be PQC enabled pretty soon. So, that's like basic hygiene. We should do this. But if you are more paranoid, and you don't know whether somebody has compromised those PQC algorithms, and you don't know when somebody will compromise those PQC algorithms, what you can do is alongside PQC, alongside even your standard TLS networks that exist today, old networks, the old HTTPS, you can deploy a simple quantum network. I mean, this is not tons of information. This is like a qubit at a time. Okay. And same thing, same principle. You get one pair, one of the pair, I get the other pair. And let's say an intruder comes in. Yep. And if you remember, I said these things are being sent over telecom frequencies, existing fiber. Yep. So, I can send these things along existing traffic and existing networks. And if somebody comes in and taps into your network, guess what happens? I measured the entanglement, which means the entanglement collapses at your end, at my end. Yep. So, I will notice that my entangled state with you has disappeared, which means somebody is breaching into the network. So, through the laws of physics, I can provably tell you Wow. that somebody is actually intruding in your network. Wow. And you can deploy this in your network today. Wow. So, that's a that's a product that's that's being sold? >> again, it's >> Still it's still demo stage? >> All of these things, I would say anything to do with the quantum network, entanglement-based quantum network, still prototype phase. We are working with design partners, some of the more paranoid ones that we just talked about. >> Yep. Energy companies, fintech, and so on. And once we are comfortable with those use cases, that's the time to go into product at scale. But I'm guessing a year. Wow. So, by the end of 2026, we might see this in production. I I'm pretty confident that'll happen. That's crazy. VJ, whenever I talk to you, my mind is blown, and you've done it again. Thanks so much for sharing. Thank you for having me.