Episode 147: Steve Girvin on Algorithms, Modalities, and the Road Ahead

In Episode 147, Steve Girvin returns for his third appearance to take stock of where quantum stands across research, hardware, and the commercial landscape. The team cover the ongoing challenge of discovering new quantum algorithms — and why better hardware may be what unlocks them — before moving into the state of the three leading modalities: superconducting qubits, trapped ions, and neutral atoms. Steve walks through the wiring problem facing superconducting systems at scale, why cold atoms are on a steeply rising learning curve, and what photonic approaches like Xanadu and PsiQuantum are betting on. The conversation closes on dynamic circuits — a hybrid model where quantum and classical computation are deeply interwoven mid-algorithm — and why that may be one of the most powerful and least understood frontiers in the field.

Transcript

SPEAKER_04 0:41

Hey Stiprin, how are you doing? Hey, Patrick. I'm doing well. Looking forward for another episode of Entangle Things.

SPEAKER_01 0:48

Oh, this is this is going to be a good one. We're joined by a repeat guest, third time's the charm. Steve, do you mind introducing yourself or reintroducing yourself to our audience?

SPEAKER_03 0:57

Sure. Hi, I'm Steve Gervin. I'm Sterling Professor of Physics and Applied Physics at Yale University.

SPEAKER_01 1:05

And we're very happy to have you back. Last time we were talking about Adams, Ryburg Adams, I think.

SPEAKER_03 1:13

Bidberg, yeah.

SPEAKER_01 1:14

Yeah. And uh, and and so I apparently I have to listen to that episode again. Uh, and um, you know, you're you're in you're in the middle of everything. You you've got a a really good seat to watch uh as things developed. Um is there anything that you're noticing that we should be talking about or thinking about? I mean, you you right now the the business side seems to really be heating up.

SPEAKER_03 1:36

Yeah, there's a lot going on uh at the at the federal government level, uh in the on the business side of things and and also you know on the the physics and and computer science uh side of things where I where I live. Um at the federal level, um uh you know, quantum continues to be a national priority and part of, you know, we're in sort of a worldwide race to try to build useful uh quantum technologies. The uh the five Department of Energy uh national quantum initiative research centers were recently uh renewed for a second five-year term. Uh so the center C2QA that I'm involved with, with the Brookhaven lab, is uh is in phase two now. Um there's um uh Dario Gill, who's the Under Secretary for Science in the Department of Energy, is pushing AI uh very heavily with his Genesis program, and a subset of that is uh connecting AI and quantum. So there's a lot of it's obviously a lot going on in the world in connection with AI, but uh AI and quantum combined two uh amazing buzzwords at the at the moment.

SPEAKER_01 3:13

Is so there's some conjecture on whether AI is really, really accelerating quantum. And a lot of people have pushed back on yes, it's helping, but it's not it's not like a tenfold increase, or or or maybe it is. I uh what's your take on how much help, how much of a tailwind is AI providing to quantum right now?

SPEAKER_03 3:31

Uh well, that's a good question. Um and there's sort of two sides to this. Some people are interested in could we use quantum to help AI? I think that's uh very far in the future. But uh AI is definitely being experimented with to help uh quantum to um there are difficult tasks like searching for new algorithms. Um there are more straightforward tasks that people have been doing for a few years now. Uh using uh reinforcement learning to turn all the knobs on the experimental apparatus. If you have 40 or 50 experimental parameters you have to tweak up to get your quantum computer to perform better, it's very difficult to do that by hand.

SPEAKER_01 4:25

Yeah, that's an obvious choice, I think.

SPEAKER_03 4:27

Automated reinforcement learning uh has is become very important now. Um there were experiments uh in the uh Vlad Sievac here at Yale a few years ago. He's now at uh Google Quantum AI doing it on a very big scale. I think that's a pretty natural uh and obvious use. But people and it's not so that's that's clearly working. Um I've I've gotten uh interested in it because uh you know uh for some reason people aren't still programming in Fortran 77, and I haven't learned Python. Have they moved to 78? Is that what I think? There was a Fortran 90, but uh but I actually have been able to write simulation codes recently by asking Claude uh uh to do it. And as far as I can tell, it's doing a good job, even though I cannot actually program in Python. Uh so it's definitely a lot of my colleagues are are using AI as a kind of uh uh very bright, a very bright but inexperienced graduate student uh who makes mistakes.

SPEAKER_01 5:49

Well, as long as they they check for the mistakes, I think they're on the right track.

SPEAKER_03 5:52

So yeah, so I don't I just at this point, I you know people are also trying to do use this lean uh formal math prover. If you can if you can formalize the question you're asking very precisely, you may be able to prove a theorem about it. Uh but how useful that's going to be in the quantum domain, I think remains to be seen. It's been proving quite useful, no pun intended for mathematics.

SPEAKER_04 6:23

So you mentioned searching for new algorithms. This is a uh uh debate, quote unquote, that Patrick and I have on the show. I'd really like to get your input on how much of searching for new algorithms is hey, we need to wait until we have working quantum computers to validate and explore, versus it's just hard to find brand new algorithms in this space. Because we're hearing both sides, right? There are people saying, look, we don't have the algorithms because look at what happened with classical computing. Once we had the working machines, right, there was a lot of interesting development. Others are saying, look, it's difficult, really, really difficult to identify like brand new algorithms. So where are we in this space?

SPEAKER_03 7:15

I think both of those statements are true. I mean, I I may have said uh the last time I was on the show that um, you know, weirdly, before there were classical computers, even mechanical ones, there were people thinking about algorithms. And uh but once there was actual hardware, a lot more people started thinking about it. It didn't, it didn't make it well, it made it slightly easier in the sense that they got frustrated because certain things were hard. And that told them what they should focus on to, you know, to make it better. But it's not uh and I think the same thing will happen uh as people gain access to quantum hardware. But really, it's just a very hard problem. Um, the way basically all quantum algorithms work is you s it's very easy to take a first step and put all the qubits in the input register into a giant superposition, and then the algorithm has to prune away, use some kind of destructive quantum interference to remove the parts of the superposition that correspond to the wrong answers and reinforce the parts that correspond to the right answers.

SPEAKER_01 8:42

It sounds so easy.

SPEAKER_03 8:44

Yeah, exactly. And you know, I gave a talk, sort of public lecture, to some high school students, and a high school student asked me a very perceptive question. If you you're writing your algorithm to solve a problem that you don't know how to solve, you don't know the answer. So how do you make it remove the wrong answers and enhance the right answers when you don't know what it is?

SPEAKER_02 9:09

Yes.

SPEAKER_03 9:10

And that's that's sort of in a nutshell why it's very, very difficult.

SPEAKER_01 9:16

Did we award him a master's degree on the spot? Aaron Ross Powell, Jr.

SPEAKER_03 9:19

Yeah, I was tempted, actually. It was a very perceptive question. So I think it's a little of both, you know, but people uh it's it's just not we haven't developed the kind of um intuition yet on how to do this, how to discover new algorithms. There is a sort of brute force uh optimization method, the way people do optimizers in compilers. You try building a circuit and see if it gives the right answer, and if it doesn't, you punish it by changing the gates and the parameters and hope it gets closer and closer to the right answer. But that's um that's very brute force, and it's a very that's a very, very, very high, exponentially high dimensional optimization problem where pure brute force isn't really gonna work. Um so uh I think it's just gonna take uh practice and experience with real hardware and thinking, but you don't you don't absolutely need the real hardware. You need to just develop some experience.

SPEAKER_01 10:39

Some intuition.

SPEAKER_03 10:40

Some intuition about how you can use quantum interference to help you out.

SPEAKER_01 10:46

So that yeah, there's a if you look at Grovers and Shores, the two most well-known named algorithms. Um, in the early days when we started this podcast over five years ago, can't believe that. Um, we we were thinking that like we were right around the corner to there being dozens of well-known named and and then we started to think that, or we started to hear with some of our guests, as well as think it ourselves, that maybe there just weren't enough NP problems. There weren't enough hard enough problems that would have this advantage. And then but since then, there's we've had cubos and other, we've seen this applied to so many different places um that it it feels more like what you're saying, which is as we get into the space, um, we'll we'll realize, oh, we might be able to apply it here, we might be able to apply it there. I I've had the same experience with AI, and and Cyprian is AI is his primary area, but cyber is mine, where I saw these great tools and I'm like, yeah, what am I going to use it for? And hearing other people's experiences of, well, I used it for this and I used it for that, kind of unlocked the, oh, yeah, I could use it over here. And I it weren't the same scenarios, but they were similar. They rhymed, as as Mark Twain might say. Right. And so you feel like we're in a rhyming scheme now where people are expanding their their worldview to include AI and what can it do for us? And I've I've read some uh some gleanings. I haven't read papers, but I've read some some reports of people using it for machine learning. And so I'm not sure if that's apocryphal or actually real. But if it if it helps the machine learning, that's a that's a really big basic, that's a very important use case.

SPEAKER_03 12:31

Yeah, I mean, I think uh we, you know, it's an amazing new tool, uh which sometimes hallucinates and lies to us.

SPEAKER_01 12:42

Uh not that not that different from a grad student, I guess.

SPEAKER_03 12:45

Exactly. But uh, you know, I think we're we don't know yet, really. I mean it's it's it's radically people are experimenting with sort of radically changing how they do work. But it's um, you know, there are a lot of there are a lot of appropriate concerns about uh if people rely on this too heavily, uh are they gonna forget how to think, or are they gonna fail to you know wake up every morning and instead of in addition to saying, could I be wrong, could uh Didn't they say that about the I think Socrates said that about the written word.

SPEAKER_01 13:28

It was making people stupid because they would not have to remember things.

SPEAKER_03 13:32

Exactly. Exactly. Yeah. So that's um it's definitely a topic that's sort of uh the merging of AI and and quantum efforts are certainly uh it's a very, very hot topic now. And it I I'm getting interested in it, but I I don't know where it's going. I will say I think it's safe to say nobody does.

SPEAKER_01 13:59

So I have a theory, and maybe I'm wrong, but I'm willing to put a stake in the ground. I believe that we are in the AI peak hype bubble. And and and I I I think just like electricity and and the internet, it's not going away. But I think that some people are gonna have to cool their jets about their predictions and things in the not too distant future. But I also see the quantum bubble coming.

SPEAKER_02 14:22

Yeah.

SPEAKER_01 14:23

And in between the two, I think we're gonna get the robotics bubble that's that's building very quickly. And I'm wondering whether the the shock, when there's a little realization of the things that AI can't do, will be offset by the other two bubbles coming, and would in and we may get a continuous wave across the three. Because AI is enabling robotics in a big way. Robotics probably won't have as big an impact on quantum, but but it might. But it I could definitely see AI helping quantum, you know, bring in the timeline. We saw that Google change their timeline to say we're gonna be quantum safe by 2029 because things are happening so quickly. And you know, we use Shore as the canary and the coal mine. It's the it's the bellwether for where are we on that on that maturity curve for quantum.

SPEAKER_03 15:14

Yeah, I mean, I think the there are two um big problems to get from here to, you know, a shore factoring machine that's not that works at a big scale. Um the primary one is to build uh a fault-tolerant computer that can correct errors without um requiring millions and millions of redundant qubits. And so quantum error correction is is having a moment now. People are beginning to be able to experiment with it on real machines, but people are coming up with a lot of very clever ideas for quantum codes which have high rate, that is, which store many logical qubits inside not so many uh uh physical qubits. That is, you don't need huge levels of redundancy. And uh that that's still an enormous challenge, and we're still at the very beginning of that, but lots of interesting things are are happening. The other problem uh really is just improving quantum control, making the gates better, faster, more accurate. And it's it's the same for many different technology technologies in a in a large sense, but very, very different in detail. So um for superconducting qubits, we're suffering from this wiring problem that we have um two or three wires coming down into the refrigerator for every single qubit. And when you get up to thousands of qubits, you have a huge uh noise and heat load on all these wires going down into the into the um refrigerator. A refrigerator. Uh Google recognized that it has this problem and recently purchased uh Atlantic Quantum to um to develop a cryogenic uh control system based on also superconducting circuits, but classical. They move little uh little lumps of magnetic field around uh to perform gates and control frequencies of qubits and things. Northrop Grumman has also been working on that. So that's a very low power dissipation kind of computation, which brings you closer to the paradigm, like in a you know, a large CPU chip, which has a hundred billion transistors in it, but only a thousand wires coming out of it.

SPEAKER_02 18:21

Yeah.

SPEAKER_03 18:22

That's sort of the direction we have to go. So that's that has to be fixed.

SPEAKER_01 18:26

Aaron Powell And with semiconductors, with standard semiconductors, geometry has become a way to overcome some of the spacing barriers. They're you know, some of the newer chips are building up as opposed to out.

SPEAKER_03 18:38

Yeah, yeah. There you um well there's a lot a number of companies that are doing um it's still 2D, but uh this flip chip arrangements where you have maybe the control circuits are on one chip and uh and the qubits are on another, and you flip them over so they're facing each other, separated by a spacer that's only a few microns uh thick. And you have to have a special machine to make sure it's the same number of microns on the left edge of this uh several centimeter chip and then on the right edge.

SPEAKER_01 19:15

And and Cyprian said many, many times, we're we're very fortunate to have the precursor of classical and the semiconductor industry to learn, or at least to like copy from and say, well, this worked there, maybe that will work here. Yeah. Um, we didn't have that back in the vacuum tube days, but you know, who knows? Maybe in the biological computer days, they'll have two examples.

SPEAKER_03 19:39

Yeah. Then on the on the uh the other area that's really exploding right now, and and which we talked a bit about last time is uh cold atom arrays or Ridberg atom arrays, where optical tightly focused laser beams are used to trap individual single neutral atoms at the focus with uh using uh forces of light. And then uh you can move them around, move them near each other to form gates, uh, move them to a far distant point to form another gate. Um the moving part is slow and takes time, yeah. Yeah. Takes time and and um the atom tends to wiggle around when you accelerate the thing that's holding it, so that can lead to um decoherence. But when they're sitting still, the you know, the they have uh coherence times that can be a second, which you know, superconnected people would kill for. Yeah. But the uh there's a sort of fundamental law that if your qubit is has a coherence time that's very, very long, it's because it doesn't couple to anything in the outside world, including your control system.

SPEAKER_02 21:04

Yeah.

SPEAKER_03 21:04

So the gates tend to be slower. Or in the c in the case of Rydberg atoms, the the gates are not super fast, but fast on the one-second timescale. But you you you send in a laser pulse that moves one of the electrons far away from the atom uh into a big blob uh uh outside, and it's and that allows it to interact strongly with other atoms, but also exposes it to sensitivity to noise and so forth. So so they need to do error correction for gates, not so much for memory. And superconductors have fast gates, but can't idle very well. So we we you know, so that everybody has some problem related to error correction, but it differs in important details.

SPEAKER_01 22:00

Brings an interesting, and this is probably an unfair question, and we can strike it if you really want. But if if there were a roulette table of modalities and you had, you know, a hundred chips, would you spread them equally because there's just no way to know which ones are in the lead? Or is it like week to week, month to month, quarter to quarter, where this one's doing better and then this one's doing better? Because we there's a lot more modalities than when we first talked to you. We first talked to you within the first year of the show, so it was over four years ago.

SPEAKER_02 22:28

Yeah.

SPEAKER_01 22:29

And um, a lot's changed. It's just, I mean, some of these modalities weren't even on the on the board back then.

SPEAKER_03 22:36

Right. Well, I think the three leading contenders are still the same, uh, superconductors, trapped ions, and neutral atoms. And there are, you know, large companies and or large startups uh uh addressing all of those. There's um there are spin qubits, quantum dot qubits. Um that's probably they are making progress, but are still pretty far behind.

SPEAKER_01 23:14

Um There's the Wildcats that are trying something completely different like Microsoft.

unknown 23:19

Yeah.

SPEAKER_01 23:20

I don't know of anybody doing is anybody doing universal gate quantum computing with photonics?

SPEAKER_03 23:25

Uh yeah, there are companies working on that. Um uh uh Xanadu in Canada and PsyQuantum, yeah. In many different countries. Uh they're taking a novel approach of uh trying to jump directly from zero to a million qubits without passing through one or two. Yeah. So that's uh that's one way to get ahead. Yeah. I mean they they're their their argument is that they're gonna build everything with the on the on the existing industrial base for building optical uh nanophotonics, you know, optical chips. And uh it's uh it's a very well-established industrial base where you have these work packages, you specify what the circuit has to do, and they then they compile it into um a sequence of fabrication commands, and and it and then it's supposed to work. But but uh you know it's much harder to do this at the quantum level than the classical level. Classical optics, if you need more signal, you can just put in more light, and if you lose a few photons, it's not a big deal. But in a quantum system, if you lose a even a single photon, the environment learns what's going on in your computation and it becomes um very collapses. So uh those, you know, there are lots of big claims out there. Uh and you know, the world could change in a minute when somebody gets some new technology suddenly working well. I mean, I would say that cold atoms are definitely having a moment. They're they're you know on a rapidly, uh steeply rising learning curve moving uh very fast. Uh but the you know, there's continued progress in um in the superconducting world as well. So I wouldn't want to I would you'd spread your chips out, which is like the smart bit. I mean, you know, uh ion traps have very high fidelity gates. They're the wall clock time as they move the ions around uh and the gates are pretty slow, but they can do, you know, uh they have some real strength in terms of fidelity of operations. So the the um, you know, even if you get error correction working, so the errors are even lower, if the wall clock time is really slow, it's it makes it harder to actually beat classical computation. And of course, the our friends in the classical computation world uh love it when somebody claims that they've got a quantum computer to do something that beats what they can do, and they sit down and and come up with a new algorithm to beat what the quantum people are doing. So that's a that's actually, I think, a very healthy competition as we we try to uh uh make each other better and better. Yeah.

SPEAKER_04 26:59

Speaking of that, um one of the things that we're seeing kind of as a directional maybe development, right, is uh embracing the idea of the hybrid approach. Because the reality is most problems, right, can't be solved with quantum only. The reality is they are probably solved mostly with classical, and then certain very difficult parts of the problem, even sure is a great example, right? Well we'll do so. Uh how do you feel about this approach? Do you think it's it's gonna be there? And kind of like an immediate follow-up is Patrick and I came to believe that as opposed to the history of classical computing, it's unlikely that we will see a winning modality. Um, at least not any time uh uh soon. So we think this is where the history will bifurcate, right? Where in quantum we will likely live with with multiple modalities, each of them being kind of geared towards specific and maybe classes of problems or or uh uh things that they can solve. So how do you see this moving forward?

SPEAKER_03 28:12

Yeah, yeah, so I think that uh that's almost certainly going to be the case, especially in the near near-ish term. Um there's uh um some things are more natural on one platform than another. And if your problem needs a certain set of gates or a certain type of um errors or something that you want to try to get rid of, you may that may affect your platform choice. The um and so I think that that will definitely pass through an era where these machines begin to become actually useful, first as scientific instruments. I mean, my friend Harry Berman from Quantinuum likes to liken the invention of the quantum computer to Anthony Van Leeuwenhoek's invention of the microscope, where he made lenses out of little tiny drops of water and looked through them and discovered uh amoeba and animicules, I think they were called. Uh and you know, who knew what that was going to be used for, right, at the beginning? Uh there's another uh there's several kinds of hybrid that are possible. You know, your your typical modern laptop or other computer has specialized chips and it has GPUs and and now TPUs and uh uh so forth. So the that could also happen, but you know, if you wanted to use cold atoms for your one-second memory and the superconductors for the fast gates, you have to find some way to do I/O and transfer between them. It's probably going to be optical. Um but that's uh that that phase will come after both are working on their own. And then the what where your question originated, um the you know, hybrid quantum classical computation, I think that's definitely uh going to be very important. I mean, you're gonna have to be tightly coupled to a classical computer to do the error correction and the decision making about what the errors are and how to fix them. But you're talking about at the algorithmic level where I do some pre-processing on my classical computer, I hand off a specific task, take the Fourier transform of this, or or you know, do uh do some linear algebra or something that the quantum computer is good at and hands it off, and then the the results come back. The um and then there's I mean, I think a very important direction is uh related to this, is goes under different names, but dynamic circuits or measurement and feed forward. The original basic idea behind quantum computers was you create some starting state in your input register, you execute a series of what are called unitary gates, and then you make a measurement on the output register, and that's you know, you can do universal computation with that. Um, but there's a more powerful computational model in which part way through the algorithm you stop and measure a subset of the quantum bits, and you get zeros and ones, possibly randomly. And then conditioned on that, then you measure those, you get some classical data, some zeros and ones. You do some classical computation to decide what, based on that, what unitary gates to apply next. And that's provably the most powerful uh computational model. And it it um it's it's not it's a little lower level, say, than just the um classical computer does something and hands it off to the quantum computer. It's deeply, the quantum and classical are deeply mixed together in the middle of the circuit. And this uh this offers a number of advantages. For example, um uh you know, one of the features of cold atoms is that you can move them around. So there's a sort of all-to-all connectivity, same thing with ion traps. Uh but as I say, the moving is slow, so that you know there's there's still problems. But that all-to-all connectivity where you can do gates between any two qubits or any group of qubits by moving them near each other, is very powerful. And typical superconducting setups do not have that. Right. But this mid-circuit measurement and feed forward actually allows you to communicate entanglement across your fabric of near-neighbor coupled superconducting qubits very rapidly and can help overcome some of these connectivity limitations. But this is all kind of um, and we talked about the difficulty of inventing new algorithms, but it it inventing new algorithms that involve some quantum, then some measurements, then a decision classically, and then some that's that's even more complex to try to understand. We've picked up, you know, we know a few tricks, you know, a little uh subroutines or things to help with this connectivity problem. But we really, it's a it's a completely wide open space in which we need much more um much more research to figure out how we can take full advantage of the power of these dynamic circuits. And when then we need hardware advances. We're beginning to be able to do this in hardware, but it's um it's still uh, you know, measurements take time. They have errors sometimes. You don't want to, you know, you might get stuck then making a mistake because you measured wrong. There are many, many challenges.

SPEAKER_04 34:57

And I would I would assume this also will stretch pretty heavily the mathematics behind it, right? I mean the the the combination of the Yeah, that's right.

SPEAKER_03 35:07

I think it's um it's because it's non-deterministic, that is, the measurements can have some randomness to them, uh, and that affects what you do next. And that's kind of uh the dynamics of that system, which is partly quantum and partly classical, is very complex. There's a lot of interest right now in the dynamics of circuits, which we simplify by just choosing random gates and then choosing random bits to measure every once in a while. People are studying that, not because it produces a specific calculation, but it it kind of describes a new phase of matter, a new state of matter where there's uh both measurement and quantum evolution happening at the same time. And that's teaching us things which don't lead specifically to new algorithms, but they teach us about what what kind of things can happen and how the spread of entanglement is killed off if you measure too rapidly, uh, but it can continue to spread if you measure slowly enough. And these are things, these are lessons that help give us some intuition. Which I would be possible we'll be able to uh to use to make make these uh hybrid algorithms, I hope.

SPEAKER_01 36:40

Crazy stuff. So we've been at this for a little bit. We could go on pretty much all day and all night if it if I know Cyprian. Um but before we wrap up, anything else that we should be talking about? Anything coming up for you that you'd want to highlight or any last thoughts?

SPEAKER_03 36:58

Well, uh I mean, I think the the one thing we didn't talk about is kind of the commercial side of things going on. There's I'll just leave it at that. There's a lot going on. There's a lot of uh investment being made. Uh the Yale startup Quantum Circuits was recently acquired by D-Wave.

SPEAKER_01 37:21

Uh I saw that.

SPEAKER_03 37:22

This morning, uh, which is the Ridberg Adam startup, uh, had a Series A for $300 million just about an hour ago. Two hours ago. Uh there's this is just huge amounts. Some of the large companies are purchasing other companies. It's just a huge amount of activity, which is you know either a sign of uh very healthy ecosystem or the beginnings of a bubble.

SPEAKER_01 37:51

So now I think we're still early. I think we're still climbing that hill.

SPEAKER_03 37:55

Yeah.

SPEAKER_01 37:56

But uh but who knows? I I we actually have some listeners that listen because they're investing in this field and they're not interested in the physics per se. I was surprised about that. I was at a conference not too long ago, and um I had a banker, uh not a banker, a uh a lawyer who was you know in charge of the IT for his firm who listens to the show. So we we it's definitely becoming a more wide net and more eclectic. I don't know if you're getting cornered more often and asked by lay people about uh quantum computing or not, but yeah, yeah, I do.

SPEAKER_03 38:30

I mean, it's a it's a um it's a difficult area. People who know the venture capital world don't generally understand the technical details and challenges, and they have to rely on they need if they're gonna succeed, they they need to rely on on good advice.

SPEAKER_01 38:52

Yeah. Yeah, it's very speculative right now, but I think I think there's enough interest in I as I said, I think we're gonna there's three bubbles and it might benefit from being the third.

SPEAKER_03 39:01

Right. And I think it's just very important for the for the government to continue to recognize that this is all going to be a long-term effort to to get this right, and there'll be ups and downs on the way. And uh I hope that um the um all of government coordinated effort that has been attempted by the last two, three administrations continues, yeah. Continues, yeah.

SPEAKER_01 39:33

Well, it's always a pleasure to have you on. We hope you join us again um in you know many times in the future. But uh thanks for taking the time to talk to us.

SPEAKER_03 39:41

My pleasure. Thank you very much, guys. Thanks, everybody. It's been a real pleasure.

SPEAKER_00 39:45

See you later.

SPEAKER_03 39:47

Bye.

SPEAKER_04 39:47

Bye, everyone.

SPEAKER_00 39:50

Cybercrime is one of the biggest threats to businesses of all sizes and industries. With almost half a million open cyber positions, the problem is compounded by the lack of available talent in the marketplace. At Pulsar Security, our elite team of highly credentialed experts collaborate with you to assess your current defenses and develop solutions tailored to your specific needs. With services ranging from cybersecurity education to advanced penetration testing and red teaming, you can start reducing your risks today. Visit PulsarSecurity.com and let's secure your digital future together.