SPEAKER_00 0:30

Welcome to Entangled Things. Quantum Computing Podcast. Hosted by Patrick and Cyprian.

SPEAKER_03 0:40

Hey Cyprian, how are you doing? Hey, Patrick. I'm doing good. Looking forward for another great episode of Entangled Things.

SPEAKER_04 0:47

I mean, I look forward to all of our episodes, but I've been looking forward to this one for a while. So we're joined by Colton and Rick. Do you uh two mind introducing yourself to our audience?

SPEAKER_01 0:56

Not at all. Um my name is Colton Dillion. I'm the CEO at PostQuant Labs, and we are the development company behind Quip Network, which we hope will become the worldwide quantum computer. Um and I'll let Rick give his intro.

SPEAKER_02 1:09

And I'm Rick Carback or Richard Carback, and uh I'm a cryptographer, I'm the CTO, and now I do spend most of my days doing quantum algorithms.

SPEAKER_04 1:17

That's awesome. That's exactly what we want to talk about. So so I've I got to meet you guys at the conference in Boston, and I said that you know, we definitely have to have these people on our show. You guys, excuse me. You guys are kind of the intersection between the things that crypto brought us and what quent quantum is bringing us. Am I right in that?

SPEAKER_01 1:41

Absolutely. So uh, you know, when you work in distributed systems, it's really, really important to have really high quality cryptography, but also uh these incentives that get people to work together. We've been very lucky with uh sort of the existing internet, the HTTP internet that you use every day, that we've been able to create these distributed systems without a whole lot of additional incentives uh beyond just the incentives of being a data center or a data provider. Um but what we've realized over the years is that this has a centralizing tendency with data, where uh if you've ever used a Google product and Google killed the product, uh now you can't use that product and nobody has it anymore. The data is gone, you can't port it to a new service. If you ever played a video game that you loved and the server went down, and there are entire communities that are dedicated to reverse engineering this software so that they can keep playing. Um, and so over time we've realized that there's a certain important aspect to the internet, which is decentralization and openness, where uh you know anyone can observe the code, anyone can see how the data is structured, and they can build new products around that data. And so uh this has manifested as crypto, which a lot of people may have bad associations with because there are a lot of uh hijinks that that are associated with frontier technologies. Um but under the hood, that's really what it's all about is decentralization, is open source software, is trying to make these things that are maintainable and that uh everyone has a vote in. And so uh when we came into quantum, uh we saw that that's not really the ethos in quantum right now. There are a few open source projects like OpenCasm or QIR or uh, you know, the Munich Quantum Valley benchmarking initiative. All of these things are really great projects, but overall the stack is really quite closed source. It's very hard to get access to the proprietary technologies that these computers are built on. Um, and we think that's an impediment, that it prevents people from adopting the technology because it's really hard to customize it. It's really hard to insert your software into the stack and uh get a different result out of it. And so that really inspired us to think about how we can change that situation, right? There there are, we say there are four major problems with quantum computing. One is that the benchmarks are not super clear. Every quantum computing provider has a different benchmark that they're trying to accelerate. Um so you know, if you're IBM, you want operations per second. That's the thing that your machine looks really good at. If you are I on Q, you want qubit count and qubit connect connectivity and maybe qubit fidelity, so you can do deeper, deeper circuits. Um, and if you're D-Wave, you want to say, well, we only care about uh what is the problem that we're solving and is there commercial application, because they tend to have uh a lot more commercial applications that are deployed than some of these other providers. And uh so you know, we wanted to solve that problem. And there's this nice aspect of decentralized software where you can create these shared data sets and these uh shared solutions. Um sort of the second component is that these machines, you know, they have to be kept near absolute zero very frequently, either uh the entire machine or the area inside the machine, and so they're expensive to run. But the additional cost of serving a customer is very, very small. You're adding watts of energy uh, you know, to the whole system in order to do a computation. Um and so because of all of this uh challenge around understanding quantum, the largest cost of serving the compute isn't actually serving the compute, it's trying to educate your customer about how quantum works. Um exactly. And and so as a result, they have to contribute a lot of time to get you up to speed as a quantum operator, as somebody who understands quantum information science, who understands the specific machine that they're working on. And so they can't just open up their API because the odds that you'll get a good result are really low. So they only want to work with these big customers who can afford the consulting that comes along with these machines, uh, or otherwise, you know, give these small uh test times to people so that they can try and figure out uh how to use an individual platform. And so this is another component that we hope that we can sort of automate away by creating this open source network where anyone who is an expert in this stuff can build software for you and then give you an interface that uh, you know, you don't have to be a quantum PhD to understand. You can just hit an API like any other software. Um and then there's a bunch of stuff about talent pool, where you know, these people are, you know, working for proprietary companies or they're working for government. So all their work is confidential or top secret. And so there's not a lot of sharing of resources. Um, if you look out there, there's a lot of reduplication of effort, building all the same quantum algorithms for all the same applications. And the reason why is because everyone keeps it under lock and key. And so for all of these reasons and more, we think it's super important to get a bigger open source ecosystem in quantum and to uh hopefully make it easier for the average developer to get up to speed in using these systems.

SPEAKER_04 7:15

So if I understand correctly, you're also building, or you or the main thing you're building is an alternative to what I thought was going to happen, which is, oh, you're a company that can't afford a quantum computer yourself, which is most of us. And so you you have to use a cloud vendor. So either D-Wave or Microsoft or somebody puts their system online so you can consume it as a cloud service. But what if I'm a hospital or a hospital group or a university group and I own a quantum computer and I'm not keeping it utilized? How do I resell? I think I can't call Microsoft and say, hey, do you can you resell some of my but this is a solution you guys have.

SPEAKER_01 7:55

Exactly. And and so that's part of the open source software is anyone can download the software and anyone can run it. So we are actually talking to uh, you know, uh the electric power board of Tennessee. They've acquired a uh IMQ computer. Um, and you know, they're interested in trying to make the time on that machine available to people uh because uh they have some experiments that they are committed to uh in terms of their work with uh Oak Ridge National Labs and some other consumers inside of that Chattanooga data center. Um but they largely don't know what it's gonna be used for day to day. And again, you have to keep these machines cool. So it's an expense. If you let it heat up, you're spending a day, maybe two days uh trying to cool it back down and then recalibrating the machine. In some cases, it's up to two months. Like if you go to AWS right now, a lot of the computers you can't actually use because they're being calibrated. Um and so what we do is we kind of fit into the cracks where if you have an SLA, if you have customers, great, serve those customers first. Um but we offer what's called a preemptible spot job. And so instead of reserving time in advance, you can bid for time on the machine against all the other people who don't have a contract. And whoever has the highest bid, they get served first. And if one of their customers comes along who has already paid for the time in advance, you get kicked off. Uh so you they can interrupt your job.

SPEAKER_00 9:21

Yeah.

SPEAKER_01 9:22

But we expect that over time we can make up in volume with all of the machines that aren't being used to make sure that your job gets served quickly. And over time, as people bid up the price of the machine, because you know, we expect that quantum computers are going to be able to break computational bottlenecks and do things that no other computer in the world can do in the same amount of time. And so that's gonna be really valuable. And uh, so what's the value to you of making sure that you get served first on that machine for a you know one-time opportunity that only comes along every six months? Do you want to take the risk on renting all of that time in advance, even though you're not gonna use 90% of it? Or do you want to bid for that time when that opportunity comes up? And you know, the value of that time will take up some percentage of the most valuable application. And to your point, if you already have compute or if you've rented compute and you're not using all of it, well, maybe you can resell it on the network by running our node software, putting in your API key, and then serving jobs with the compute that you're not using.

SPEAKER_03 10:26

What is the what is the unit of measurement uh against which the bidding is is is done? I'm curious to understand. Because you have to have a common denominator across multiple providers of capacity, right?

SPEAKER_01 10:39

Yeah, so think of it like oil, right? Oil has many different grades, uh, many different uh levels of processing, but ultimately, you know, we we create certain categories of it, like Brent Crude, and we put it on a market, and people bid against that category as sort of a uh standard for all oil. Um and so this is very much similar, right? We're commoditizing quantum compute as a commodity, and you can essentially bid based on the jobs that are meant to be done. So the network uh as is normal for what's called a proof-of-work network, you have to prove that you've done work at any given time. There's always a random problem waiting for you where you can prove that your computer is available. Um, because we don't want to reward people who are just like sending a message to the network saying, Oh yeah, we're available, and really their quantum computer isn't doing anything. So we require that you solve a very difficult problem that uh generally a quantum computer will tend to outperform on. And this is proof that your computer was available, that somebody could have requested a job for it. And so the network gives these computers uh compute credits just for being available. And then you, as a consumer, you can look at the library of open source solvers and you can say, hey, uh, you know, I've got a logistics company, I want to solve a combinatorial optimization problem around packing my boxes into a bunch of trucks and choosing how I'm going to route those boxes in order to maximize my profit. And uh that would be like an application in the library. And if you request that, then there's a baseline cost, which is the uh you know, total chip space that we're taking up, right? And the total circuit depth uh that needs to be applied on that chip space. Um and then there's an additional cost, which is sort of the floating tip that you give to the operator. And so if you want to be served first, you increase the size of your tip. If you don't care if you're happy to wait for your answer, then you can give a very low tip. And when the network's not serving anyone, that operator's incentivized to serve your job anyway because they'll get the extra income from your bid.

SPEAKER_04 12:50

Is that priced in a coin, a quip coin, or is that in dollars?

SPEAKER_01 12:54

Yeah, so so both, right? Ultimately, that's the beautiful thing about cryptocurrency markets is that they have very, very liquid markets where there are people who are trying to arbitrage away the extra value of your token against every other asset on Earth. And that's only going to get better over time because now we're starting to bring traditional assets uh into the cryptocurrency ecosystems like Tesla Slap stock or SpaceX stock or uh whoever uh you might be interested in holding on-chain. And so there are people who are trying to arbitrage away these uh discrepancies in value. And uh so the quip token has to be a native currency. Uh, and the reason that is, is because if somebody tries to cheat the system and they live in China or they live in Vietnam or they live in India or they live in Argentina, as a US domicile, I'm gonna find it very difficult to sue them or to take them to court over the damage that they've done to the protocol. And so what we do is we create this compute credit that stands in for money. And uh, if you do something bad, then the protocol reserves the right to take away that token, take away those compute credits. And that's just an easy way to do dispute resolution across borders in a world where anyone can join the network and they may be a bad actor. Um, and so you know, it is everything's denominated in the compute credit, but that compute credit always has a very uh liquid floating price to US dollars or two euros or to whatever currency you're interested in trading from.

SPEAKER_02 14:31

I think our our expectation is your your interface is to, you know, it's a choose your own adventure interface. So if you only want to think about US dollars and you don't care about any of this, it looks like US dollars, and I put my credit card in and everything is done for me.

SPEAKER_04 14:46

Cool.

SPEAKER_01 14:47

Yeah, I you you don't think about this when you buy oil at the gas pump, but there is a very complex ecosystem of futures and derivatives that are making sure that the gas station right there is only paying a certain amount for gas every year, and they're just charging you a floating price around whatever gas they can get on top of that. Um, and so you know, that's the wonderful thing about these very uh mature financial markets is it allows you to plan in advance what your needs are. So if you're a shipping company and you need to use a lot of oil, the fact that there's a conflict in the Strait of Hormuz doesn't blow up your business because you already bought all your oil in advance. Yeah, and there was a very uh active financial market to help you make the decisions about how much that costs. And so we're doing the exact same thing. It's just instead of the commodity being oil, which helps your ship get around the world, it's compute, which helps your business to make decisions that it needs to make every single day.

SPEAKER_04 15:44

So is is this I I know you guys are working on at least I've seen a lot of buzz about you helping secure Bitcoin wallets against ECC vulnerabilities that are expected closer on the horizon than RSA vulnerabilities per se. Is that that where a lot of the activity has started to manifest initially?

SPEAKER_01 16:05

So that was actually our original thesis was was we saw, you know, Rick can talk more about this, but he's been a cryptographer for two decades and uh specifically focused on the quantum resistance problem.

SPEAKER_02 16:16

And uh when we first got 11 years ago, I started focusing and did some early post-quantum cryptography research and work, and I said, okay, this is somewhat of a solvable problem. You're just gonna use a lot more resources to solve it, and then kind of you know put it down and moved on to other things that I thought were more pressing in terms of like privacy protecting technology, things like that. That's that's generally where where I like to play. Cool. Um and and uh things changed in the last three years, so that's all I'm gonna say. Completely shifted my entire uh uh career around because it became very evident that this is gonna happen uh probably within you know by 2030, but more likely, uh I think now we're thinking it might be earlier, and we're about to start entering a gray zone of the computers are plenty big enough, you know, they have enough qubits. It's just you know, the coherence isn't there or the air correction isn't there to break the crypto. So we just don't know, like, has someone figured any of these out? And that could happen privately. So um I suspect our our current estimate of that is is in March of 2028. We're gonna enter that gray zone. Yeah. Um, but it really depends on um or atomic and um I think INQ has the other computer that's coming out with the 10,000 qubits. Could be wrong on that one.

SPEAKER_01 17:31

Yeah. Um and yeah, so so that was originally what got us into this space was uh, you know, we've known in Bitcoin that uh cryptographically relevant quantum computers were going to be a problem for you know 10 years, 15 years. Uh I think Satoshi posted about it in 2011 saying, yeah, we're gonna have to upgrade the algorithms. Um and so when we saw the Google Willow announcement uh at the end of 2024, we said, hey, you know, maybe there's a scaling law. You know, they connected a bunch of chips together, and for every additional chip they connected, they were able to use exponentially fewer qubits on error correction and exponentially more to actually do computation. Um and we started looking into it. We put together a very dumb model at quantumdoomclock.com where we take all of the publicly available performance characteristics of these chips and we just plot out uh, you know, if all of these performance characteristics are growing exponentially, when should we expect to see a cryptographically relevant quantum computer? And it's a 10% chance by the middle of 2028. And uh, you know, European governments, the US government have all advanced their timelines. You know, in Europe now you have to have a plan for quantum resistance if you are a government agency uh by the end of the year. If if you are a US government contractor and you want a new software contract, by 2027, your software has to be quantum resistant. Um, but when we first started going out, we we created a very simple solution that doesn't require these networks, which you know are distributed. You have to get everyone to agree to upgrade together. Um, there wasn't a lot of consensus about whether this was even a problem a year and a half ago. And uh as we started having these conversations, we said to ourselves, well, how can we make this more urgent? How can we show people that these computers are getting uh more powerful? And so we came up with this idea of doing a proof-of-work blockchain where the proof of work was the hidden subgroup problem, where you can uh basically convert any hidden subgroup problem, whether that's like a graph isomorphism problem or uh some other pattern finding problem factorization into an elliptic curve key discrete logarithm problem. And so you can show you, hey, we've published this public key that matches a problem. Can you find the private key that matches that public key and show that you've broken the key? And uh again, this can be applied to really useful forms of work, like saying, hey, does this data cluster over here look exactly like this data cluster over here, which turns out to be exponentially difficult. Um and uh in the process of doing that, we said, hey, this provides a few great features. One, it's a canary. It tells people how capable are quantum computers and when should you start to be worried that they might be able to break your key. But two, it brings together this cluster of quantum computers. And so what do you do with all of these quantum computers once they're all working on the same problems? And that really unlocked our imagination in terms of like, well, uh quantum networking is starting to come along. We've we've shared a GHZ state over kilometers between four different data centers. Uh so one data center emitted uh three states to three other data centers. And uh, you know, we were able to show that this was a reasonably uh probable thing that you could reliably get the outcome from. And so, how far away is quantum networking? Well, we believe that horizontal scaling is probably going to be the most important aspect of quantum computing once we figure out all of the problems between now and that ideal future where we have a quantum internet. Because, you know, famously, the most important aspect of quantum is that if you can find one more qubit, it doesn't matter how big the system is, if you can connect that qubit to the rest of your system, you've doubled your representational space. And so the power of your quantum computer is doubled. And so that's ultimately what we want to do is we want to be that protocol that helps people connect together, that punishes bad actors, that gets all these quantum computers cooperating to solve the hardest problems in the in the world, because there's no other system that is as large as this distributed network. And so it really will solve impossible problems that you can't do on any other machine.

SPEAKER_04 21:45

So, do you expect the the quick coin to be something we uh uh surprisingly, we have a lot of people who listen to the show for investment because they see the the wave of quantum being somewhat masked by the the hype of the AI wave, but when it arrives, Arrives, it's going to arrive and probably by the end of this decade, as you've said. Is the quick coin something people can buy, hold, and speculate with, or is that not something on the table?

SPEAKER_01 22:11

Yeah, so we don't encourage speculation. Uh, you know, all too. It it happens anyways, and people speculate on gold and oil and wheat and all of these things. Um, but it we do expect that quantum compute will only become more valuable. Um, you know, we've talked to some cloud providers where they feel that uh NISC systems, the noisy intermediate scale quantum systems that are available today, don't have a lot of demand uh, or the demand is very spotty just because they are noisy, it's hard to get a great result from them. But once we have fault tolerance systems, they expect that the line will never end, that there will always be people waiting in line to use the computer. And so uh if you expect that quantum compute is going to become more scarce over time and more difficult to acquire, and you're going to need that quantum compute, then absolutely you should purchase the token and make sure that you have some of the token on hand so you can access the compute when you need it. Um and we do expect that this will be available before the end of the year. Um, no promises. Uh we say in the crypto ecosystem that it's always soon, trademarked. Uh and so we are actively working right now. We have uh our smart contract programming language uh should be going live before the end of the month. So you'll be able to uh formulate uh optimization problems under a constraint programming language and we'll automatically convert that into whatever target quantum hardware needs to uh solve that uh optimization problem.

SPEAKER_02 23:42

If you're a programmer, you can access it now. We have a few people already filing tickets, and uh the response right now is yes, we know we're we're working on it. Um so it's all open source. You can go to github.com slash quip network uh and check it out.

SPEAKER_03 23:56

There there is another ecosystem that has another trademark, which is within the next decade. Um that's AI or is that fusion? The fusion.

SPEAKER_01 24:06

Yeah. Yeah. And that that is the trouble with uh frontier technology, is is that it always feels just around the corner. And and that's what separates good VCs from bad VCs and lucky entrepreneurs from unlucky entrepreneurs is whether it really is around the corner.

SPEAKER_03 24:24

So so your your decentralized network uh is based on proof of work. Uh is that correct?

SPEAKER_01 24:30

Correct. So go ahead.

SPEAKER_03 24:32

That if my understanding is correct, then uh the the the quip coin is basically mined by solving those very difficult problems, right? Correct.

SPEAKER_01 24:42

So so it's proof of useful work, correct? Correct. So so the proof of work is not just you know grinding hashes to find a random number that's below a certain threshold. Uh our initial proof of work is solving uh icing models. And if you're familiar with icing models, they can be applied to quadratic unconstrained binary optimization. So if you have a bunch of uh features of a problem that trade off against each other, where if you choose more of this feature, it's really hard to get more of this feature, um, then you can formulate this for uh quantum annealers and gate-based quantum computers so that they can solve this uh with a little bit of edge over classical computers. We we find that with the D-Wave Advantage 2s that we have available on the network today, that uh if you face off an advantage two against 80H 100 GPUs, which are essentially watt equivalent, they'll consume the same amount of energy in the same amount of time. Uh the D Wave computer will deliver an answer about 30 times faster than the ADH100 GPUs, and it'll be a better answer about 92% of the time. Um and and so you know, there is an edge if you are using one of these machines versus using one of these large clusters. Now, your initial capex on the D-Wave machine is going to be higher than purchasing ADH100 GPUs right now. But even if you scale up the capex so that it's about equivalent, uh the GPUs win about 66% of the time versus the D-Wave Advantage too. And uh, even still, you won't be able to get the same level of quality of a solution as you can with the quantum computers. And so that's part of what we're trying to show with this network is these proofs of works really create this public benchmark where now you can see exactly how are all of these machines performing on these problems. And if they're useful proofs of work where you can substitute in a person's real problem for that proof, uh, then you can also show there's a public record of people having demand for this problem. And uh so it really helps to solve this benchmarking issue of what's the right benchmark to search for. And so over time we want to add more proofs of work. Like right now we're working on a random circuit sampling subnet, where essentially the network uh records that a random circuit was proposed and that you solved it in a short enough period of time to be certain that it's quantum. And so, because you have this network recording all of the events, it's very difficult to fool the entire network at once and to say cache a circuit because the circuit is being generated on the spot by the entropy over the whole network. And then once you've generated all of the quantum randomness, then the rest of the network will validate it and show that it fits uh the distribution, um the Porter Thomas distribution that is expected for a random a quantum random circuit sample. And so this is sort of the first use of gate-based platforms that uh, you know, you can't do this on classical computers. And uh we expect that there will be more proofs of work, including this the hidden subgroup uh proof of work. And eventually these will all come together in order to form this uh you know united difficulty function of all of the different proofs of work that have gone into the blockchain.

SPEAKER_03 28:01

So if if I want to solve a problem and I want to do it through through your network, then I have to be able to model my problem against your smart contract language, right?

SPEAKER_01 28:13

Uh yes. So so ultimately we are limited by what the virtual machine can represent. And uh then there's a surface language that goes over that virtual machine that we hope will uh you know make it easier for people who don't know the difference.

SPEAKER_02 28:28

You're touching on one of the strengths, right? So when you when you represent it in our in our stack, we have the ability to target multiple different machines at once. So you're writing your code once and you're able to uh very trivially check it against every available system on the on the platform and find the one that might perform best for you. One of the things that we've recently found is we've been we've you know we've been active on this useful proof of work for six months or so, and we we now have a large corpus of very hard-to-solve problems. And the question, and we still haven't finished the analysis, but the question is is the quantum computer better on certain problems than the classical computer? And we see differences, but we haven't been able to characterize them yet. And we have noticed a particular pattern uh with anticorrelation that the optimization uh annealers, the quantum annealers are very good at. So there's like lots of really interesting things falling out of the work we're doing already. And we're we're you know, we're literally in the middle of year one here.

SPEAKER_01 29:28

So things are going good. And and we've kept it constrained uh because you know, if you try and eat a whale at once, it's not gonna work. You gotta do it one bite at a time. Um, but a lot of the research that I'm doing is actually on a hybrid intermediate representation using interaction nets and the ZX calculus to uh sort of represent universal computation uh under one convergent uh paradigm, uh, where essentially you can switch between the classical and quantum realms by just giving the graph a wire color. Um and so, you know, this is some cool research. Uh in theory, we might be able to do like distributed compilation and uh, you know, pass around one intermediate representation as sort of a homoiconic uh representation of the underlying compute so that you don't have to transform in and out of these uh different IRs. Um but it it'll take some time to get that to maturity and then to get adoption for it. Um and so that's some cool research that should be coming out of this, which will make it so that anything is representable uh in the virtual machine. But uh yeah.

SPEAKER_03 30:37

That's actually kind of preempting my question that I had. Like given a lot of the problems are essentially going to solve going to be solved by a hybrid approach, right, between classical and and quantum. Like take sure, for example, right? Maybe the poster child example, where you use quantum for just uh uh a very specific problem, right? Um are you planning to also provide capabilities around like mixed capacities? Hey, you can solve this in with a quantum computer, you can solve this with classical. Is that something that you folks are looking for, or are you just focused on quantum?

SPEAKER_01 31:14

Uh so absolutely, we believe that hybrid is the paradigm that will continue to outperform. Even if you have a fault-tolerant quantum computer, there's still the problem of transforming the data into something that the fault tolerant quantum computer can uh outperform. And so there will always be a component of the pipeline, which is looking at these large data sets, chopping them up, reducing each of those uh gadgets or each of those um subgraphs of the larger computation and uh optimizing it, and then virtualizing it and splitting it across circuits or serializing the circuits onto the same platform. Um and then once you've done all of that transformation, transformation, transforming the answer back out and making sure that any of the uh graph cuts or other transformations that you did before the circuit come back out of the result at the end. And so we absolutely expect that hybrid is here to stay. You know, you don't generally use GPUs as a single platform solver, they're combined with CPUs uh or they're combined with with other platforms to enable their power. Um and so the question is what are the trade-offs there, right? Um there's this really interesting space in quantum circuit land of like, do you want to optimize on T count or do you want to optimize on Tefoli gates? Like, how often do you want to measure? How often do you want to pull out uh an answer that you can then uh parallel process with uh classical compute while you're waiting for the rest of the quantum computation to come out. And so this is again what we hope becomes a utility for the network, is that it becomes really easy to see the difference in performance here, that um we've sort of separated out the computation layers so that uh if you're a quantum algorithm designer, you focus on the quantum algorithm and then you expose that API to all the classical developers. And if they want to build smart contracts on top of your quantum algorithm, then they can look at all the different applications and they can be the one who says, Oh, yeah, load up a bunch of boxes and I'll do a bin packing problem and then hand it off to the quantum computer, or load up a satisfiability problem, and uh I will uh change it from finance into something the quantum computer can operate on. And because you have this unified API, it makes it really, really easy for people who aren't quantum experts to continue to develop with you and to collaborate. Um, and that's really what we hope to bring to this industry is uh one, that's sort of on-ramp for uh people who aren't familiar with quantum information systems, um, but two, that spirit of collaboration and composability that made web one and web two so successful.

SPEAKER_03 33:50

So folks are essentially commoditizing computing capacity.

SPEAKER_02 33:54

Exactly. That's the idea. It's commoditizing it, it's making it really easy to access and try it. And then it's trying to, when you buy the computer, make it still use, you know, make it make it possible to recoup some of that investment, right? Even if you can't use all of it.

SPEAKER_01 34:10

And this was uh, you know, a very rich space for NVIDIA to play in, right? Like before CUDA, GPUs were not commoditized. It was it was a very specialized uh space to uh that was mostly focused at ray tracing, focused at people who make uh video games and people who do uh visual effects. And then in 2007, NVIDIA released a software package called CUDA that made it really easy to work with graphics cards, where you didn't have to know about L1 and L2 cash or you know what the size of your scratch space on the GPU was. Um, you could just program uh like any other language. And this unlocked all these applications. Suddenly GPUs were used in finance, suddenly they were used in scientific computing. You know, seven years later was the first time that we split a AI training uh run across two GPUs. And so that was 2014, seven years after the production of CUDA. And now, 10 years after that, AI is the thing that nobody can stop talking about. It's the most important thing in the world to tech investors right now. And so we expect that quantum is going to be very similar, that once you make an idiom that is easy for classical programmers to develop on and to understand that it will blow up uh and blow out all of the applications of quantum and make it something that is used day to day in businesses around the world.

SPEAKER_04 35:31

Agreed. We've been at this for a little while. I think we could go on for another another hour at least. So we definitely want to get you guys in. But we do have to wrap up soon. I hate to be the one that throw cold water on the party, but what uh what what are the last things we need to talk about before we wrap up today?

SPEAKER_01 35:49

Yeah, I I mean the last things from my side is that we are fully open source, right? We want to build a community. If you this interests you, if you are the type of person who likes tinkering with quantum algorithms or who likes tinkering with new technology, please come join us. Uh, you can go to links.quip.network uh and you can join our Discord, you can join our Telegram. Uh we have docs. Uh our GitHub is fully open source, so you can review all the technology. If you want to run a node on your computer at home, you can validate the outputs of the quantum computers and you can compete against them to benchmark them, and you can earn some quantum compute credits for doing that work. We're in testnet right now, so we expect to release the mainnet later this year. Um, and that's when the tokens will be on uh tradable markets. Uh, but right now it's all testnet. And if you're a researcher, uh please contact us. We can get you um some dev keys and we can get you up to speed. And uh we would love to collaborate on your research and introduce you to the rest of our community and make sure that you have access to these machines that are notoriously difficult to get access to unless you've got hundreds of thousands of dollars in budget.

SPEAKER_04 36:58

I think that was very well said. Um, and I think we should probably leave it there. We'll definitely be talking to you guys in the future if you're if this wasn't too excruciating of an experience. Um, but uh but thanks for joining us.

SPEAKER_01 37:09

An absolute pleasure. So uh would be more than happy to come back and thank you for having us on. Yeah, thank you very much.

SPEAKER_04 37:15

It's been a great talk. Thank you. Thanks, everybody. We'll see you next time.

SPEAKER_00 37:18

Goodbye.

SPEAKER_04 37:19

Bye, everyone.

SPEAKER_00 37:22

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.