Yannik Schrade, co-founder and CEO of Arcium, welcome. Thanks for having me, Peter. I'm very excited. I've wanted to get more and more Solana ecosystem projects on. And so I'm grateful that you and the team have taken the time to do this. Before we get into Arcium, I have a lot of questions. I think you guys have a very interesting product that I think meets a market need. We'd love to hear about you. How'd you get into crypto? And specifically, what led you into the Solana ecosystem versus building on any other ecosystem? Yeah, excellent question. So within our team, mean, nowadays we have a lot of PhDs who got into crypto basically very late, who have really been just into cryptographic and AI research. But for us founders, I'm actually the one who got into crypto the latest, I guess. So my first touching point with crypto was when After high school, I initially studied law. And then during studying law, I met a lawyer who was into crypto. yeah, he had a presentation about Ethereum. So we were building some smart contracts there, sort of a legal tech implementation. And that was in the time where, I don't know, this notion of code as law was more prominent. So it sort of made sense to say, hey, every contract will be a smart contract. So that's when I first got into touch with crypto. And then really getting deep into crypto was when founding Arcium. And I met most of my co-founders during studying mathematics and computer science at university. And we met through our shared interest in cryptography. And yeah, we were at the local blockchain club in Munich. and decided to build something new about privacy using new kinds of cryptography. So that's really how I got into crypto. Not from a trading perspective or financial perspective, but really wanting to build something with cryptography. And the way that these new types of protocols that allow for privacy function is that they really go hand in hand with decentralization and distributed ledgers. So it was the logical fit to build something in crypto. And yeah, that's how I got into crypto. What was your second question? How come you, of all the ecosystems that you could have built, Arcium, what led you to choose the Solana ecosystem? Yeah, so it actually really was the first ecosystem that I personally got into besides this dabbling with some Solidity contracts in the past. And what we really liked about it is, at that time, this narrative of composability was even bigger in Solana, where you said, hey, this is this state machine where you can have composable smart contracts. You can have composable programs. all those different applications work together in a composable way. And that was entirely different from what you were used to in the Ethereum ecosystem. So that's one of the key aspects that made us choose Solana because it allowed us to add privacy. And the kind of privacy that we've been building with Arqium has greatly evolved over time. But even when starting, we always had this idea of having composable privacy where you have public ledgers with public state. And then at the same time, you're able to have some private things on top of the blockchain. And we wanted that to be composable. So you didn't have to, for example, go to some dedicated network that offers privacy as a big use case, but instead say, hey, this is a fast distributed ledger where people are building amazing applications. It would be perfect if those applications then had access to composable privacy. So that was one of the key aspects why we chose Solana. And on top of that, I think from a mindset, we are very closely aligned with Solana, with the IBRL mindset, increase bandwidth, reduce latency, being able to make things faster, more practical, and actually possible for businesses and people to use. And I think on that layer, also aligned very nicely. And very early on when we started, really had the chance to meet all of the Solana core folks from Tollee over all of the other founders and Solana Labs folks back in the day when they were running those Solana hacker houses. So those week long, I guess, where developers were locked in and building their applications. So that was extremely powerful. Being able to work this closely with the core devs and iterate fast over ideas and design decisions was extremely valuable. So that really made us a core part of the Solana ecosystem. You bring something up that I think is very impressive in how Solana, the foundation, does developer marketing. They have these Solana hacker houses where they get as close to the developer as possible. The developers always have a place to go, they work with each other, and they grow together, and they build together, and they iterate quick. I think that is really fascinating and something that I think the Ethereum ecosystem has done in the past, but I don't think they've done it as intentional as the Solana ecosystem has. Yeah, and for example, this year in February, we were sponsoring and co-organizing MountainDao, which was a month-long hacker house, if you will, in Salt Lake City, where, again, all of the big famous folks flew in and gave hand-on advice to developers. There's developers who started building some application one month ago. Now they get hands-on advice from those big founders. I think that's incredibly valuable. And this direct space where you can have face-to-face discussions, yeah, that has been extremely valuable. So nowadays, it has become a bit more decentralized. There's more and more, I guess, community-organized spots popping up because the foundation itself actually has stopped running the hacker houses. But Mountain DAO, think is a great example where twice a year there's those one month long retreats. Yeah, I agree. Let's get into Arcium. Maybe if you could describe Arcium in two ways. Maybe the first way, if you were speaking to someone not technical, how would you describe Arcium? And then maybe to a developer who will be potentially a builder on Arcium, how would you describe Arcium to them? Yeah, so for a non-technical person, and the way we frame Arcium is as the encrypted supercomputer, which gives you a very top-level idea of what we are doing. We are performing computing, and it's encrypted. So basically, I would say you have a black box, a very secure black box that nobody can attack. Nobody can. and peek into that is able to run any kind of computer program in a fully encrypted and secure way. That's what we're building. And for more technical developer explanation, what that means is that we actually don't build some big supercomputer in our basement and run the computations there, because that would, I guess, be a single point of failure, right? Instead, we're making use of our distributed network. And this distributed network run so-called secure multi-party computation, which is a kind of cryptographic protocol that allows for the private and verifiable processing of computations. And so by instantiating these cryptographic protocols that we've developed in our distributed open network, anyone can run any computation in a fully encrypted way, and in fully encrypted, fully private way. And that's so powerful. because the status quo, both in crypto and in traditional computing, really is that your data can exist in multiple phases. So also there's this phase where your data sits statically on some server. Most of the data that sits statically on some server is actually encrypted. If you have some user credentials when you store something in iCloud, most of that stuff usually is actually encrypted and stored in a way there. So it's only accessible to you in the ideal case. The problem with that is that when you have to process that data, when you have to process that stuff in the cloud, it has to be decrypted. I mean, that's a very easy concept to grasp, something that is encrypted and you don't know what it is, how should you process that? So that's the big problem that we're solving that we, with Arcium, enable you to process this stuff, even though it's fully encrypted. So that's the big powerful shift. And that's what we are calling privacy 2.0, where privacy 1.0 is limited. And it's really just, I have some encrypted stuff. I can store it. And I can maybe share it. But with Arcium, you don't need to share it. You're able to run computations over encrypted stuff. And we can collectively run computations over all of our encrypted stuff without having to share it. yet we are able to get the outputs of these kinds of computer programs. So is it correct to say that the problem that Arcium solves is you're able to run computation on encrypted data without having to decrypt the data and then share that data with parties? Is that right? Yes, yeah. So you're able to, we like to call it, run computations over encrypted shared state. So I think a good example is if you're a hospital, you're sitting on a lot of very sensitive patient data, right? And there's a lot of hospitals. And all of this data sits in isolated data silos that, for very good reasons, cannot be shared. And so With Arqium leveraging Arqium's technology, what all of these hospitals are able to is they can encrypt their data and, for example, train an AI model on top of that data without anyone ever seeing the data of any of the other parties. And at the end, everyone can have access to this model. But what's also possible, and I think that's also can be quite mind blowing, is that after everybody has their sensitive encrypted data and that is being merged into this model that is being trained, the model itself can also be encrypted. So nobody sees the model, right? And then some doctor who has some information about a patient and wants to predict the risk of some disease, they simply encrypt this patient's information and run an inference against the encrypted model. And they get back a result only they can see. And now they know, OK, the risk is 70 % of heart disease without them seeing any of the data, without them seeing the model, and them getting the output back. So it's possible to build these entirely end-to-end encrypted systems where it's computations running all the time in a fully encrypted way. No, that's very interesting. I've heard that similar example with healthcare. And the example that I've heard is, you know, some projects use trusted execution environments to house the data. And you need a private key to access that data, but that data is still, it's encrypted in this T and it's viewable by those who have the private key to it. And it sounds like, and with every, you know, a kind of encryption method, there's different trade-offs. And Arcium uses primarily multi-party computation. What trade-offs are there between Tees, MPC, and potentially other methods also, and why you guys went with MPC? Yeah. So one of the big issues with trusted execution environments already lies in the name trusted execution environment. So my personal mission really is to eliminate single points of failure. And eliminating single points of failure usually means we cannot have some trusted party. And in the case of trusted execution environments, there's a set of trusted parties which you have to trust. in order to gain privacy. And so for the listeners, what a trusted execution environment is, it's basically a dedicated computer chip that comes with some privacy and security promises from the manufacturer. And how those chips function, because you hand out a chip to some random guy, and then the promise basically is that, hey, you can just encrypt data, run it on my chip that I have in my basement, and I can't see the data. That's the promise. And so how Intel and AMD, for example, achieve that is they have their proprietary manufacturing supply line. And within their factories, they're generating decryption keys. And those decryption keys then are being fused into the chip. And so what happens with a trusted execution environment is you encrypt some data. send it to the trusted execution environment. Before you do that, you basically verify that the trusted execution environment hasn't been tampered with. And then you send the data. The trusted execution environment decrypts the data, processes it, and sends it back. That would be the ideal case. The problem really is that we've seen a myriad of exploits with these kinds of systems. So basically, every other week, There's some new big exploit that is being found. And especially in the blockchain setting, there have been blockchains entirely relying on trusted execution environments. And we've seen them fail fatally, whereas every data had been exposed because there was some relatively simple hack when you had access to the hardware, for example. And so a few months ago, we've seen a hack that cost$10 to perform with AMD's trusted execution environments, where again, you're able to exfiltrate all of this information. And so your trust really relies on whoever is running that environment, which in a distributed setting, you don't know who's running that. And on top of that, you have to trust some corporation that manufactures that stuff that they actually make it secure. And for example, there. key generation factory isn't leaked, the attestation process, which is used to verify that the untempered nature of such a device isn't infiltrated. And so it's really this linear supply chain, which is a chain of single points of failure. And it's relatively straightforward to attack those. in traditional chip manufacturing, We've seen a lot of these exploits as well, where, for example, someone had access to a signing key to upgrade the firmware of a chip and then being able to attack millions of devices. So that's why I see that as extremely dangerous to rely on these systems, because also all or most of the performance improvements that these systems have implemented introduce many more potential flaws. And with secure multi-party computation, what we are trying to do is to not have to trust anyone. We are building a system where you don't have to trust Intel. And importantly, we building a system that is future-proof. So what I don't want to see is to have some network that relies on some proprietary manufactured hardware, which in a few years, will be discontinued because something new arises. What I instead want is a cryptographic system that can run forever. And that's what we building. We don't need to have trust assumptions, and these systems can run till eternity in a secure and private way. And so how MPC functions is not on the hardware level, but instead on the cryptographic software level, where we make use of distribution and encryption. And so the only trust assumption that you need to have with an Arcium is that there's one single honest actor. So we've designed a new protocol that makes use of multi-party computation. And multi-party computation as a term already communicates this notion of, there's multiple parties involved. And so many of these systems that have been implemented in the past run on the so-called honest majority trust model, which is similar to what a lot of blockchains run on, which is more than half of participants have to behave honestly in order to provide privacy, for example. With Arcium, that's not the case. With Arcium, we've strived for the lowest possible trust assumption, which is only one single honest participant has to be honest. And that's easily achievable if you're permissionless. and makes the system way more scalable because you can operate with smaller sets of participants running these computations. And anyone can achieve a relative trust assumption of 0 because you wouldn't know, basically, if you're dishonest together with all of the other parties or not. So that's making the system incredibly nice, not relying on these trust assumptions against Intel, which is in my opinion, actually insane to be selling this as privacy, especially in the distributed setting. think trusted execution environments can work nicely together with MPC to even further reduce trust. But TEs, the way that they're framed nowadays, are incredibly dangerous and really are single points of failure that mainly work in an environment where you're operating with a trusted party that you have a bilateral contract with. If you're signing some contract with AWS, you should be able to some degree to trust them. However, you still don't know if there's some backdoor to those devices, and there can all the time be some new vulnerability or exploit. Yeah. No, I thank you for sharing the trade-offs between trusted execution environments and MPC. And I agree with you. You whenever you introduce hardware requirements into an algorithm, it just increases complexity. And whenever you increase complexity, the potential for failure just goes up. so those trade-offs really matter. Let's talk about the, you mentioned earlier about the healthcare example. So I take it that one of the target customers or target users for Arcium will be businesses that need some level of protection for their data using some level of encryption. What about other target customers that would use Arcium? Yeah, so what's interesting to me, and that's actually what excites me most about the kind of technology that we're building with Arcium, is that we're catering to a lot of different kinds of customers. So the first type of customer we're catering to is dApps on the blockchain. So that can be dark pools, so enabling private trading. on the blockchain, which in turn will both enable privacy and more power to individuals, but also will allow for institutionals to be onboarded onto the blockchain. Because nowadays, more than 60 % of the daily US spot volume happens off exchange within trusted dark pools. And so on the blockchain with Arcium, you can build a trustless dark pool, which will bring way more adoption to blockchains. At the same time, it enables you to have private stablecoin payments. And all these core financial and DeFi applications can be made private, at the same time being able to add compliance. think that's also very interesting. The developers using Arcium can, for example, simply specify, OK, this is encrypted state. And now I want this encrypted data to be visible to some specific party. Developers don't have to learn any new cryptography for that. They simply write their normal computer program, the normal smart contract, and specify this should be encrypted. This should be visible to party X. This comes from party A, something like that. And so that's the most straightforward applications, I think, for on-chain stuff. But what we've realized, basically, that there's not a single application. that doesn't benefit from having some level of privacy and confidentiality. So we have teams building with us on-chain machine learning marketplaces where people can provide their proprietary models and entire workflows. So they have some Python code with some models in them, and they want to monetize those models. But they can't just post them on the blockchain, because then I'm just going to steal the model and run it locally on my machine and won't have to pay anything to them. So they need those models to remain proprietary. At the same time, whoever wants to use those models doesn't want to send them their data they want to run these computations on. And at the same time, they want the output to be verifiable. And so we're solving for that. So with Arcium, it becomes possible to simply wrap all of that in a smart contract where someone can provide this proprietary encrypted model. Someone else pays for using that model with a fee and provides their encrypted data. Arcium combines the both, runs a computation without anyone ever seeing the data or the model. And at the end, only the party providing the data is able to decrypt the output. At the same time, whoever provided the model will get the economic benefit from that. And so with a similar setup, it becomes possible to also monetize data in a decentralized way, where you can then aggregate a lot of encrypted data without having to share that, train models, keep those models encrypted to make them proprietary, and then whoever uses those models pays fees, which then can be distributed back to the people providing the original data. So I think that's extremely interesting. At the same time, besides all of these web-free applications, and there's millions of applications, I guess, that can only be enabled by having this kind of encrypted computing power to them, where you need to store some private information on chain. We are also enabling traditional companies to use our technology, not just companies, but also governments and public organizations. And so what we did last year, is we acquired our largest Web2 competitor. So they've been around for more than eight years building private AI for threat fight, defense, and health care. And we've acquired a technology and core team and merged that into Arqium. So we're offering all of this technology within Arqium to be able to run these AI models, train them, all of that. And that's highly interesting to health care. to Threadfy, to Defense, and normal businesses. Because it enables also a new thought model you can think in as a company. Because you're able to make use of your data and maybe the data of your competitor without each having to share their data, yet being able to get some collective benefit from it. So I think that's very interesting. You mentioned about the acquisition of, which I find very fascinating because you don't hear a lot of restructuring or acquisitions in the Web3 space. Tell us about the thinking that went behind, know, growing through acquisition and what was that experience like and how's the integration going with the Infer team and the... embedding the new technology that you acquired into the Arcium stack. How's that going? Yeah, it's going great. So we today shared our paper about explainable privacy preserving AI for tree ensemble models, which is very exciting and really has been the work of those people that we've onboarded from Infer into Arcium. And what I really found interesting is that all of these folks are extremely excited about building the encrypted supercomputer to build something open that anyone can use that runs entirely on this encryption. And before that, what they were doing is more of this SaaS thing where everything is proprietary and you manually onboard customers. But now what we've built together with this team is an entirely new compiler, which you can add to your smart contract or Python, which then and is able to turn your normal computer program into an encrypted computer program. so it's really this open platform now, and they're extremely excited about that because a lot of them come out of the research space. Most of them are PhDs who've published a lot of papers about MPC, cryptography, and their intersection with AI. So I think this openness is very powerful and that excites them a lot. And so we've onboarded them without them having basically any prior crypto experience and successfully onboarded them to be excited about it. So I think that's extremely valuable and it makes simply more sense to run this kind of cryptographic secure computing in a distributed open setting rather than being this closed source company. So I think it was a logical step to acquire them and integrate their technology. Because especially for decentralized AI, decentralized AI won't function without privacy preserving computing. Because I think it's a strange idea to think about training models on highly sensitive, valuable data and having all of those models be public. Because. what are the economic incentives, right? So by adding privacy, you can add new economic incentives without requiring on some open AI, which isn't open, to store this model and own this model. Now everybody can own the model without having this direct access to it. So I think that's very interesting also when thinking about AGI down the line. But yeah, in general, it has been extremely exciting to be one of the Web3 companies acquiring a Web2 competitor. I think that's really fascinating and a good use case for how we can grow the Web3 space. And acquisition is not something that I've seen done very much. So thank you for sharing a little bit about that. Tell us about, now Arcium is built on the Solana layer one, but is Arcium, is it chain agnostic meaning that it's services and cryptography services and offerings? Can that be used by any application on any chain? Yes. Solana really is the first network that we're supporting. And it's our home, I guess. But the way you can think of it really is Arcium is its own network. It's distributed nodes that get partitioned into computation clusters. And those computation clusters then run those encrypted computations. But Arcium on its own is a stateless network. So You basically have the blockchain and from the blockchain it calls Arcium and tells Arcium, hey, this is encrypted data. Please run the computation for me. Arcium does that and settles it back. And so each individual blockchain, the first one being Solana functions as coordination, state, consensus layer, which is extremely nice because then the Arcium network is really fine tuned for its best application, which is just running confidential computations. I see. You shared a few use cases. One was a trading use case of dark pools and maintaining or preserving privacy on your trading strategies. You also shared a healthcare use case. And then another use case you shared was this idea around private AI, running AI models on your private data. I'm seeing that use case more and more from other privacy preserving projects. Tell us about that and how Because that requires a lot of imagination, you know, for me to think about all my private data and like what AI can do with that data. Tell us more about that use case because it's a very fascinating one. Yes. So really, I think there's this new notion, actually, which is quite nice, of individuals and businesses becoming more interested in privacy by using chat GPT and others, when randomly you ask chat GPT some question. And then it tells you something about what you asked it one year ago. And you're like, what's happening right here. I thought we just had a new conversation. So it's remembering everything. That's OpenAI using all of that to enhance their models even further. And there's way more data. think Firefox last week updated their terms of service where they say, we can monetize your data however we like, which I guess Makes sense, right? But at the same time, I, as an individual, would like to have control over it. At the same time, I I'm in the privacy space, But I'm not saying that it shouldn't be the way that companies can use the data or we can train AI that is more powerful on data. But it should be done in a way where the data is actually private, where there's no risk of that data being leaked. And without... detected that we are building at Arceum, it's simply not possible to do that in a trustless way. And now it's possible. Now you can have your most sensitive data, right? So imagine you have some health tracker, your entire history on your phone, your patient file, your family history, everything. You can use all of that of every single human being and train some models on that, and then make those models more powerful and never expose any single bit of information, never have anyone besides you yourself in control of the data. And for a business, that's important as well, right? For a business, all of your sensitive business data should remain proprietary to you. And that's one of the big constraints, actually, that businesses face when trying to use chat GPT and others, because they have to send all of their sensitive data to these. So it might actually be a more pressing issue for businesses than individuals in this point in time. So that's, think, what excites me about it. Being able to cater to basically anyone, to individuals and sets of individuals, I guess, in the form of companies, but also governments, right? The defense base, right? Being able to securely monitor things. I think it's also interesting to think about surveillance. I think we will always have surveillance in our societies, but wouldn't it be nice if we could make this kind of surveillance verifiable and in a way where the video data that is being recorded is never shared with anyone and only if a face of some criminal that the society has deemed to be a criminal. is identified, then that will be shared. And anyone can verify that. And anyone knows who is deemed as a criminal. So I think we can add a lot of verifiability and privacy to our systems and, in the same time, make them more powerful. And that's really this. I think the whole point about privacy 2.0 is making systems more powerful through this kind of encrypted computing. And that's one of the. the aspects that are lacking by what I call privacy 1.0 providers, where it's really just tedious to use this kind of privacy. The user, the developer, every party involved has to go out of their way. And instead, we're saying, OK, we add composable privacy to any application. Yeah, yeah, speaking of trade-offs, there's always a trade-off between privacy, safety, convenience. The more private something is, the less convenient it is typically, but more safe. it runs along those three different variables. And so making it easy for the user to have privacy and it's also convenient, I think is a... is a big value proposition definitely for developers and for app developers and for businesses. And when you look at human behavior, it makes sense why people use the same password for a million different things. It's because it's easy to remember. And so people really tend to prefer convenience and ease. But if they can have security and it's easy, then it's even better. 100%, right? you don't want to, I mean, I think it's entirely wrong to also be making privacy a political issue. I think that has been tried in the past and it has failed, Where you're the guy standing on the street with design, hey, we need to have privacy, right? I think that's entirely wrong approach. I think every human should have privacy. think there's a right to privacy. But I think the solution to that is actually to make the systems better, to improve UX, remove frictions, and at the same time, allow for more powerful applications than just, hey, your application is now private, right? To allow for things like, now you can train models over data you could have never used. So tell us about, you shared with us that you went to Mountain Dow last week or two weeks ago and then ETH Denver, believe also. And so the Arcium team, you're doing quite a bit in terms of developer marketing and getting more developers to build on and use Arcium technologies. Tell us about that. How is that going? And also tell us about onboarding community members that might not be technical. Because with every project in the Web3 space, it's important to, you know, cater to also community members that are not technical and giving a place and making sure that they feel like they belong. How's that going? So I think one of my biggest personal achievements is coming up with gmpc, which is how we greet ourselves in the Arcium community. So spreading the word about npc. But actually, my idea is that nobody should have to know about npc. Nobody should have to know about CK or any of these types of things. Instead, I want all of that to be abstracted away. I really want them to just grasp this concept of what's possible and then build things with that. And so that's why we're spending so much time on developer tooling, where we are abstracting all of these things away. And in the past, what Teams did was really introduce some new paradigms, some new programming languages, whatever, which has always failed because you don't want to do that. You don't. want the developer to learn a new type of programming, where you now program something that is private, that is encrypted. The developer has to learn about encryption schemes and decryption schemes and how to manage all of that. All of that is abstracted away. So I think that's one of the key hurdles of onboarding new developers, taking all of these difficulties away and simply enabling them to do what they can do best, which is build new applications, and I guess be creative, right? Because all of these tedious things now can go away. And what's super exciting is really seeing all different kinds of teams building with us. Small one-person teams coming up with novel use cases, like for example, building a DocuSign on chain where you can have encrypted data directly on chain. So encrypted documents on chain. And what he's building right now is you upload those documents on Solana in an encrypted way. And then you can give access control and say, hey, Peter, you should sign the document. So you can also decrypt the document and read it. And you can sign it. And all can verify, hey, it's on the blockchain. At the same time, what's possible is I can say, hey, my lawyer should have access to page three of this document, right? So information reduction. We can redact specific information and post it publicly. So all of that is possible with encrypted computing. And it's one of the most simple use cases, right? It's super straightforward. So I'm really excited about what he's building with Solsign, for example. At the same time, we have large teams building with us. Or we've started working with some big web 2 players, which hopefully we'll announce soon. So yeah, I guess what we are working with is trying to handle all of this demand. That's exciting. SoulSign sounds fascinating. Having used DocuSign and PandaDocs and all these other web 2 signing projects, it always baffles my mind. Why is there not a web 3 version of this yet? I think if there was a good legitimate one, it would get so much usage. I mean, that's all I would use. Yes, I think so too for me as well. You have privacy. You have it run on some better payment rails as well. And there's this newly introduced, I think, cryptographic chain and information reduction. And you can play around with a lot of cool things. And I think down the line, you can combine that with even more. At some point, you can have your entire payroll on chain, which also gets executed privately based on the contract you've signed, which is stored in an encrypted way on chain. But some encrypted ML model is reading that. then so I think there's a lot of cool things. Yeah. Yeah. Tell us about the Arcium roadmap. One thing that I like that you are doing is you have three phases and you're calling it Arc 1, Arc 2, and Arc 3. And it's like your roadmap to mainnet. And I like that you're using, you know, it's part of the name, so now it's memorable, you know, and it's repeatable. And whenever you hear something repeated over and over again, it starts to create in your mind, you remember it. And so Arcium. Arc 1, Arc 2, Arc 3. Tell us about the roadmap. Where is Arceum and what people can expect in terms of Maynet lunch? Yes, so we are going into a public test net very, very soon. So we'll announce the precise date very soon. And at that point, anyone can start building their applications. And the funny thing actually is that you can actually already start building production applications with that, because if you solely rely on, for example, doing private AI training. You can do that in Testnet as good as you can do in Mainnet, but you don't have to pay anyone for it, I guess. So that's extremely exciting for us. And after public Testnet, we'll be moving into Mainnet. So it's actually a very, very dense timeline. The team is. working very hard, doing a lot of class, doing a lot of cryptographic work all the time, optimizing things, making things faster, making things easier. And so it's a pleasure to be onboarding more developers and working with them and then really having those, yeah, constructing this feedback loop of being able to then iterate even more and fine tune everything for the needs of developers. So that's the phase we're in right Is there for the public test net, are there activities for non-developers to do so that they can be part of it also? Yes, we have a few cool things planned. Okay, well that's exciting. I'm part of a test net for another, for a ZK project and they're doing some pretty interesting things for people that are not developers, but is helping to grow their community. And so if there's anything I could recommend, that's definitely something to do. make sure that there's some fun things for non-developers to do. You can massively grow your community that way. Yeah, 100%. one other interesting thing that Arcium did. the acquisition is one interesting thing from a kind of a market perspective, but also Arcium did a rebrand recently, or I I'm not exactly sure when, when the rebrand happened, but I'd love to hear about the thinking behind the rebrand because, uh, I've actually spoken to a couple of projects that have told me that they are thinking of rebranding, but they're scared to do it because of they don't know how to do it. and all the work that's involved. But a rebrand is something that Arcium did. Tell us about the thinking behind that, like why you did that and how is that going for you? Yeah, we actually, some people might call it pivoted. I really like to call it evolved. So how we started was as elusive on Solana offering a new type of transactional privacy. That was our starting point to be able to transact privately on the blockchain, swap privately, right? These kinds of more simple applications. So these applications were simple enough for us to use zero knowledge proofs. What we designed back then was so-called decentralized compliance. So we wanted to build a privacy, private transaction system where anyone could remain safe at the same time, right? Because they wouldn't have to fear illicit actors. So the design we came up with back then was you have those zero-knowledge proof-based transactions, which is just me. signing something with my private data, submitting a zero-knowledge proof to the blockchain, and then transferring it privately. What happened with every single of these transactions was it created some encrypted message. And that encrypted message could be processed, again, by MPC-based confidential computing. So what's running as this black box computation is has the sender been added to the blacklist. And so it would just screen after the fact without seeing anything about what has happened in the transactions. But it was able to screen against a public blacklist. And the public blacklist, is something that decentralized consensus can come up with. So a Bybit hack happens. You propose, this address is the Bybit hacker. And that can also be partially, I guess, automated and then with zero knowledge proofs, can say, this address is directly related to the Bybit Hacker. And so they are also added to that. So a very nice architecture in that where you can enforce compliance over private transactions in a decentralized way. And the big powerful thing really is that a screening happens after the fact because there are systems like privacy pools, I think, that also came up when we started doing that, which enforce this kind of privacy when you deposit money. But at that point in time, nobody yet knows that it's the Bybit Hacker, right? So our system really had this benefit of allowing for that. But what we've realized at that point is how complex it is to build such an MPC protocol for screening those computations, right? So we wanted to make it as trustless as possible, which means this on this majority trust assumption. which I talked about earlier, lowest possible trust assumption. Super cool. At the same time, in practice, this honest majority always introduced some censorship mechanism, because a single bad player could normally DDoS these kinds of computations, which would have been bad. So we came up with a protocol that eliminated that. So a censorship resistance protocol that cryptographically identifies any kind of misbehavior and can punish the corresponding party. So after doing that, we also added verifiability, a lot of things. And what we realized as well when developing that is why are we only building this powerful technology for compliance reasons? Because it's way more powerful to be used for DefAI applications, for... machine learning for AI, right, for everything, even traditional industries. And so that's when we realized, OK, we should shift from privacy 1.0 to privacy 2.0. And we wanted to reflect that in our name and branding as well. This evolution from being a protocol on a blockchain to being a generalized computing network that allows for anything to be encrypted while everything can be computed. That makes sense. I appreciate you sharing that. I think the rebranding is a challenging thing for a lot of projects to be thinking about, but I'm hearing it more and more from projects that they do want to rebrand because they've evolved or pivoted somehow. so I think hearing from your experiences, and it makes sense. The rebrand makes sense. And there's other types of rebrands where there's kind of have to do it because of, know, there's some reputational issue that they were dealing with from the past. But in Arcium's case, was, you evolved, you you, you, you learn more about the problem, about the developers and the opportunities, and you added different technologies and offerings in your stack. And so you just needed something else that would represent you better. And Arcium is it. In the last few minutes that we have left, I'd love to hear about, you shared with us about the public test net that's coming up and that's quite exciting. In how should people be thinking about, you know, we've heard a lot about privacy and compliance and want to hear from you on your thoughts around, you know, decentralized privacy, decentralized compliance. Like, why should we be Why is that, why should that be like a really important thing to us, you know, for people that are in crypto and web three? Yeah. Yeah, Peter, what's your social security number? If you mind sharing. OK, perfect. I think that's Peter just shared an encrypted version of a social security number. So I think there's a lot of stuff that needs to be used when processing different things, giving you a credit scoring or whatever, which you wouldn't be willing to maybe share with every single human in the world. and for eternity because it's on a distributed public ledger. So I think there's very simple use cases where everybody realizes, hey, yeah, I want privacy, which is payments, for example, the most simple form of application, I guess, to be built on our blockchain. But what's important is that I want to be in a position, and that's why I'm building Arcium, is where the individual doesn't have to think about it. I don't want to have to convince the individual because that's a battle that I am not able to win because they will go the path of least resistance. And so privacy has to become something that makes stuff more powerful and has to be as straightforward to be used that people don't have to care about it. that it doesn't become something extraordinary, that every website uses HTTPS instead of HTTP, right? That's the point I want to reach with what we're building. And in general, why you should care about this kind of technology, I would say because it's incredibly valuable and one of the best applications of distributed systems. Being able to run this kind of black box computations with no trust involved is only possible by combining this cryptography with distributed networks. And so it's the perfect intersection for blockchains. And if you're into AI, right, I think what this enables those three things paired together is an entirely new form of intelligence. And so I think that's what makes this exciting. new, more powerful concepts to think about. Well, that's exciting indeed. And Janek Schrade from Arcium, thank you so much for taking the time to meet with us. Very excited for what Arcium has ahead of itself with the public testnet. And we'll share all of the URLs and links in the show notes so people can learn more about Arcium. Thank you so much. Thanks for having me, Peter. It was my pleasure.