[00:02:05] Is that fair to say, like, what are your thoughts as the person sitting at this organization?

[00:02:10] Paolo Ardoino: So definitely the stablecoin Tether is, has always been run as a lean operation. So the most important personnel in the stablecoin business is the reserve management and the compliance team that includes also the investigation team.

[00:02:28] So on one side, you want to make sure that you can properly manage the reserves. You need the most important thing that the stablecoin issuer needs to do is to give back the money to anyone that asked them. Right. So at any point in time, you get huge redemption, 20, 30, 40, 50, and so on percent, you need to be able to be paid that immediately.

[00:02:49] So that is why I think our finance and research management team is incredible. Starting from 2022, we remember we have been attacked [00:03:00] by fear tree and a few others publicly saying that we didn’t have reserves. They wanted to cause a bankrupt. We were able to process. redemptions in just two days and 20+ billion dollars in just 25 days of redemption. So that was 25 percent of our reserves.

[00:03:15] So there is basically almost no banking institution that was able to survive through even 10 percent redemption. So that is basically the team of a stablecoin is very early. And then for, because on one side you issue a token. And so there’s a bit of technology there, of course, but on the other side, you need to buy key bills and other liquid assets, mostly is gold and, a few bitcoins and it’s it right.

[00:03:39] So it’s very easy from the company perspective and the resource management. That is also over collateralized by 7 billion. So around 5 percent of the total reserve. So we are 105 percent collateralized. That’s very unique as well.

[00:03:53] Preston Pysh: And that’s not talking about, that’s not talking about the retained earnings of the business. So that you would effectively be even more [00:04:00] over collateralized than at, if you would include the retained earnings of the business.

[00:04:04] Paolo Ardoino: We would be another 20 billion over collateralized. So again, it’s a very unique situation. I think that in this world, everyone tries to do two things or copy two things.

[00:04:16] One is Tether, or the other one is MicroStrategy. And I think that the good thing about to Tether is that we portion of the earnings and the profits. We are keeping them in the stablecoin business to further collateralize the stablecoin. Now we are 105%, but we are very methodic in that exercise. And with the rest, we have been buying Bitcoin and investment, right? Companies like Rumble and others.

[00:04:39] So I think that the last latest at the station that they’re at around, we were taking 90, 000 Bitcoin and around four to five tons of gold. So I love the fact that on one side, we can, we are basically the most important company when it comes to the U.S. Dollar, but on the other side, we, with the profits we [00:05:00] get, we are investing in assets that are resisting to inflation much better way than the U.S. Dollar. That are gold and Bitcoin one more, especially this is actually the best assets in longterm to have.

[00:05:12] And the beauty again, is that on top of that, our user base have the luxury of being exposed to these two assets that whatever happens in the worst case scenarios of the work, they will appreciate in value against any other currency, if the world goes towards that, they might Republic like in the 1920, then Bitcoin and gold would go to the moon. And I think that having a little bit of a sprinkle of those in our reserves is something that makes that are much more appreciated than any other stablecoin in the world, because a, we are over collateralized, even if it goes to 0, basically, we would be pull your shares. Reserve as a stablecoin, but since the Bitcoin going to zero will be unlikely if on the other side, if fiat goes to [00:06:00] zero, our users have much better chances rather than anyone else, any other stablecoin holder.

[00:06:07] Preston Pysh: Whenever I look at the current situation, it reminds me of this Jeff Bezos quote that he says, your margin is my opportunity. And when we look at the SAP 121 repeal that just recently happened, I would think every bank on Wall Street is wanting to somehow play in this particular space and kind of take a bite at what you guys have demonstrated in style of what you’re doing.

[00:06:30] So how do you think about that competition moving forward? Do you agree with them with SAP 121 kind of playing an instrumental role in their ability to now participate and play in this space? How do you see some of that going forward?

[00:06:44] Paolo Ardoino: I think there is definitely will be increased competition, but I think that a few days ago, we had I had this speech, which title sounds very cocky, a bit obnoxious. The title is once Tether, once in [00:07:00] a century company.

[00:07:01] The reason why I gave that speech and I talk about Tether in that way is that people don’t realize that the success of USDT It’s not because we were first or step out. But Tether built and is currently keep building and investing in the biggest physical and digital distribution network in the world.

[00:07:23] So it’s not about creating a stablecoin. Everyone can create a stablecoin, every single bank. But the problem is that if you issue a stablecoin in the U.S. and you focus on its distribution in the U.S., that is, first of all, is meaningless, in my opinion, in the sense that in the U.S. You have already 15, 000 ways to pay people, like PayPal, Zelle, CashShop, bank wires, credit cards, debit cards, FedNow, whatever, ACH.

[00:07:51] There are so many ways to pay people. So the dollars in the U.S. are already digital. So it’s not like you’re just reinventing the wheel. You’re going [00:08:00] to sell a bucket of ice to an estimate. But if you go outside the U.S., then there are hundreds of millions or well, several billions of people. They are desperately trying to get on the dollar.

[00:08:12] They are buying dollar cash, dollar cash, but they would be more happy with the digital dollar. And that is basically what us it is about is about bringing 3 billion people that are called unbanked and never had a chance to have to be part of the financial system, not because there are bad people, just there are too poor of being of interest to the banks.

[00:08:34] Because if you are a bank, you need your every single account holder. To generate at least 100 per year in fees. And if your average salary in Haiti is 134 per day, you know, you cannot pay 100 yearly to a bank. So the bank will not accept you as a customer. Many of these people don’t even have a payslip.

[00:08:54] They are not rich. They struggle even with job opportunities. So. USDT is the [00:09:00] dollar for them is the only solution for them to be part of the financial system is a way for them to grow dollars, not under a pillow, but on a wallet and a digital wallet. And while in the poor countries, they have, they struggle with the financial inclusion.

[00:09:15] They have access to smartphones. So a smartphone. Can host a wallet and can have USDT and Bitcoin. That is the way. But see, this is what we have been building. So we have invested in common companies that are building thousands and thousands and, and, or own or leverage hundreds of thousands of chasks around the world from Africa, Central South America, Asia, you know, in the past, for the last 20 years, while we were sold.

[00:09:43] With the concept of globalization, any PC would have, imagine that you are a company and you’ve owned a distribution network, a physical distribution network with trust, you wanted to raise money and you would go to a PC, any PC would have told you, sell the physical [00:10:00] distribution network, sell the trust and rent them back, it’s a waste of money, but the world is going towards localization.

[00:10:06] The world is going towards a different place now and owning or being able to leverage physical and digital distribution networks is, will be the most important thing, not only to distribute stablecoins or Bitcoin, but also to distribute education or any other application, any single piece of technology we go through these networks.

[00:10:26] So we are not just building a physical and USDT, but also we are building a network that can sell or educate. So that is the fundamental difference between us and someone actually trying to copy our business model, sitting on a very nice office at the 40th floor in Wall Street.

[00:10:49] Preston Pysh: That is pretty fascinating. So what you’re really getting at here, if I could. Summarize some of this, and we’re going to get into a lot of this in more detail. This is your whole punch technology. That’s [00:11:00] peer to peer. That’s not relying on any type of server technology or any type of server in the middle to basically sensor that flow of communication.

[00:11:09] This is. While you guys are getting into AI and trying to do AI at the phone level, or the processing of that data, that private data is happening locally and not in some cloud that some parasite organization is sucking all that. And there’s many other things that you guys are diving into. Would you say that this is the all encompassing strategy that you’re referring to here that you’re talking about that the banks, the Wall Street banks are not going to be able to compete on, on a much grander level. Is that what you’re saying?

[00:11:38] Paolo Ardoino: Yeah, I’m seeing exactly that. And this is that if you, we recently made different investments or announcements, investments in rebel is a social media platform, but then we made investments in, um, and we have hired more of the best teams in artificial intentions. We have all punch as our telecommunication technology are [00:12:00] all that you couldn’t look at every single investment thinking.

[00:12:02] Oh, well, they’re just playing with different toys. But actually, if we look at more deeply at our strategy, what we are building, even. Just take the AI side. So AI and hedge devices. So the ability to run it and do inference on edge devices, imagine every single device can run AI agents that are fully local to the device.

[00:12:25] They will have their own identity through. Bitcoin 39, for example, anyway, Bitcoin identities. And then all these devices can talk to each other without any central server. So I using the whole bunch of technology and each one of these agents will have a wallet built by with WDK. That is our wallet development team.

[00:12:44] That is an open source library to build wallets. So I think that with Tether, we can build. The biggest AI agent network in the world that could consist in billions and billions of AI agents, all talking to each other, peer to [00:13:00] peer, all unstoppable, all able to deal with each other and appear to be a faction with peer to peer money.

[00:13:07] That will be a grandiose project. It’s very, very hard to pull off, but we are building the technology that goes into that direction. It’s a way to build AI that is for the people, empowers people, not an AI that tries to steal data from the people just for the benefit for a few companies. So I think that every one of us Should never GI shouldn’t have AI that can take on part of our tasks that are the most boring.

[00:13:33] This AI should have a wallet that is through Stordial, can interact with other AI platforms, all in a fully pure, pure, decentralized way. No one is thinking about that in these terms. For the very same reason that if you go to a VC and you tell them, Oh, well, I’m building an AI system where I don’t have a control.

[00:13:53] The VC will kick out you out from the office immediately because without control, you cannot make money. [00:14:00] But so you need to have incentives. You need to create in the eyes of the VC. You need to hold your customer by the balls in order to. Make a return on investment, but we tell her we are very not in a sense that the 13.

[00:14:14] 7 billion mark that you talk at beginning allows us to be a unique company. So on one side, we have the philosophy that is the Bitcoin philosophy. On the other side, so we have three things. We have philosophy, we have capital and we have tech vision. So if we had only philosophy, so part of the big creators and tech vision, and we didn’t have the money, it would be very impossible to do anything because VCs will create.

[00:14:42] Preston Pysh: You’re playing a much longer game. You’re playing a way longer game with a much bigger.

[00:14:45] Paolo Ardoino: Yeah, exactly. And we don’t have to create shortcuts or we don’t have to misalign incentives with the freedom so we can, our incentive is to create the most important and widespread [00:15:00] networking platform, peer to peer for AI agents and for people to communicate to each other and send money to each other to be resilient.

[00:15:10] To any sort of catastrophe. So a part of the privacy aspect is very important. You don’t want to give your most important intimate thoughts to open AI. The other most important topic is if something happens, something bad happens in the world, you want this system to continue to run. And that no technology today is it will make that happen unless you use you go fully peer to peer, which we are doing.

[00:15:34] So we might be able to make less money out of it, but it’s just the right thing to do. And we can do it just because we have the capital from the stablecoin business in order to build these crazy things.

[00:15:45] Preston Pysh: Wow. I was going to add, so I’m going to ask this question anyway, even though I think that you just explained why maybe this isn’t the case.

[00:15:53] So here’s the question. Do you feel like you’re at a slight disadvantage not being a publicly traded company because you can’t [00:16:00] access public markets and basically do what Michael Saylor’s doing by issuing more shares, transmuting that into Bitcoin, either in preferred convertible debt markets, or just common markets?

[00:16:11] I think I know what your answer is going to be for this, but I’m curious to hear it.

[00:16:14] Paolo Ardoino: Sino is a very, well, it has a very peculiar company, right? Um, strategy B now it’s called. Yeah. He can be a public company because his main role is executing the MicroStrategy play. So just through more Bitcoin.

[00:16:28] That’s fun. The software side of a side of the business is basically handle the side, right? So that is not core. Tether, on the other side, is still able, without issuing any debt, to buy a good amount of Bitcoin. I would call 90, 000 Bitcoin not a cheap change, but we feel that our mission as Tether is to build something that is complementary to Bitcoin.

[00:16:52] So if in life we have only Bitcoin, and we have Bitcoin that is held for the vast majority in public traded [00:17:00] companies or ETFs. Imagine like the worst case scenario would be 21 million Bitcoin held into public traded companies and ETFs. How Bitcoin will look like at that moment when people need it the most?

[00:17:13] Will people be able to use Bitcoin if it’s all captured within public traded companies and ETFs? So of course Bitcoin will remain Bitcoin. But at the same time, people might not have access. If there is a third world war, how people can access to Bitcoin or use Bitcoin if it’s all captured out in a few private companies.

[00:17:34] And also, even if you have Bitcoin available and that scenario of Bitcoin being captured by a few private companies would not be realistic. Even if people had Bitcoin, but if people didn’t have the ability to talk to each other in a peer to peer way with the same precondition of Bitcoin, but for telecommunications, how the world would look like.

[00:17:55] So, if I have freedom of mind, that is Bitcoin, but I don’t have freedom of speech, [00:18:00] I don’t have freedom of communications. Are we really free even if we have Bitcoin? I don’t think so. So there is so much to build, especially now that Bitcoin is almost like the four phases of acceptance. We went through, as Bitcoiners, we went through, first no one cared about us, then they mocked us, and then they fought us, and then they accepted us.

[00:18:19] Fun. In 2034, they accepted us. Okay, now what? And what we learned from Bitcoin that we can apply to the rest of the world, or the rest of technology. That I think is the biggest achievement and strategy of Tether, is trying to apply the philosophy of Bitcoin and the tech vision of Bitcoin that is peer to peer money, but being peer to peer tech.

[00:18:41] Bring back internet as it was before. Bring, create a system that empowers people to be free and keep control over their own data rather than just keeping control of their own money as with Bitcoin. So I don’t want to be a public company because I don’t want to have to report every three months to a junior analyst that [00:19:00] will tell me, Oh, Paolo, but You know, you could have to make much more money if you have done that and that and that.

[00:19:06] No, I want to be able to be free to build the technology for the work that I see happening rather than having to squeeze every single last drop of profits from the company.

[00:19:18] I want to frame this for folks just so they can kind of understand where kind of the root of that question comes from. So, If we just look at the profits that were generated from the treasuries that Tether sitting on, it was about 7 billion last year, I’m taking out the gains that were attributed to bitcoins growth.

[00:19:35] So if we just look at those, which I would attribute to the quote unquote operations of Tether, which is about 7 billion. This is about 70 times larger than micro strategies profits, just to kind of put, and I should start calling it strategy, but the company makes about 100 million a year, and if they’re not, if Tethers knocking out 7 billion, that’s about 70 times more profit.

[00:19:56] That they can allocate and direct anywhere that they want. [00:20:00] And I think that that’s important framing because when we look at how much Bitcoin MicroStrategy or strategy has been able to accumulate over the past year, two years versus how much Bitcoin Tether has accumulated over that same period of time, even though their profits are 70 X, that of MicroStrategy, there’s a very large difference in the amount of Bitcoin that’s being accumulated by both of these companies.

[00:20:24] And so. I would say that they were able to do that simply because they were a publicly traded company. He’s able to tap into all these very financial engineering type methods in order to attract more Bitcoin onto his balance sheet. And that was kind of the root of the question. But I want to go to this idea that you’re building something that is way bigger than just tokenizing sovereign securities.

[00:20:49] And What the, I guess when we get to the essence of what it is, you’re doing, is there just a desire to build [00:21:00] freedom like technologies for the world? Is that the driving force of the company and the founders and the owners or did you? Start working on whole punch technology. And it’s like, oh, well, we can use it here and we can use it there.

[00:21:13] Talk to me about the essence behind this grander strategy that you guys are trying to perform.

[00:21:20] The world is going to be more and more predictable in the next decades. And I think that everything that we have in our hands when it comes to technology will fail us. So anything that you have on your smartphone today, if there is a catastrophe or there is a war, a global war, will stop working.

[00:21:37] Everything that will happen that you have in your phone will stop working. Why, as humanity, we built for the last, well, let’s say only 20 years or 30 years in technology, we built technology that will fail us in the moment we need it the most. How can we can be the most intelligent species in the world if We built something that the moment we needed the most, it would fail us.[00:22:00] [00:22:00] I mean, the reason why we did that is because in order to make money, you need to have a control over technology. And if you need to have a control over technology, you need to make it in a way that is centralized around you, whereas around the company. So the entire incentives for technology that should be a common good have been misaligned with actually the people.

[00:22:21] And we’re only aligned with the corporates. And so when I think about technology, this is me being a lover of sci fi. Always thinking about what would happen in the worst case. Now I’m like, I need to build something that they know that will be a legacy, even if that are dummies. So when I look at Tether in general, I think, what if tomorrow Tether dies, everything that we have been building will disappear?

[00:22:46] Well, that is my fundamental question, right? So will that happen? And when I look at Google, if tomorrow Google disappears, so the search engine will die, the email will die, the docs will die, and so on. But if we [00:23:00] build technologies like HolePunch, or peer to peer AI, and similar, If the company, our company disappears, humanity can still leverage that technology and keep using it because it’s open source.

[00:23:11] So I think what we are, with Tender, we exceeded our expectations in how much money we could make. Like in 2014, if you were asking me what would be the biggest number in market cap that the USDT would achieve, It was between 1 billion, but 1 billion was really, I would have already exceeded really multiple times.

[00:23:31] It was just roughly like 1 billion. But now we are 141 billion and counting. And so, and the amount of returns we made on that market cap is just insane. Up to 2022, we did make little money because the interest rates were low, but now interest rates are very high. So at some point, you have to look at yourself.

[00:23:51] You look at the company and you start thinking, okay, fine. I got, I’m in this position. I have an important, very important capital. I [00:24:00] know how to build technology. What if tomorrow, how I can make technology be resilient to me? Because when I say technology needs to be resilient, Should also be resilient to ourselves as a company, right?

[00:24:14] So even with Tether in the stablecoin business, we USDT, we create USDT that now yields around 7 million per year, right? Without doing much, but also we create a tenor goal that is very small compared to USDT, but their goal is successful would yield zero profits to them. The reason why we created Tether Gold, any VC would have told you, or any highly threaded company would never have done that.

[00:24:41] Because why a company should create another product that could, if successful, cannibalize its entire profit margin? The reason why we did that is simple. If someone has to do, every single company should build products that potentially will, could put out of business the company, because if they don’t do it, someone else [00:25:00] will do it.

[00:25:00] And that’s how we think about Tether. We should always challenge ourselves before challenging others when we build any piece of technology or any new product. And it’s for the best of humanity. Again, we don’t have yachts, we don’t have planes. Who cares? I mean, we are simple people. So we exceeded our expectations. We are going to use all our resources to build something that we will be proud of in the future.

[00:25:23] Preston Pysh: I want to get into this new announcement that you guys had about routing Tether on top of Lightning that happened recently in El Salvador. This, as a Bitcoiner, this is exciting. I’m curious how you think through the reliability of the Lightning Network, whether you think that this is going to take much of your market share of the issuance of Tether.

[00:25:44] Today, my understanding is that a majority of Tether issuance is happening on Solana and Tron. And a smaller portion is on aetherium, and now you’re going to start issuing tokens on the Lightning Network via the taproot asset protocol. Talk to us [00:26:00] about this decision, how you see this kind of playing out moving forward, mostly from an incentive standpoint.

[00:26:06] So when I’m thinking through the routing of this, I would think that it’s. Just as fast as those other networks, but cheaper from a cost from the user standpoint, like if I’m going to route Tether from a wallet that I have to yours over the Lightning Network, the fees associated with that, I would imagine are significantly less than those other networks. And so do you see that incentive driving further adoption or how do you see that kind of playing out?

[00:26:31] Paolo Ardoino: Network is in line the structure the core idea around black network is to create peer to peer channels and people and you have been hearing me saying peer to peer multiple times in this discussion so from a pure technological standpoint if you want to scale to the work.

[00:26:49] It’s always cute to make a technology that can serve 10 million people or 100 million people, but what if, can that really scale to 7 billion people and let’s say [00:27:00] 50 billion AI agents or machines in the next 10 years? So, because we shouldn’t only think about us as humans, if machine to machine payments are going to be fundamental and machines will do much more high payments than us.

[00:27:14] And so can any existing technology sustain 100 billion payments per day, or 1 trillion payments per day. I think the only way, from a pure technological perspective, I think there is no doubt it’s anyone that says differently. I think they have their incentives are wrong. The only way to do it is peer to peer.

[00:27:34] It’s creating channels among people. The problem of blockchain layers, I’m not even signing Bitcoin, but the problem with blockchain layers is that they have a single share state where every single node needs to know the entire state to make sure there is no double spending. That is incredibly inefficient.

[00:27:50] So even if you make the best layer one super fast, with a very short log time and so on, that will not eventually be enough, [00:28:00] because there is always more need, there will be always more need for through. Then you can say, well, then I will create a layer two or layer three, The actual intelligent way to solve things is to create a peer to peer challenge, where why someone else should receive the status update of, well, a transaction that I’m sending to you, Preston, why they should care.

[00:28:22] As long as we have a way for us, you and me, to trustless verify that I’m not cheating you and you don’t cheat me, why someone else should care. And the fact that someone else now is involved in looking at everything that the others are doing, Is what is limited because if you even take Solana or theory with all these faster chains.

[00:28:42] They still have a single share state and that is very limiting and the layer twos are single sphere state as well. So they will be limited and so on and so forth. And their trade off of these layer twos is that they just reduce their security and they will concentrate validators ended up with one single validator.

[00:28:59] [00:29:00] So completely against the entire premise of decentralization blockchains, the only way is going to appear. So what I think will happen is that USDT on Lightning. Will be for a very, very fundamental change can actually take over all the machine to machine payments if properly implemented. That is where you do need the most scalability. And on top of that, though is secured by the Bitcoin network. So on one side you have your channel that will use basically inputs and outputs will be on the Bitcoin network that is more secure network in the world. So even if there will be 1 trillion USDT on the Bitcoin network, it would be safe because the underly economics of Bitcoin will be enough to make it, to not create incentives for people to cheat the system and trying to double spend or create issues with the settling the channels, the like channels.

[00:29:54] So in my opinion is technologically is the right way to do it. Peer to peer [00:30:00] channels on top of the Bitcoin network. Now, the question is how soon and fast people will realize the necessity of moving to this layer. And so there will be a requirement of a lot of attention because in this moment, there is still little usage of blockchain technologies, but that will change in the next year. So it’s not evident yet, at least to normal people, it’s not evident yet that there is a niche or scaling issue that will become apparent in the future on standard blockchains. So it will need, creation will need patience to see that growing.

[00:30:34] Preston Pysh: That’s part of my frustration with a lot of the discussions that are happening in government bodies right now is I don’t think that they understand that point at all. And I’m just worried that they throw some of their weight behind some of these other protocols without really kind of understanding what sound engineering looks like 10 years from now.

[00:30:50] And those are going to be lessons of pain if they continue to lean into some of these other protocols, in my humble opinion. But yeah, that’s really exciting. Have you guys issued anything [00:31:00] on taproot asset yet? And if not, when are you guys looking to do that?

[00:31:03] Paolo Ardoino: So we’re still in exploratory mode, right? So the technology became available very recently. So we’ll need to do some very heavy metal testing and so support, but we took our decision as also as Bitfinex and Tenor, we are supporting another protocol called Kind of a similar ethos and and philosophy, just different type of implementation and then the preconditions. But we plan to support Bitcoin tech, keep investing Bitcoin tech.

[00:31:29] I believe that I would say that Tether and  Bitfinex are probably the two with the full growth funds. We are the three biggest investors in anything that is Bitcoin tech.

[00:31:41] Preston Pysh: Do you, with the recent acquisition there with rumble, where you guys have a controlling, I think what I read, it was about a seven or 800 million deal.

[00:31:48] And then you have a controlling interest in the company. Would you say that this is rolling? It’s not controlling. Okay. I’m sorry. Would you say that this relationship is for kind of implementing some of your ideas with [00:32:00] whole punch technology into what they already have in order to kind of leverage the social graph that they’ve already built out, or how would you describe that from a strategic sense, if you, if you don’t mind sharing how you guys think about it from that perspective.

[00:32:11] Paolo Ardoino: So the investment is interesting. So Rumble will not share Rumble wallet. Not many people know that Rumble has a very peculiar user base. They have 70 million active users. They have basically the entire wealth America, almost the entire wealth America they have through their channel. So they were able to sell over the last two years, 800 million worth of gold.

[00:32:38] Imagine a rumble wallet with Bitcoin could, what could do. So that is one on top of that, I think rumble cloud is a great product. And they build their own cloud solution because they were attracted by, for such a long time by the former establishment on be shut down. Remember when parlor was shut [00:33:00] down, they were threatened as well, but they survived because they had their own cloud solution, but there is a step further.

[00:33:06] Using whole batches of technology to create a resilient video distribution network is going to be a game changer, I think, for companies like Rumble. So there are many things that we could develop within Rumble together. We have both great technologies within the two companies. And so the opportunity is just insane.

[00:33:26] Just even for the locational part of the Rumble wallet, educating people that you could have a nice, simple wallet with Bitcoin and stablecoins inside without any altcoin or puppet apps. That would be a refreshing change, right? So that’s how we are tackling things. We want to use all these investments as a way to educate people, but also building better technology.

[00:33:50] Preston Pysh: All right, I want to try to cover AI and just for me to try to understand the comments that you had earlier because I find this really fascinating localized [00:34:00] AI processing and then somehow networked with all the other localized AI processing in order to have a bigger, better model. Is that kind of what you were saying earlier that we’re basically able to leverage these various localized AI’s in order to have a better collective response for harder questions? Is that, am I following you there with what you were saying earlier?

[00:34:23] Paolo Ardoino: Kind of. So you have to see this from multiple points of view. So first of all, deepy was quite interesting in the scenarios and entertained. The reason is that deep Seeq obviously is not the direct word and not saying, oh, everyone could have done that.

[00:34:38] So team behind Deepy is great, but the approach that they had is to, a technological engineer should be the obvious one is David and . So you don’t have one single model that does everything and is a gigantic monolithic model. But you split the model in multiple that are more refined and [00:35:00] more vertical.

[00:35:01] It’s the concept of modality versus modularity is not new. It’s something that is very, very obvious. So it’s just crazy that companies like open AI, the fact that open AI and the others didn’t do that yet, or in a proper way is just because they grew so much so fast. They prefer to brute force AI rather than actual do the right type of engineering and intellectual challenging work.

[00:35:25] To make a great and usable and sustainable for the future. They prefer to announce 500 billion dollars of investment in new data centers. So, think about the tree seed, right? So, how pure economic you can hire 1000 researchers. Pay 1, 000, 000 change, right? That would be 1, 000, 000 per year. And you have 1, 000 of the best researchers in the world.

[00:35:49] Or you can buy 15, 000 GPUs more with the same amount of money. What do you think is the best or the most intelligent strategy? Is trying to buy another [00:36:00] 15, 000 GPUs to try to brute force and just add a little bit of capacity or trying to solve the problem in a better way. Of course, you should try always to solve the problem in a better way.

[00:36:10] That is exactly what DeepSea did. Rather than announcing terabytes of data centers and millions of GPUs and so on. That is, it feels like really going back to the history of computing, like where you didn’t know how to solve a problem. You just added another server. I think I’m not sure if in the last interview that we had, we discussed about one of my preferred books is the Cathedral and Bazaar of Eric S. Raymond that explains is from the 90s book explains the problem of. Operating systems that are building monolithically as cathedrals, or operating systems and technology built in a more modular way as bazaars. Cathedrals are beautiful, but if you remove one pillar, the cathedral will fall down. [00:37:00] Instead, with bazaars, even if you bomb a bazaar, that will just move 50 meters away and will just reconstruct itself because it’s made by people.

[00:37:08] Bazaars are Built on peer to peer interaction, people interaction. But cathedrals are just, there is an architect and they cost 200 years to build a lot of money and you just need one pillar that’s safe for everything to fall down. And so, when DeepSeek was, or came out and people were screaming about this revolution, it is a revolution, but was the right way to approach technology and AI development from the get go.

[00:37:34] And so imagine if we could expand the concept of deep sea, not just on single or multipurpose models with the, but imagine instead of having few verticals, you have an almost infinite number of models with a peer to peer orchestration system without any central servers. And with proper incentives that, for example, imagine that you go to [00:38:00] a university, a chemistry university, where researchers are releasing research papers on new chemistry developments.

[00:38:09] The university could build a model that is super deeply trained in chemistry with all the most advanced chemistry techniques. And that to be part of this gigantic peer to peer network of information. That is what the Internet, actually to me that is Intel, is what the Internet should have been from the beginning.

[00:38:28] A very beautiful, cause, well, sparse amount of information that can be researched. that can be also sold, right? So you can have models that can be rented, or you can pay per query on a model. The orchestrator could take care of routing payments across this infinite number of AI agents to make sure that either everyone has the right incentives to have the best information always available within these models.

[00:38:56] But there is no way, even the amount of information that humanity creates, there [00:39:00] is no way to create one single model. It’s not sustainable. It will fail people eventually, but we believe that with our capital and with our technology being open source, we are going to make it the right way.

[00:39:12] Preston Pysh: It’s kind of interesting that he is taking this path, the 500 billion that you’re talking about, and the Davis, Microsoft said that they were good for 80 billion, and then what was it, three days later, this DeepSeek thing came out, and for people that aren’t intimately familiar with everything that we’re talking about, so DeepSeek Did this for tens of millions of dollars, came out with a model.

[00:39:34] It’s effectively comparable to GPT 4.0 as far as the performance goes. And they did it for pennies on the dollar compared to where OpenAI is at. What I really agree with you on Paolo is just this idea that in the fine tuning. That’s really where I think they’re going to be able to make the difference is by hiring more people that are going through, because in the fine tuning stages, so just so people understand AI, I’m going to break it down for you [00:40:00] the best I can in like 30 seconds.

[00:40:02] You take all the data, you curate the data, you then go through the development of what’s called a base model by compressing all this data. So imagine all the information on the Internet, you’re going to compress that down. You need GPUs to do this. And the chat GPT folks are saying, let’s just. Compress this, the data on the internet, even further by owning more and more GPUs to the tune of 500 billion worth of GPUs.

[00:40:24] But then after you compress all of that data down, then you go into what’s called the fine tuning phase of this, where you’re basically teaching it how to become an assistant. So you’ve got to take all that compression that happened on the internet. You then have to train it on another AI through this training, which has humans in the loop.

[00:40:41] People going through there saying, this is what a proper response should look like. You do that hundreds of thousands of times and then you apply that to this base model in order to come up with the AI and what Paul is getting at is he’s saying that fine tuning phase. Let’s say we want the model to be an expert in mathematics or an expert in [00:41:00] biology or an expert in you name it specialty.

[00:41:02] Spending and reinvesting in those areas to make the fine tuning and even the curation of the data in the first place. Like, why are we taking all of the internet’s data on communications, the large language models, and then applying it to the fine tuning of just biology. We should match all of the data, all the communications and biology to create a base model.

[00:41:23] Then we should do the fine tuning on biology with experts in the field. And we’re going to come up with a way better solution than some gigantic model that’s plowing all this. And Paolo, you’re way better versed on this than me, so please correct me if I’m saying anything out of turn here or not correctly, but I think that’s the gist of what you were getting at.

[00:41:41] Paolo Ardoino: Yes. And plus, what if tomorrow OpenAI disappears? What if basically we create a society where every single person relies on trajectory, and then something happens to open AI as society, we are going through a change also in the job market rely more and more on AI. But [00:42:00] if we rely on one or two companies to give us that AI, what if those two companies will fail?

[00:42:05] Society will disrupt because suddenly people need to go back to square one and will be almost left hanky because their skill set will change over time. The only way to make it help in society is to create resiliency in AI, making sure that models can run on a hedge device. Models, people should be able to train easily their own models.

[00:42:32] People should be able to fine tune their own models very easily. Their private information should remain theirs. That is also another important part. And that when we will have that peer to peer network of AI agents, doesn’t matter if companies come and go. People that rely on society, that rely on AI, will be able to continue to use AI no matter what.

[00:42:54] That’s, again, very important. We cannot leave the future of our society, [00:43:00] especially in such a very drastic change that will bring AI into the hands of a couple of companies. That will end up badly. We need to empower people, empower companies to retain control over their data. Even just think about data ownership.

[00:43:16] I saw an article with that was saying that meta, some employee meta took like 80 gigabytes or 80 terabytes of, I don’t remember, the unit to measure of documents and books from, from Torrent. Like in the future, people realize more and more that what they’re, the content that they’re creating will be just used and stolen by centralized AI companies to build.

[00:43:39] Better models that we only serve the purpose of making more money for such companies in the future, people realize that it is very important to build. If they will create content, they should own the content. And that content should be used to train their own models rather than someone else.

[00:43:56] Preston Pysh: Yeah. Help me understand just from a technical standpoint, if you’re doing [00:44:00] this localized, like the model that I’m downloading onto my phone, I think from a storage standpoint will be okay.

[00:44:05] But from a processing standpoint, from a GPU standpoint, for an advanced model, something that’s giving you really good intelligence and responses, I would think that that would be really computationally intensive for call it a smartphone. Why is that wrong?

[00:44:19] Paolo Ardoino: So there are different ways to look at this.

[00:44:21] First of all, if you want to have a model that knows everything, sure. It will be very difficult because then on a smartphone, when a hatch Captain Hatch smartphone might be able to run between three and seven billion parameters models, you cannot run a 700 billion parameters model. But I think that what will happen is that again, if you have is that more modernization in terms of knowledge, and also if you have access to a peer to peer.

[00:44:45] Infinite network of knowledge, then you will seek the information from who has the best and you can download specific models that will help you in that specific moment in time. So, it’s not that you will run the entire Internet in your poke, but you will have [00:45:00] control over it. That is most importantly, you will know what’s happening.

[00:45:04] The conversational part will remain because you can with 1 billion tokens, you can do a great, you can have a great generic conversational AI that, of course, will not have the entire. Intelligence of the world, but we’ll know where to find it. That’s really important.

[00:45:18] Preston Pysh: Do you find that the demand for GPUs at people’s homes to conduct a lot of that processing and then shoot it back over to the phone? Like, basically, you’re off boarding the processing to a localized GPU. I mean, you can do a lot of that through encryption. It doesn’t have to be at your house.

[00:45:34] Paolo Ardoino: Yeah, but keep in mind that the GPUs on phones are becoming better and better. Everyone. Now, if you compare an iPhone 10 with an iPhone 15, when it comes to GPU, it’s like 20, 30 times more powerful.

[00:45:47] And imagine what an iPhone will be now that everyone is investing in AI and hardware, an iPhone 20 would be insanely powerful. So that most of the tasks that you need. Can be done local, [00:46:00] so I think that we are going to be at a point where you can have a local AI that does fine tuning on all your emails and all your chats and become your best assistant.

[00:46:10] Now, the other option that I think will be very interesting, and we are looking into that as well, and our technology already enables out the box is to have. An AI box at home, like a sort of like Apple TV, but instead of focusing on TV, you have a GPU, a contained GPU from nVIDIA, but more powerful than what you would have on your phone that can offload all the tasks of processing of your family’s data, all the photos, tagging the photos of when you go on holiday with your children, all that.

[00:46:42] By the way, think about it. Right now you are using someone else cloud services to tag all the photos and categorize all the photos of your children’s handscan of scary. Yeah, so you should use your own hardware because your hardware is already capable to do so you should use backups that you own. [00:47:00] And then if you want a farther backup in cloud, you should do it in an encrypted way rather than giving all your clean and clear images to someone else. Again, you never know how that information will be used.

[00:47:12] Preston Pysh: Okay, real fast. So you guys are setting up shop in El Salvador. That’s where the company has been officially relocated to. I read that you guys are building a 70 story building in El Salvador and hiring. You’re doubling your head count at the company. Half of which will be people from El Salvador. Anything you want to add on here about moving the company to El Salvador?

[00:47:34] Paolo Ardoino: El Salvador is a great place. It’s a beautiful place. Every single time I go and I go there very frequently. I’m also a citizen now in El Salvador. I go there very frequently. Every single time the city is more advanced.

[00:47:48] It’s cleaner. There are cranes and buildings are happening everywhere. The food is better. Like every, there is investment across the entire society. It’s so fascinating [00:48:00] to see it live. Sometimes you hear like, Oh, Singapore developed over the last 30 years, 40 years. And back then you didn’t have smartphones to document all that.

[00:48:11] Sure. Some TV crews were there and taking like documentaries. But now we live in the digital era. You can see the change. You can feel the change at a daily basis almost. And the bouquet is the representation of what an illuminated politician should be. I, unfortunately, I’ve been living almost, almost all my life in countries where politicians were not that great.

[00:48:34] Everyone was promising things and never, no change was ever happening when I saw Katie. I saw, I thought, finally, there is someone that actually can hold the title of politician and look himself in the mirror a single day. So, I think it’s very important to support El Salvador in this transition because we’ll be an example for the rest of the world.

[00:48:55] Of course, Bitcoin adoption is great to have in the country, but more importantly, I [00:49:00] think the example that else would work in a receptive when it comes to the change and the hope that they can bring to other countries is just an incomparable.

[00:49:09] Preston Pysh: All right, Paolo, I want to stick to the time that I told you we were going to have the interview complete.

[00:49:15] I really appreciate your time. Anything else you want to leave with the audience or things that you want them to check out as we close things up?

[00:49:22] Paolo Ardoino: Just stay tuned. I think that there will be many more announcements in the couple of weeks or crazy things that we are developing. So just, just follow us.

[00:49:29] Preston Pysh: Well, I appreciate your time and love getting into the engineering stuff with you, as I’m sure people can tell through the conversation, but thank you for your time and coming on. I really appreciate it.

[00:49:41] Paolo Ardoino: Thank you for having me.

[00:49:43] Ottro: Thank you for listening to TIP. Make sure to follow Bitcoin Fundamentals on your favorite podcast app and never miss out on episodes.

[00:49:52] To access our show notes, transcripts or courses go to theinvestorspodcast.com. This show is for entertainment [00:50:00] purposes only before making any decision, consult a professional. This show is copyrighted by The Investor’s Podcast Network. Written permission must be granted before syndication or rebroadcasting.