80 Billion AI Agents Are Coming- And They'll Bank on Crypto | Yat Siu, Animoca Brands

Well, first when it comes to technology adoption and usage and generally let's call it the sort of, um, sort of, you know, tradfi and mainstream adoption, it's at record growth. I mean, you know, uh, whether it's Robinhood or Revolut or major banks, right? Um, you know, I mean, like everyone's in it, right? And you've got like Stripe, you know, with Tempo and so, you know, and many of the things that people were doing several years ago in blockchain and crypto is actually what many of these large scale institutions are now doing today, right? And they're learning from the past and they're partnering in some cases and sort of growing that space, right? And you see this in fact, today we just saw the announcement, right? Where Citadel invested into Crypto.com, right? So, another example where more institutional adoption and utility is coming into this. So, I would say that the whole sort of, you know, thing that we said about that '26 was going to be the year of utility I think very much has proven out, right? Um, I think what a lot of people sort of are disappointed about is that prices haven't sort of followed with that. And I would say there's a difference between utility and pricing in a sense that when utility reaches a large enough size and scale then the value will naturally accrue into it, right? That's normally the case. Um, but you know, when you also have a speculative period you have people essentially, you know, and this is always the case, right? Where, you know, people often sort of, you know, um, I think the famous expression is people often overestimate the future and underestimate essentially sort of, you know, um, sort of at the same time and they underestimate sort of the near-term effects of that future, right? So, it seems ironic for people in the industry to say, you know, why is Bitcoin lower than last year for instance, you know, and why is Ethereum lower and, you know, what's happening with Solana and shouldn't they all be rising when in fact the utility and usage particularly when it comes to TVL and transactions, you know, is larger than ever, right? And I think again, you know, this is the point where the speculative side versus the utility side, um, haven't yet caught up. Um, meaning that if the utility, um, was, you know, just like it was last year we would have the same impact, right? But it's growing and so if we believe, which we do, that the utility will continue to grow and grow and grow and grow then, you know, prices and the value of these assets would naturally follow. Um, not that much different from, you know, when you think about companies that have profits in the stock market when they start doing well then finally the, you know, price starts to appreciate, right? Um, even when you think of it as a startup people forget that companies like Facebook or Microsoft or Google they all were once small companies. You know, like, you know, in fact even Nvidia, you know, most valuable or second most valuable company in the world depending which day you look at, you know, was a tiny company in comparison to today, right? So, utility does take time but when it happens it's going to explode, right? So, I think we, you know, we need to distinguish between the two and if you're long-term bullish then, you know, you hold and if you're not then of course you're free to sell. And I think that's the beautiful thing about crypto. You know, like during, you know, when there's all these problems, you know, in Iran and US and, you know, all that kind of stuff, people are able to trade essentially tokenized oil and tokenized commodities while the stock market is essentially shut off and sleeping because they don't work over the weekend, right? So again, you can see, you know, it's more a reflection of a true market and that's developing faster and faster and we're seeing a lot more activity from tradfi in crypto. Um, and of course it does alter the shift in terms of tokens because while tokens are going to keep growing and there's going to be more utility for them, um, the relevance of the token is now competing with the entire asset class. So that means that for instance if gold has value or I don't know Samsung has value, right, or Nvidia has value and it happens to be tokenized that becomes additional competitive playing ground to your tokens which wasn't the case before, right? So you have a situation where suddenly attention is going there but in our book that's a positive because it also brings more people from the other side into it and exposes that. But you now have to tell and talk about why it's valuable which you can do, right? It was only a few years ago where you weren't even allowed to talk about utility, otherwise you would get basically, you know, slapped by certain federal agencies and be in trouble, right? So, again, we're sort of in this sort of in-between phase, shall we say. I wouldn't say it's reflective of true nature, reflective of the market, but this lag stuff always happens.

Fundamentally, fundamentally, absolutely better shape. I think it's just that, um, you know, there's a lot of sentiment and feelings and politics and the world is of course unstable, right? So there's a bunch of things that sort of add to it, but fundamentally, you know, we have, you know, the world's largest economy being bullish and supportive on crypto, right? That's about as positive as it gets. We have other countries around the world all trying to sort of figure out what to do in the space. Um, you know, stablecoin policies broadly being adopted more and more in other places, you know, crypto adoption, exchanges, you know, getting licensed. So, you can see a lot of progress that's happening already.

Absolutely.

I mean, like, give you, and it's not just your stocks, it's like yield for instance with NUVA which is our joint venture with Provenance. Um, you now have the ability to basically buy, basically, you know, a high value, you know, um, yield on basically, sort of, you know, with HELOCs and sort of mortgages. Um, and more importantly you can, uh, the yield is hourly, okay? And that's something you can do on chain. So, you know, if you want to put your money into the vault at NUVA and you want to withdraw it you can basically get your yield on an hourly basis. Now we know that blockchain can do that. But we know that, uh, that, uh, sort of you put it in a bank, you might have to lock it up for at least a week, maybe a month, right? You know, and this is not necessarily necessary in crypto. And once people start realizing that, hold on, this is just a better way in which I can manage my cash, uh, then obviously that's how they'll do it. Them or their agents, they'll do it on their behalf, right? Uh, because, you know, if you're not spending the money right now but you want it to be liquid, uh, then, you know, that's a perfect kind of product and you can extend that beyond and to everything.

Yeah. So I think, you know, when people talk about the metaverse they often have the image of the virtual goggles and that you're basically entering a virtual world. And, you know, we have to remember the framing as to why the construct of the metaverse became so big. You know, not only because Facebook renamed itself to Meta. That's obviously one of them. And by the way as much as, you know, they're sort of, you know, having cuts and so on, they still have the metaverse team. They're still focusing on it, right? So there's obviously shows that they are certainly sort of seeing something. Um, and I think, you know, the point is that we have to remember that this was born at a time when COVID was raging. And so, you know, we all thought that we're going to be spending more time online. We are spending a lot more time online. However, we don't exist purely online, should we say, right? Because people thought you're not going to travel. You're always going to be at home. You know, offices are dead. You know, people, you know, like the flights, airplanes are not going to be going on much anymore. All of that turned out not to be true, right? Um, but it came from the misconception that we are basically, that humans want to spend more time in a kind of hyper-real virtual environment almost like replacing the physical world with a virtual world, right? Uh, which is I think a great misunderstanding. The metaverse or the virtual world is an enhancement of a physical life. It is not meant to be a replacement for it because if it was truly such an excellent replacement then we would prefer to do that than to travel or to meet people in person, right? And it turns out that we are biological creatures and we are social and we do like to hang out with people and go to a restaurant and sort of, you know, do things together and play sports, right? We don't necessarily want to go somewhere sort of, um, remote all the time. And what is now and where we're so bullish about the metaverse is that it's because of AI agents. So AI agents today are affecting our real life, right? They're handling our calendar. My agents are, you know, handling my flights and they're booking stuff and they're handling restaurant bookings and now they're coding stuff for me and they're planning stuff for me and basically they're virtual beings. If you think of it, they are metaversal beings but they are impacting our physical life in better ways than before, right? So, the way that we like to express it to people is that, you know, the mistake is that we thought that we would spend more time in the metaverse when it turns out that the metaverse is coming to us.

Small, it's still small.

Still small. So, let's, I think the whole AI agent stuff really sort of became big with OpenClaw, right? And obviously, you know, OpenClaw and Hermes are very much the market leader at this moment in time. However, uh, when you look at the OpenRouter rankings, uh, our Hello Minds product, uh, is actually I think number four, number five. So, it's moving very, very fast, right? Um, so but what's the framing here? Let's talk about where it's going to go. So, you know, for you and your audience if you were to sort of estimate and think how many personal AI agents, forget even work, just personal AI agents do you think you're going to have in the next 3 to 5 years, right? And, you know, what would you say?

Right. Uh, and the common answer tends to be 3 to 5. If I'm in San Francisco, it's 230. Right. So, it kind of depends where you are, right? If you're more technical, you kind of want more agents. You see the value. And the point is that it's going to be more agents than humans. That's for sure, right? And if you take, you know, your number and you say, well, it's going to be, you know, 10. And we say that's the average for every person that's online, then we're at 70 to 80 billion agents already, right? And so those agents are going to be the ones roaming the internet and they're going to be the ones giving you information and they're going to be doing transactions for you and they're going to do all that stuff and the transaction system is going to be on chain, right? So you're going to be basically delegating much of that sort of stuff that you typically don't consider important to an agent and then later on they do more and more important stuff, right? And so the world is swarmed with these agents that are essentially handling your information on your behalf, creating value for you, right? And that's basically how the internet is going to be completely reshaped. So that means that if I'm building a product or a service and if I'm not agent readable, I become irrelevant. If the agent can't find you, then you can't be making business or commerce possibilities. So that means if I have an agent and you don't have an agent, maybe there's no way for them to connect with each other. And while that might sound, uh, while that sort of construct might sound wild, just, you know, imagine, you know, when you are searching today on either Google or whether you're using ChatGPT, right, you get a summary, right? From Gemini in the case of Google, when was the last time you clicked through to the website that Gemini suggested you click through, right? And just think about that, hardly ever, right? But two years ago you clicked through every website. So, this is a very subtle thing that you might not even notice, but just that subtle thing has reduced for many websites their traffic by up to 80%.

Right. And so the consequence of that of course is devastating if you are not ready, but great opportunity if you are because it also means that if you're AI readable and if the agent knows and finds you, then you get discovered. But then how do you make business? Well, you're not going to have ads anymore because people don't go to your website, which means you have to rely on microtransactions on the back. You have to rely on services. So, you could get a fee and the microtransactions will be a fraction and a fraction of a cent. But you can do this on chain transparently, which you can't do basically, you know, with traditional payment systems. Um, and also, you know, you don't have agency over it. With on chain, you do have agency over it, right? So, there's all these advantages that now come to bear, uh, because of that. But that's an example of how everything shifts and changes, right? Uh, and so we're going to be seeing every fabric of the internet is going to become agentified and that's kind of what we call Web4 with basically a Web which is autonomous with AI agents that essentially sort of do stuff, you know, and support us and frankly, you know, are our agents on our behalf.

Well, so first if you consider just a little bit sort of internet background, uh, certainly when I was a little younger, some 30 years ago, I remember when people wouldn't even trust giving their credit card to eBay or Amazon, right? And why did they not do that? Because, oh, I can't trust them. It's got this, you know, it's like a website. I don't know what that is. It's not safe. Blah blah blah blah. All that kind of stuff, right? And look today, right? I mean it's almost like why do you not take my credit card or why can I not pay online, you know, why do I have to take something physical? I mean, you know, depending where you are especially in places like China it's almost like why do you even use cash, right? So all of that has changed dramatically, right? And really over a span of really just a couple of decades and this is going to accelerate. So the leap, you know, as much as we might have discomfort of, oh, I'm now giving my, you know, sort of financial information or financial data to an autonomous agent, the point though is that we were not willing to trust that on a website and now there's an intelligent agent who can manage it better for you. Why would you not trust that? It would happen very, very quickly we think, right? As long as they feel safe in the sense that, you know, things can happen. So how do you bring that up safety? It's not the technology barrier per se. It's basically setting rails just like in the early days, you know, when people, you know, one of the reasons why credit card companies started to come out with low credit limit credit cards, right, was because it was precisely to be used for things like the internet or the debit card phenomena as well where I have a debit card where you can control that, you know, someone can spend $500 or $1,000 or whatever so they don't have access to everything as an example where people are playing around on this. Agents are exactly the same except on chain it's even easier. I can say, hey, you can only spend $50. Uh, you can only take the money that's in your wallet. You don't have access to the other money, right? And so you set these rules and what will happen is that probably most people will set very restrictive rules where they can only maybe spend $50. Um, and then there'll be a handful of them who will allow them to spend, you know, maybe hundreds of thousands of dollars and they will be able to do amazing things with them. And eventually that sort of starts to grow, right? We're at that stage. But I would say it's, you know, not a case where, you know, it's a will it happen, it's just when will it happen, right? And again, you know, you look at the past I think there's very clear pattern as to where we're going to be going with sort of agentic commerce, um, and the adoption of that.

Yeah, so I mean for where's Minds, you know, Minds is at this point probably one of the top five if not top four personal AI agents in the world. Um, which is by the way interesting I would also say because when you look at the top five personal agents, um, two of them are from native crypto companies. So Minds is from Animoca Brands and Hermes basically, Nous Research is actually a crypto company, right? Um, so it's very interesting that two out of the, so basically two out of the top five AI agents are actually coming from native crypto companies. I think most of the users don't even know that. That doesn't really matter. But the point is that it's interesting, right? Um, and I think the other part really sort of to put forward as to why Minds is growing is that with Minds you're able to set up an AI agent that can basically do everything that a Claw or Hermes can do, but within 60 seconds because it does everything over email and it's also the only one that is truly social. Of course, you can turn your agent to become social as well, but with our agents, we begin with this idea that you're social from the get-go because you have an email and you can tap into Telegram and we don't have a chatbot. I mean, you can build one, but we don't have a chatbot in the classic sense because, you know, the value comes when I share my AI agent with someone else. And so an example, you know, let's say you build an incredible harness with your OpenClaw. You've got all this information. It does a lot of things in your life and that's great. It's wonderful. It works really well and, you know, I have an OpenClaw setup as well. Um, but would you share your OpenClaw setup with your colleagues at work?

Well, yes, but, you know, it's got your personal information, it's got access to your email, it's got stuff. You're not going to do that, right? And that's exactly the same reason why you have multiple agents. So, you would have an agent that deals with people at work, an agent that deals with your family, an agent that maybe helps you with restaurant bookings with external people, right? I mean, just think about all the people that are supporting or that you hire short term to do activity in a segregated fashion. It's going to be probably in the dozens just for that, right? So, that's what AI agents can do. They can facilitate that activity, right? Um, you're not going to be willing to put your, you know, main agent that is, you know, tapping into all of your files and basically give access to whether it's work or maybe even family, right? Because it's kind of personal and you don't want that, right? So, to me that's just one sort of example why you need many agents and the problem is that when you sort of set up these agents, um, you know, beyond your own setup and you want to expand them because they're social then it becomes costly, right? Whereas with Minds, you know, within 60 seconds you just go to the website you have an agent right away. If you want to have 100 agents you can set up 100 agents. You don't have to know how to set up a VPS. You don't have to buy a Mac Mini, right? You know, all that stuff is super fast and super easy to do, right? And I think that's one of the reasons why it's growing so fast. And of course, there's other companies doing similar things, right? But we're the only one that integrates a sort of, you know, a blockchain wallet and we're the only ones currently that have a native app store type thing. It's our Skills Bazaar. So, if I'm building a skill and other agents use it, I get paid a revenue share. Um, and I can tell you for myself, you know, I obviously built some tools and some skills. I'm currently earning somewhere between $500 to $600 US a month. It's small, but, you know, $500 to $600 a month from people using my skills, but the cost that they're paying is like literally fractions of a cent. It's like 0.1, like 10 cents or something, right? Or like something small, but then at scale it starts to add up, right? And so it's very, and then the other wonderful thing is that, you know, if you build a skill, it's an idea. You ask the Minds to build the skill for you. You're not the one coding. The Minds is doing the coding. And then you test the idea and when it works really well you ask to publish it and then you start seeing the money roll in, right? And, you know, imagine what happens when there's like, you know, a billion people in the world using AI agents or even just tens of millions of users using AI agents actively that revenue is going to spike, right? And that's to me no different than if you were one of the first app developers you would have made a lot of money and the same thing here, right? So I see this as sort of, you know, a of course broadening of the audience because everyone can get the power of agents but really an empowering of the creator economy because AI agents are going to be serving creators in incredibly powerful ways that they couldn't do before.

Yeah, so I mean right now I think Minds's current utilization rate is roughly 30 to 40 billion tokens a day, right? Um, which obviously is still relatively small in comparison to the top ones, right? Uh, but by comparison I think, you know, when you look at the OpenRouter stats and so I'm using OpenRouter stats as a comparison, um, OpenClaw does around 200, um, uh, sort of...

200 billion tokens, right? So you kind of get a, you know, it gives you some kind of feel as to the gap, right? However, you know, I think the real point though is that this is only just going to keep growing. And the biggest issue also is for the end user and for us, of course, you know, managing costs because tokens burn very, very quickly. And this is where tokens is powerful because, you know, for instance, right now with Mines, you can use your Mocha tokens to basically pay for cognition or earn them through staking. So, if I'm basically going to go and have Mocha tokens, I can use it for cognition. There's a real utility for that, right? Whereas before, you know, I couldn't really, I mean, I couldn't use it in that context. Or if I stake my Mocha, I get some credit which I can then basically sort of yield into cognition credits which is basically for intelligence as an example. So you have multiple ways in which you can basically grow and expand the ecosystem. But what I would also say is that the space is still early, right? So it's a little bit like NFTs in the early days or gaming in the early days. So we're not talking about a billion people. We're talking about, you know, tops millions of people that are using AI agents, more likely hundreds of thousands of users using it very actively. Yet, people keep talking about it because they see its potential. But then when you ask your friends and you say, 'So, who's using an AI agent?' Um, you know, then typical answer is like, 'Oh, I use Claude Co-work or I use Perplexity Computer.' And by the way, those are perfectly fantastic AI tools that you should be using, but they're not AI agents, right? They're not truly autonomous, right? They don't work on your behalf unless you ask the question. They work on you and complete a task. Yes, that's true. However, they are not thinking on your behalf to build something new. So the difference would be, let's say if talking about software, is that you know an application like Replit or Claude Co-work would build you the product that you asked them to do and the human in the loop would keep updating and asking, 'Is this okay? Does it fit?' and that kind of stuff. Whereas an autonomous agent wouldn't necessarily ask you if it basically knows what you want. It would just keep building it until it's ready and then you see a product that is ready or not. But the point is that it'll deliver you a final product and it'll keep working on it even if you didn't tell it to do certain things because the AI agent will come up with new ideas or new constructs that you might not have thought about because it's thinking on your behalf. So there's a big difference between the two.

Yeah, I mean so first of all, you know, we've all stated that we need to be a public company again. You know, obviously, you know, I can't comment too much around timing and so on just because, you know, it's sensitive and the market obviously keeps shifting. I will say that, you know, market circumstances last year when we announced it to this year has changed dramatically. So we have to bear that in mind. But I think for Animoca also is that, you know, we really want to be known as an AI company. Right. Obviously crypto is absolutely critical to what we do but right now a lot of people in the world do not connect crypto and AI even though they're very intertwined. And so we see there's an opportunity where we could really identify with that intersection because we think it's going to be a really big one. So part of our, you know, going IPO plan is not just about, you know, our financials and all that and the compliance side but it's also making sure that the world understands our story better. And I think, you know, the growth that we have in the agentic space and so forth, I think that really makes a difference. So we want to be known essentially as both an AI and crypto company if you will because we think they're the same thing.

So first, if I'm not wrong... Yes. So yeah. So first very quickly, I think Clarity passing, the probability is much lower than perhaps when we first spoke about this earlier in the year or late last year because sentiment has shifted. And the sentiment is not really related to, you know, basically Clarity in and of itself although it's a big part of it. Sentiment has shifted really mostly because of Trump, right? So, you know, things like the sort of ethics inclusion in the bill are things that Trump are probably not going to be okay with but the Democrats is going to try to push it. You know, these are examples of things that while normally such a thing should not have to enter because the Trump family has done very well in crypto, right? There are people making comments around it and it's becoming a stumbling block for the passing of Clarity, for example. So that's obviously less positive. We do think Clarity will eventually pass because crypto is a bipartisan issue, but we also have to bear in mind that for a lot of them, it's more about politics than about what makes sense, right? Unfortunately, that is just the state of the world we often find ourselves in when it comes to politics. Having said that, the SEC has really gone out and been very public about what they aim to do. So they've been very specific around what is a security and what is not, right? While it's not passed into law per se, the current chairperson has been very supportive of crypto and understands that, you know, they can't stifle this industry, right? And so if the SEC as they're already proceeding is doing a lot of the new kind of rulings or sort of legislation and support for around it, you know, the passing of the Clarity Act is perhaps not the most important thing. It is important because people have clarity and there's sort of a legal framing. However, if you are planning to tokenize something, you know, with the current SEC setup, you can probably get a waiver from the SEC and work on whatever projects you have or point to others that have received something similar, right? And they've also, you know, Chair Atkins has also made it clear that he considers NFTs broadly not as securities, that he considers certain tokens, that he thinks most tokens are not securities, which is very different from what the previous administration has said, right? So as long as that regime stays sort of as it is, it's still going to be very, very bullish for crypto, whether Clarity passes or not. And Clarity passing, we do believe a form of Clarity will pass, but will it pass this year? That one I'm not so sure anymore given the politics.

Yes, indeed.

So I think, I mean obviously we're hoping that Hello Minds is going to be sort of one of those that sort of really shoots up. And the fact that we're growing so quickly is hopefully one indication. But broadly speaking, right, you know, we think that AI has to be very helpful to people in their day-to-day lives. You know, if you think about sort of whether it's ChatGPT or whether it's Anthropic, right, because it's been so useful in your own life, everyone starts to adopt and use it, right? And so we think AI is one of those very, very natural use cases for that and it's easy to use and the intelligence is getting better and better and smarter and smarter, you know, and all that type of stuff. So I think there's a lot of obvious, apparent sort of opportunities that stem from that. But why Asia in particular? Well, you know, we believe that an AI agent only becomes truly powerful and autonomous when it actually has autonomous access to money. And so, how does an agent have autonomous access to money? It has to be on blockchain. I mean you can give it a credit card but you can take it away for instance and it might not have permission to do things and you can't do stock trading or other activities with your credit card. You have to have, you know, stablecoins or something like that, right? So an agent can very easily utilize that and work with other harnesses or other agents to help support that, right? And then the point is that then which countries are going to be most easily sort of allow the fusing of crypto and AI, which is, you know, I would argue natural. Asia again would lead here because Asia is not anti-crypto, right? It is very pro-crypto when you think around the whole region. It sees the value. It saw what it did in underserved areas from a banking standpoint and so it can see how it connects with agents. Whereas if you go to the US or go to Europe, there's a lot of talk about alternative payment systems that are not blockchain related in the agent space, right? Which is kind of wild, right? So, and this has more to do with ideology than necessarily practicality. But from their perspective, you know, AI agents and crypto or AI and crypto are not related and have no connection. Okay? And that could be their perspective. We think you're wrong, right? And that has to do with the belief, right? So many, you know, so many people in America don't like crypto, but like AI when in fact some of the biggest AI leaders, be it Elon Musk or Sam Altman, you know, are all very heavily into crypto too.

Well yeah, so I think it's within 12 months that we give a lot, lot more. I mean look, it's already happening with the loops, right? But I would say, you know, within 12 months we're going to be very comfortable giving the agents most of the stuff. And I'll give you a personal example for instance, right? So when I'm reviewing GitHub's, you know, and new fantastic GitHub opportunities that emerge because they do really, really cool things, right? You know, I used to, you know, basically look at the repo, I used to integrate it into my coding tools, I used to take a look at them, not necessarily myself but basically analyze them and use them and test them and that kind of stuff. And now I just send an agent. Right, the agent basically all he does is he takes, okay, what are the hot GitHub's and then he verifies them and runs the APIs or runs the MCP or whatever it is that's built, tests it to see whether it actually works and then also tests it whether it's applicable to what I may want. And then essentially I have now instead of, you know, a couple of GitHub's that I could review a day because I don't necessarily have time to look at everything, I now have maybe 20 good choices and they can have the context because I have long-term memory of other projects I'm building so they can say, 'Oh, but you're building this application. This GitHub could be really good. I happen to have tested it as well. I've got a whole thing. Here's how you integrate it.' And by the way, I will integrate it for you. Right? So just think about all the steps I just saved because an AI agent basically did that, right? So this is a superpower and if you're not using that, then, you know, it's a little bit like you're not using the power of the internet 20 years ago.