What if the next crypto bull market isn't driven by people, but by AI agents? In this conversation, Yat Siu lays out a provocative thesis. AI could become the biggest catalyst crypto has ever seen, creating millions of autonomous agents that need wallets, own assets, and transact onchain. He explains why altcoins may become the commodities of the AI economy.
You're already spending tokens for OpenAI and Anthropic and Gemini. That's actually a spend in tokens right now. Imagine turning that into a commodity. It's a representation of compute and energy and there's real utility. There's an actual burn mechanism that actually has purpose.
Why NFTs could make a surprising comeback.
NFTs are essentially a symbol of the wealth class in crypto. As crypto becomes wealthier again, NFTs become that status symbol. I also believe that AI agents are going to be buyers of NFTs as well.
And why today's crypto market is missing the one ingredient that fueled every previous cycle.
And the other thing I think that's missing in our industry much more is we got to bring back the fun. NFTs were fun. The metaverse is fun, right? Gaming is fun, right? We got to bring that in.
If Yat is right, the biggest opportunity in crypto may be hiding in plain sight. Let's go.
Let's do.
Is this thing on? I'm nearly invigorated by actually eating food. For the first time today. It was a big move for me.
Oh, you did it fast or something or what?
No, I just never got a chance, but I had a two o'clock break, so it was good.
It works, too. Awesome. So, I was with Dan Tapiro earlier and he told me to ask Yat about Animoca Minds.
That's what he told me. I said, 'Give me some good insights. I want to hear about it.'
Yes. AI agents. Yeah. Are we on to talk?
We're on. Oh, there. Let's go. Let's go.
Well, just very quickly. So, Animoca Minds or now called HelloMinds.ai is basically our agentic platform. You can set up your own AI agent. It's very similar to OpenClaw with one exception: you basically don't have to set anything up. You just put it over email. It communicates with you over email, over Telegram, over WhatsApp. But more importantly, it has a personality. It's more like a coworker or a friend. We're still in beta. We just relaunched the website. We have thousands of skills now. It's wild the kind of stuff that's happening. For instance, I asked it to write a Wordle game. Instead of going to the New York Times, within 10 minutes it constructs a game custom made for me and I just play it with my friends. It gives you a window of how the world is changing. And by the way, we think that AI agents are going to save blockchain.
I'm now remembering that you sent this to me in DM. And I put it on my 'I'm going to check this out' list and didn't. So now that you've said it to me in person, I've definitely got to check it out. I mean, for instance, for yourself, how many AI agents do you think you'll have in the future?
And I put it on my 'I'm going to check this out' list and didn't. So now that you've said it to me in person, I've definitely got to check it out. How many AI agents do you think you'll have in the future? Incalculable. I literally think it'll be hundreds, you know?
So, I have 200 right now, right? And yes, I agree with you. So, now think about it, right? If every human is going to have tens, if not hundreds of AI agents, then how many AI agents are we going to have? We're going to have hundreds of billions of AI agents. They're going to be all floating around on the internet. They're going to have their own wallets. They're going to transact with each other. They're going to find stuff for you. And basically, that's actually the use case for blockchain where we're going to bring the billions onchain, but not because they're onchain, because their agents are onchain. They're going to do business with each other. They're going to trade things. They're going to buy things from each other at microtransaction fees that won't work with a Visa or Mastercard, but will totally work on something that's onchain.
Yeah, we've had these conversations now for four or five years. How did we miss the part when we were saying how are we going to get a billion people onchain that we didn't ever need to—those are people, right? We can have trillions of people transacting.
But they're there on our behalf, right? And I think this is the powerful thing because this is not just a shift from, hey, we have agents doing stuff for us—it's a shift of the entire internet. Just think about app discovery. Right now we go to a website, the app store, Steam, to a place or location because that's how we discover stuff. But what if you have an AI agent that finds stuff for you? You no longer have to go to the app store. They'll discover everything for you and negotiate with other AI agents, other platforms to find the best product, cheaper, better, faster. The kind of stuff blockchain is good for. Or for instance, the dating scene. You're not going to go on Tinder or Bumble. You're just going to ask your AI agent, 'Hey, I'm looking for these kind of dates.' They're going to find you a connection with other people looking for that. There's not going to be a platform. They'll link to each other and they'll probably be pre-vetted because the agent asks questions and figures out what they're looking for. Much more detailed than some app where you swipe left or right. Unless that's fun for you. That's a different story, right?
What happened to meeting a girl at a bar?
Well, that can still happen. But my point is the experience of that kind of efficiency extends to everyone. When you think about the back end, we're totally being monetized through advertising. This year, the forecast is close to a trillion dollars, like $950, $960 billion in online advertising. But what is that? It's a transaction of our time and attention. We just never get paid for it. I'm using Instagram, TikTok. They make money on the advertising, someone pays for that. For that, I get a service. That's now all going to be abstracted by AI agents on the back. When was the last time you clicked through on a website after you did your search?
Right. You got your answer. That's the point. Normal traffic is down. So they get paid through microtransactions instead of the traditional forms of advertising.
I think of all the iterations of these ideas we've gone through over the years. Everything to earn, right? Play-to-earn, game-to-earn. Or is it just the agent does it for us instead? There's trillions of dollars in—
It's a different form of earn. One way to think of everything on blockchain is that for the last six or seven years, we humans were the P for the agents. We were the playing ground, the testing ground, and then we made it work. We went from CryptoKitties would kill Ethereum because it couldn't handle that, to now trillions of dollars of volume of transactions, no problem. We still haven't grown onchain traffic. However, that's where the agents come in.
So, is it over for the previous narratives where we actually cared about mainstream adoption from actual humans?
I think mainstream adoption for humans will come, but it will come because agents are there, because agents represent business opportunities in and of itself. We don't have enough opportunities onchain that exist. When you have billions of agents doing stuff, we expect a future where agents and humans interact with each other also onchain. But right now, if you go onchain, it's kind of lonely. We're still only at 40, 50, 60 million. We need to be at hundreds of millions. AI agents will come first on our behalf, but then also drive us into that area. That made us think the metaverse is not dead. It's just that we thought about the metaverse in the wrong way—instead of going into the metaverse with goggles and devices, the metaverse is coming to us in the form of AI agents.
The most interesting iteration of the metaverse was always something that interacts with your daily life rather than you having to completely—like you're driving by and you click on something and the ad goes and purchases that for you. Rather than having to put on goggles and exist in a cartoon world.
Exactly. For instance, I wrote a very simple agent that just told it to do a price comparison on local supermarkets. I discovered in Hong Kong there was a 75% spread between grocery goods for things like eggs, milk, and water. You don't think about that because you go to the supermarket and buy stuff. Just that type of stuff can give you huge savings, and it impacts your real life. Never mind travel booking and scheduling—even the small basic stuff. Because it's an AI agent, it comes to you with suggestions as opposed to a chatbot where you ask it and it gives an answer valid only for that moment, as opposed to one that keeps looking for stuff and optimizes for you.
So the value accrual of all this—is it to stablecoins? Is it to the chains that stablecoins are moving on?
I think it's going to be in altcoins. The tokens that use inference become the commodities for those AI systems. Think about it—you're already spending tokens for OpenAI, Anthropic, and Gemini. That's actually a spend in tokens. Now imagine turning that into a commodity. It's a representation of compute and energy with real utility. There's an actual burn mechanism that has purpose. You can hedge, you can trade, no different than commodities in the physical world like semiconductor chips. Of course, stablecoins are going to be a big winner in utility and usage. But that doesn't mean stablecoins themselves don't appreciate. If you're in the business of stablecoins, it's great because that's a natural onboarder. I think the core assets like Bitcoin, Ethereum, the majors, Solana—they'll do fine because they're going to be using those chains. However, the value pool for the chain mechanism isn't going to be quite as big because I view them like telcos—really important, infrastructure growth will be there, but they're not capturing value the same way because if they capture too much value, people don't use them because fees are too high.
Right. The faster, cheaper, fraction of a penny doesn't add up enough and that gets arbed away and competed away over and over. Unless you have 100 billion agents onchain, right? And then microtransactions count.
I just wonder when you say it'll be altcoins and tokens—is it altcoins and tokens that have yet to be created? Or is there some bid for coin number 74 on CoinMarketCap that we so deeply believed in last cycle?
I think one, it has a lot to do with community building. If the tokens have a community and can attract the use of this agentic system, you've got a base. Size matters. If you have a reasonably sized community with a reasonably good token with some liquidity, you can use that as a base as opposed to a brand new project. I'm not saying it can't be a brand new project, there's just an advantage around that. But is that project going to be adopting AI? Are they going to do agentic AI? That's a different question. But I definitely see value in the altcoins that are out there and some of the majors.
So they'll have to be sort of reskinned for what their purpose is, but wonderful utility. That's so fascinating to me. Is that where you are entirely focused right now?
Well, no. One of the big news we also had as a business is our joint venture Anchorpoint received a stablecoin license. There's only been two stablecoin license receivers in Hong Kong from the HKMA. Our joint venture with Standard Chartered and HSBC. That's one of the big news. The other one is we're doing a lot with RWAs. I think art is another example. Arts are going to be another winning category because institutions and agents around the world are going to want to buy certain types of assets that deliver actual yield, and that's something they could do onchain—better, faster, cheaper. You expect that area to grow. And of course, the agentic stuff is a big focus area because we think everything we've built so far was in service for these agents.
Stablecoins of things other than US dollars, right? What are we at? 99% probably of stablecoins are dollars.
Yes. Well, that's also because most stablecoins haven't launched yet or weren't able to launch.
And I think it's an interesting point that Hong Kong was able to launch a regulated stablecoin before the Genius licenses, which I think is due for next year. There seems to be a lot more optimism around clarity at this moment in time. But let's see. I saw Coinbase's stock pop a little bit because people thought there was a compromise. Circle went up 20% or something. Coinbase went up as well, but Circle saw the bulk of that. It makes sense that the market would price that.
Exactly. I think generally positive on this area. And again, how do you demystify the process? My mom's not going to buy Bitcoin. She's not going to set up a MetaMask. But the agent will do that for her.
Well, that's the interesting part—we always talked about how will grandma use this. She'll never save her private keys, but now all grandma needs is an agent platform and she'll never know what's happening.
I think there's plenty of things that someone might want to buy onchain more and more, but the problem is how do I do it even if I want it, or do I want it badly enough? The masses just want convenience. I think the AI agent, which dissolves the traditional interface, will do that.
So stablecoins and foreign currencies other than the dollar—do you think they're going to be popular, that there'll be a place for them? I think so too, but people seem to think only the dollar can work.
