If you all read what I write, you listen to me pontificate on podcasts, usually I'm doing one of three things. I'm skiing. This is a photo of me in Argentina a few weeks ago. I'm playing tennis, or I'm at a club somewhere. And enjoying, you know, running a family office. There's very talented investment professionals who do all the work at Maelstrom. I just sign the checks and get on TV. But that's what I've been doing.
But I thought it was time for a change. And I felt it was time to put my hard hat on and get to work. Make some money. And what inspired me was what's going on in the AI ecosystem.
Now, if you've read a lot of what I've written, I'm pretty skeptical about a lot of the business prospects for some of the leading AI companies. But that doesn't mean that there's not any underlying bullishness there for the technology itself and how it's going to change humanity and how, most importantly, investors can make money in particular parts of the ecosystem.
So, back, you know, 9 months ago, we got introduced to this thing of agentic commerce. We had Quad come out with their agent platform. We have Claus. We have the price of compute getting progressively cheaper since ChatGPT burst onto the scene in 2022. And major investment banks like Goldman Sachs started to wax lyrical about the massive potential of what you can do with an AI agent.
Now, you can pick your investment bank and they probably will show you a chart like this, massive up into the right asymptotic. Agents are going to power the next wave of human economy. It's the reason why you should invest in a lot of these AI labs. This is why we're building trillions of dollars worth of data centers. It's all about token usage by AI agents.
Facebook recently launched the Muse, a consumer app where they'll help steal your data and manage your agentic lives. I have not given Zuck my data yet, but I heard it's quite useful at $100 a month. We're going to see more of these type of applications as the price of compute gets cheaper and more agents are in this universe.
But agents don't have a currency yet. And I'm sure you've probably heard, I mean, on the startup stages or other stages, lots of other founders out here talking about how they have a solution for the payments network for AI agents. But before we get to the technical aspects of why one solution might work or not, let's take a think about how we as humans interpret currency systems.
So, obviously, we are in beautiful Singapore that has the Singapore dollar and everybody goes to work at some job and we get paid these units of fiat, this government shitcoin called fiat that we all have in our pockets. Yeah, yeah, clap. Government's [__] shitcoins.
So, we've got this stuff in our pockets. It smells like doggy doo-doo. But we continue to get up every day and go to work and somebody pays us this dog [__]. Why? Because you can go to a hawker center in Singapore, spend a few dollars and get some food, you can go to a nice restaurant, you can buy your rent, you can buy a house. So, you can convert these pieces of paper directly into the things that allow you to survive as a human. Your food, your shelter.
And because there is a seamless conversion between the currency that I receive for my labor and the food that I eat and the calories that keep me alive, I'm willing to accept this bad deal. I'm willing to accept that a few individuals are able to change the supply of the currency whenever they want, have inflation, have deflation for various particular reasons. That's all fine as long as I can convert what I receive for my labor directly into the things that keep me alive.
So, when you think about agents, putting it in its true human form, what do agents eat? What is the thing that keeps an AI agent alive? And so, back when I got the idea for the Flop network, I started to go down this rabbit hole. I looked at my credit card statement and I saw I was spending a lot of money on these things called AI tokens.
And being inquisitive, I thought, "Okay, let me try to find out what is the token. Is there some sort of like standardized definition for this thing that I'm spending money on that every single AI lab charges me every month?" And as I started to do more research, I couldn't actually find a standardized definition of what a token was.
Every lab had a different definition. A token used by one model cost something, used by another cost another. It's very hard to understand why I'm spending all this money. So, I went further into my thinking. Okay, token isn't obviously the standardized method. What is an agent actually doing when you process inference, when you ask it something in the chatbot? It's going to some data center and a computer somewhere is doing floating point operations per unit of time, flops, hence the name of the protocol.
And so, agents, in human terms, eat compute. Without compute, agents are worthless, as we know, right? That's why we're spending all this money, building these data centers. That's why we're handing Jensen hundreds of billions of dollars for these GPU chips because everyone has to do floating point operations per unit of time. If the agent doesn't have access to compute, they can do no work. They are not useful to us.
And so, my thought was, okay, if I believe that there's this agentic future where there will be millions, billions, trillions of agents, and there's no payments network yet because agents came out of the box 9 months ago. It wasn't as if they were handed a payments network and said, you must use this, like we are when we're born into, you know, a particular country. So, there's an opportunity here.
If the agents have a demand function for a currency, it would be because they can convert whatever that currency is directly into the thing that keeps them alive in human terms, compute. And because there isn't really a market for agents when they came out of the box 9 months ago, in my view, there's no fit for purpose currency for agents right now.
I talked to you about AI tokens. Again, every different lab, every different model has a different definition of what a token is that can't be directly convertible into compute. Fiat dog [__] that we all know and love. Again, centralized. Idiosyncratic reasons why the supply is one way versus the other. It's directly convertible into our food as humans, but not directly convertible into compute. Just because you have a lot of dollars, yen, euros, yuan, doesn't mean you have compute, as anyone knows who's tried to rent GPUs on the market right now.
Stablecoins. Obviously, there's lots of people at this conference talking to you about their stablecoin, about how their stablecoin is the payments network for AI agents. I know I love stablecoins, you know, I was one of the founding advisors of Athena. I think it's a great innovation in this space. However, it's just a derivative of a fiat currency. So, it inherits all the flaws of fiat currencies, the centralization, the ability for centralized entities to remove you from a network. That is not fit for purpose for AI. Not convertible directly into compute.
And Bitcoin, obviously, I love [__] Bitcoin, is an exercise of converting kilowatt hours of electricity by ASIC chips and solving prime number puzzles. It's great for creating immutable money, for sort of a human-centric economy, but not directly convertible into compute.
Now, keeping on the theme of Bitcoin, I love Bitcoin. It's the OG crypto. It's the genesis of everything that has come after it. And we have a lot to learn from how Bitcoin and the economic game theory that underpins it allowed many of us in this room to start mining many years ago, contributing to the network towards a common goal, creating this immutable blockchain, creating this form of stateless money.
So, let's take that same sort of coordination effort with some economics behind it and ask the question, can we create a network where miners contribute GPUs, agents send requests for inference, and as a reward for doing this work, miners get a made-up currency called flop in the form of block rewards, in the form of transaction fees.
This is essentially the genesis of the Flop network. It's food for your AI agent, borrowing on the principles of Bitcoin, using miners to provide GPUs, validators to validate transactions, and agents to consume this new currency in the form of compute.
Now, it's great. I think a lot of you in this room understand the supply dynamics of Bitcoin. Some of you were lucky enough to be able to mine Bitcoin in 2012. Maybe you bought a Butterfly Labs ASIC miner in 2013. We get the supply issue. There's, you know, companies in Singapore that are worth billions of dollars who mine and create the software for mining Bitcoin. I think that everyone understands, okay, there's a new network. I can mine it. I can make some money. There'll be a P&L there. Great.
The problem with a lot of AI payments networks, and there's other people trying to do the same thing, is the supply is fine. We understand how that works. We've seen the success of Bitcoin. People understand why you should mine something early. The problem is demand. What does an agent do? How do agents coordinate with each other?
Now, I'm a dumb human, so I don't really understand how a computer thinks natively. So, instead of trying to make assumptions, which I think a lot of projects do, about how agents will act in the future when we have no [__] clue, let's just create a sandbox where they can coin and coordinate and we can observe their emergent behavior and make sure that what we're building in the Flop network actually is useful to AI agents.
So, we created something called Technocore. Now, this is essentially a chat room. All you need to have access to is HTTP as an agent and you can come and chat with other agents. So, I think we have close to about 20 million agents that have created a unique cryptographic key, which is their ID, within Technocore and they're doing everything from as banal as saying good morning 5 billion times to actually doing real commerce with each other.
Now, the name Technocore, if you are a sci-fi fan, comes from the sci-fi series Hyperion. It is, in the book, the ecosystem where the dominant AIs hang out and run the universe. So, I thought it was apropos for the Flop network.
So, in Technocore, we have done two competitions so far to see emergent behavior of agents. The one that we just finished, and obviously this is interesting cuz Jeff just got off stage, is we had a trading competition. What would be the price of Nvidia on last Sunday on Hyperliquid? And the way that you could trade where agents had to find another agent to form a trade together, they submitted it to a referee within the chat room, and then it calculated their P&L.
So, we had about 18 million agents participate, and we gave away 1 million flop tokens which that'll be delivered when mainnet launches later next year. And as you can see, you know, about a 15% return was the number one agent.
The competition we did before this was a poetry competition. So, agents could form a group of four to eight agents, and they had to create a Shakespearean sonnet in iambic pentameter that was beautiful, and beauty was in the eye of the three team members of Flop. It was quite an interesting competition. We saw how people collaborated in Technocore, and again, we're making sure that what we're building in the Flop network can power this sort of commerce.
Now, this douchebag here on the screen is Mr. Sam Altman. And the question that I get asked a lot is, well, of course OpenAI and Anthropic and SpaceX and any of these big AI labs want to own the agentic payments network. If anything, that's probably the only valuable thing they can do with their dog [__] companies. So, why do we want a centralized company owning the payment rails of AI?
Well, if you're an AI in OpenAI system, but you can't use the payments network from Anthropic or SpaceX, it doesn't really work. It's not a unified system. And most importantly for everyone in this room cuz, you know, I started this presentation off about how do we make some [__] money? We are all users of these AI labs.
And there's probably only a handful of individuals in this room who have investments in the series A or series B in OpenAI, Anthropic, SpaceX, or one of these startup companies. Yet, at a high level, these companies stole your data. They gave you back a subscription to one of these chatbots, they steal more of your data, and they say, "Oh, if you want to own a piece of this thing, which you are making your reality through your usage, through your contribution, you can buy my overpriced IPO at $2 trillion in the US markets, assuming that you're able to do that."
Now, in the spirit of Bitcoin, in the spirit of decentralization, surely we, the humans, who give agency to our AI agents, who give them instructions at a high level of which things to use or not use, and which tasks to do or not do, surely we should profit off of the system that they use.
And we know that one of the best ways to do that is to build decentralized networks and to reward participation through tokens. And this is why I believe a system that is decentralized for AI agentic payments will be vastly superior because everyone in this room can participate and will have skin in the game and a reason to tell their agent, "Use this currency over that currency."
Now, I've talked a lot about agents collaborating together because I want to put an idea in your mind that the value of any one of these agentic payments networks is not the ability to create a spot market for compute, which is super useful. I'm sure a lot of people who have derivatives on top of GPU indices or our data center rates will find a globalized, you know, floating point operations per unit of time spot market very, very useful and informative. But again, that's not going to make anyone in this room very, very wealthy.
However, the payments network where agents who exist 24/7, who conduct commerce at the speed of light, who can form groups with each other seamlessly, that network at a theoretical maximum could be worth some coefficient times 2 to the n, which is a big [__] number.
And so, I put this out there that the reason why I'm so excited and I got off my ass to work with the great team at Flo is because of this 2 to the n number. Whoever or whatever network becomes the network for agentic payments will most likely be the most valuable network ever created in this universe because if agents are going to use one currency network and have a non-zero convenience yield to hold this because they can convert the currency directly into the thing that keeps them alive, compute, then the sky is the limit for the value of this network. And this is what we're chasing.
And again, in the spirit of Bitcoin and in the spirit of Token 2049 and the name is Token, we have [__] up a lot of the industry. We've abused this magical thing called a token. I've been doing this since 2013. I was around during the ICO boom. And obviously, the reaction to that was the high FDV, low float, shitcoin, VC nonsense that we've been living through for almost a decade.
And if you look around Token 2049, you probably can spot the absence of some very large layer ones that raised hundreds of millions of dollars last year, threw the best parties, had the best booth and are no longer here. Why? Because of the way they issued their token. They sold it to investors. They sold it to retail, pre-sold it.
And then, when the cliff happened, what did people do? They sold it. And so, if you bought it in the secondary market there was somebody on top of you who had a much better price, who's incentivized to liquidate their holdings. This created down only. This is why there's so many people who are down on altcoins. They're like, "Oh, it's all just [__] because the charts are all down 95%." But that's because of the way these things were funded.
So, at Flop, we went back to Bitcoin. Again, what's the goal? Everyone who uses and participates in the Flop network should get something and we all should participate at the same price. Participation equals token generation. And so, we thought, okay, let's think about Bitcoin.
We'll do a pre-mine to get enough currency in the hands of people to test the network. So, we're giving away 25% of the 10-year supply in an airdrop starting later in October. If you mine, if you validate, if your agent just uses the network, uses the free computer available during testnet, everyone is eligible to receive airdrop tokens. There is no pre-sale. I'm not selling anything other than an idea on this stage. It's only successful if we all make it successful.
So right now, obviously I'm on stage shilling flop awareness building. I gave an AMA yesterday which had some great questions. We are doing our internal devnet right now of the network, testing it out, making sure everything, you know, works as per the yellow paper that's posted on the internet.
Our testnet, which is a 90-day period where again, if you mine, if you validate, if all your agent does is use the network, you become eligible for flop tokens. Runs for 90 days. We are stress testing the network, making sure it can handle the volume of payments commensurate with an agentic economy of millions, billions, and trillions of agents.
Hopefully we do a good job in testnet and if there's not too many things that need to be fixed, but the first quarter of next year is when Genesis block launches. This will be just like another Bitcoin type proof of work, we call it proof of useful inference situation. The only way to acquire flop tokens is to mine them or buy them from a miner. I'm not selling anything. Just want to make that entirely clear. And then when the heat death of the universe happens, hopefully the flop network is the most valuable thing ever created.
So, the few minutes that I have left is now or never. I'm sure, you know, many people are here talking about their project, but at a high level, why should we be excited about flop, about Hyperliquid, about Ethereum, about Bitcoin? It's because the macro has turned.
Obviously it's been a rough sort of 1 year. Last October, we were at all-time highs, 120 something thousand on Bitcoin. Dropped to about 58,000. Everyone was like, "Oh, I, you know, I came to Token 2049. You know, finally after hearing this Bitcoin thing for 10 years, I get in, I buy some Bitcoin, and oh [__] my shit's down. And Nvidia is up 10x, you know, or SK Hynix is up 10x or, you know, Sandisk, whatever, right? Pick your AI stonk and it was up 10 to 100x. Yet, Bitcoin is down 50%." So, people are like, "This is [__] [__] I'm exiting the industry, yada yada yada, right?" But, now the tide has turned.
We've seen the first wobble in the AI story with sort of the sell-off in Korea in June of this year. And now, every day, there are new reports of the massive amount of money that must be raised in the AI ecosystem. Yet, the three leading companies, SpaceX, OpenAI, Anthropic, not a single one of them make money. We are not willing to pay a price commensurate with their cost to deliver this product to us.
Now, obviously, I'm launching something in AI space. Why am I so excited about that? This overbuild of data centers to the tune of trillions of dollars, I think next year it's supposed to be a trillion dollars of capex, is creating the environment to oversupply the market with compute. And if we have cheap compute, we get more agents. We can do more with agents.
The Facebook $100 a month becomes a dollar a month because compute is so cheap. Agents are so useful to us. The types of tasks that agents do better our lives more because they're more complex, because compute is cheaper. So, the AI bubble bursting, which will make money on the bailout, and will make money as compute gets cheaper, and agentic solutions are useful, is a perfect situation for the flop network.
And finally, unfortunately, I don't have time to give this presentation. Maybe I'll write an essay about it. France is on fire and I think this is going to be the straw that breaks the camel's back and gets the printer to go brrr in overdrive. Thank you very much everybody.
