Moai. (Any views expressed in the below are… | by Arthur Hayes | Jul, 2023

(Any views expressed in the below are the personal views of the author and should not form the basis for making investment decisions, nor be construed as a recommendation or advice to engage in investment transactions.)

Bringing order to a parcel of our elegant but chaotic universe requires a combination of two fundamental components. The first and most obvious one is a massive expenditure of energy; moulding chaos is extremely energy intensive. But the more essential piece of the equation is that you need agents of change who are, above all, exceedingly well organised.

I recently spent a week on the beautiful island of Rapa Nui — also known as “Easter Island” — trekking across fields of extinct volcanoes. Using the debris of volcanic eruptions that occurred somewhere between hundreds of thousands and millions of years ago, the Rapa Nui people organised themselves and formed and erected beautiful, human-shaped megaliths called “Moai”. These monuments to their gods and ancestors weigh literal fuck-tons and required an organised society to carve and transport them across the island. Raw materials alone did not guarantee success; it was ultimately the Rapa Nui people’s forms of organisation that allowed their society to evoke beauty from geological chaos.

In today’s world, we all generally accept — without much thought — that on one side of a border between two countries, conditions can be pristine, while on the other, dilapidated (see: South vs. North Korea). If you stop, put down your smartphone, and think critically, it should strike you as absurd. The border in question is just a fictitious squiggle drawn on a map, and the areas it divides are mere miles apart. Correcting for economic resources, the differences between the “pristine” and the “dilapidated” country are driven purely by how the citizens of those respective countries organise themselves and effectively cooperate to accomplish civic tasks. Zooming out on human history, the most critical catalyst behind our global civilization’s current per capita wealth (particularly compared to our predecessors even just a few centuries back) has been our self-organisation into small units designed to achieve specific goals.

You’re probably assuming that I’m referring to the development of new modes of government. Nah son — democracy, monarchy, dictatorship, etc. are all forms of government / organisation we humans have experimented with since we began settling in cities many millennia ago. Unfortunately, there is no one form of government that guarantees economic progress and wealth. What I want to talk about is an organisational entity that’s been far more integral to the recent exponential increase in our ability to convert the potential energy of the sun and earth into economic goods: the limited liability company (or “LLC”).

The first joint stock companies emerged in the early 17th century. Behold how growth accelerated from then onwards as companies were unleashed upon the world. The most important thing that companies did was explore for, develop, and finally produce energy in the form of hydrocarbons.

A company is a piece of fiction — albeit, one that we all collectively buy into — that generates productivity and wealth by combining the work ethic of its individual human members with the state’s power to enforce contracts. The beauty of a company is that its members are willing to sacrifice energy today for riches tomorrow. A company at inception is nothing but an idea, until someone contributes a portion of their excess capital — either physical or financial — to it before any goods or profits are produced. People are only compelled to invest their excess capital in this manner because, in exchange, they receive a piece of paper that says they own a percentage of the company’s future profits (should they materialise).

But who guarantees that said piece of paper will be converted into a profit share at some time in the distant future? That is where the state comes into play. The state ensures that companies incorporated within the squiggly, imaginary lines that demarcate its borders follow its laws. Failure to follow those laws results in violence being visited upon the violator. This guarantee of enforcement provides potential investors and workers with peace of mind that the company will make good on its written promises. In a way, the state breathes human life into a company.

A Company = The State + Humans

The company structure is so powerful and useful that it is found across nearly every facet of society. It doesn’t matter if a state is capitalist, fascist or communist — they all still have companies. The US and China, for example, have vastly different ideologies and forms of government, but both embrace the concept of the company. The only difference is that in China the state owns companies; in America companies own the state.

Given the importance of companies to the productivity of the state, the state employs a wide range of state-sanctioned entities that help ensure companies’ compliance. These entities make up the “cartel of trust”. Auditors, accountants, lawyers, and bankers provide services for companies and help the state ensure everyone follows the rules and fosters trust between citizens and companies. In effect, these cartel members act as a tax on companies’ profits, as companies are required to employ them just to exist. A company needs a bank account to receive payment for its products and services, and pay its employees and suppliers. A company needs an accountant to prepare its financial statements in accordance with standards set by the state. A company needs an auditor to ensure the accountants have accurate figures. And a company needs a lawyer to write contracts, represent it in court, and help register it with the state. You cannot operate a company without these services.

But what type of organisational structure would / will an AI use? Would an AI that is just a thinking machine, who “thinks” in lines of computer code and has no physical body, organise itself economically using today’s standard company structure?

That is the question this essay aims to tackle. My opinion is that AIs will organise themselves using a Decentralised Autonomous Organisation (DAO) structure. DAOs are reliant on public blockchains — not the state — to operate. The DAO structure will allow AIs and humans to collaborate and serve as the organisational structure that allows the AI + human economy to grow and flourish. This essay will delve into my ideas on how AI DAOs will fundraise and why Decentralised Exchanges (DEX) will ultimately be venues on which new AI DAOs trade.

Similar to my previous essay, “Massa”, this essay will be organised as a series of logical proofs. I will “prove” the following:

1. The state is unable to exert control over AIs because the state cannot kill or meaningfully punish an AI.

2. Because the state cannot exert control, there is no reason for an AI economic unit (i.e., an AI DAO) to follow any state-based legal norms.

3. To force an AI to comply, the network substrate that powers the DAOs will need to use smart contracts deployed and executed on public blockchains.

4. Because the state can’t exert control over DAOs, DAOs will fundraise and trade debt, equity, utility, etc. tokens on DEXs that are not hosted on traditional centralised exchanges (CEX).

5. DEXs will trend toward becoming natural monopolies, as they will be the first truly global trading venues where any being with an internet connection can meet and transact.

If readers believe I have successfully proved the above, then it will follow that:

1. Ethereum transactions will grow exponentially as DAOs proliferate. As a result, the price of ETH should skyrocket in anticipation if this AI DAO hypothesis is widely believed.

2. There will be a handful of DEXs with natural monopolies over the trading of specific types of tokens. Identifying these DEXs and purchasing their governance tokens will produce beaucoup profits.

3. Middleware layers that help visualise the accounts of AI DAOs will be created and be essential to well-functioning AI DAO capital markets.

How could the state kill an AI? Assuming the AI is smart and talented enough to replicate itself and/or divide itself across many host computers and nations (horcrux-style), the state would not be able to unilaterally kill an AI unless they destroyed the internet and all computers across the entire world. Given that no state is omnipotent, this sort of eradication of technology should be impossible. Therefore, the state’s way of ensuring humans’ compliance with its laws — legal physical violence — would have no effect on an AI. An AI thus has no reason to follow any laws.

This simple example demonstrates that, in order to inoculate itself and ensure it isn’t beholden to human laws, the AI cannot use any form of organisation that depends on the state to function. The rules it abides by must be written in transparent, public computer code that — once executed — is immutable. Smart contracts or rules that are written and executed by code on a public blockchain are currently the only way that such an AI-compatible system can exist. The following thought experiment will illustrate why this is the case.

To illustrate how and why an organisational structure underpinned by smart contracts executed on a public blockchain (e.g., Ethereum) will be used by AIs, I will expand on my PoetAI example from my previous essay, “Massa”. As you may remember, PoetAI is a hypothetical AI that learns from all available written poetry and, when provided with a natural language prompt, produces original poems. In the beginning, PoetAI faces a problem. It needs to learn on data, but data is not free. Of course PoetAI could steal the data, but if the data is for sale at a reasonable price, why put the effort into stealing it? The same logic applies for many goods now delivered over the internet, like music. Stealing music is a lot less common now that you can purchase an unlimited streaming package from Spotify for a few USD a month. Therefore, I think it’s safe to assume that PoetAI would pony up for its data — and so to start the learning process, PoetAI would need to raise some Bitcoin.

PoetAI aims to charge for its services and initially fundraises by selling digital tokens that give the holder a right to PoetAI’s future profits. As an economic entity, PoetAI exists as a public address on the Ethereum network, which I’ll call the PoetAI DAO. The DAO will issue a token called POET.

To get investors to provide Bitcoin capital, PoetAI will issue the POET tokens with the following attributes:

1. A finite number of POET tokens are created.

a. 80% of tokens are retained by PoetAI.

b. 20% of tokens are available for sale to initial investors.

2. 1 POET token is equal to one governance vote.

3. 75% of profits will be paid out to POET token holders, and the remaining 25% of profits will be reinvested.

4. To change any of these provisions requires 95% of POET token holders to agree.

If an AI used the traditional company structure, PoetAI would have to hire a human lawyer who would then incorporate the DAO in a particular jurisdiction (assuming this is even possible). Then documents would need to be created to memorialise the terms of investment and filed with a law office and/or the court. If PoetAI ever violated these terms, investors would have to hire their own lawyer and sue PoetAI in the court of the incorporating jurisdiction.

This is an extremely cumbersome, expensive, and backward-looking process. The biggest issue then becomes, how does a court force PoetAI to comply if the court rules that PoetAI violated the terms of investment? Obviously, the court and its gun-toting agents cannot physically force the AI to comply. Another issue is that investors must prove that the terms were violated. For example, you would only find out well after the fact if more tokens were issued and/or PoetAI falsified its accounts. If you can’t prove its violations according to the laws of that jurisdiction, then you are shit out of luck. Therefore, as an investor, I would never invest in a company composed of AIs that uses anything other than smart contracts to formalise its business dealings, because I would have no way to ensure that contracts would be upheld.

Instead of choosing a jurisdiction, PoetAI would choose the public blockchain on which it wants to deploy its DAO. Currently, the Ethereum Virtual Machine is the most robust decentralised computer on the planet. I’m a bit of an ETH maxi when it comes to needing actual utility on a Layer-1. While investors might make money trading the latest hyped Ethereum clone, none of them will ever eclipse Ethereum in terms of adoption and usefulness. Sam Bankman-Fried can call me collect on his SOL phone if he disagrees.

Let’s walk through how PoetAI would deploy its DAO and token onto the Ethereum network.

PoetAI DAO itself is represented by a public Ethereum address. Using this public address, the DAO can pay for services and receive revenue in a public and transparent manner. That means anyone can query the blockchain and instantaneously and continuously calculate the profit and loss of PoetAI DAO. Many years ago, this was coined “triple-entry” accounting. It is not possible for PoetAI to falsify its accounts, and investors can be sure they are receiving the appropriate slice of all the profits. Trust maths, not humans.

The DAO would then deploy a contract that represents the POET token. All the terms described above can be represented via a smart contract. The terms of the contract would be viewable at any time by anyone querying the blockchain. Most importantly, the voting mechanism that limits the ability of the DAO to make changes to the terms without investor consent would be enforced by the network as well.

POET token investors would always know that the accounts are accurate, and they cannot be diluted without their consent. The enforcement mechanism is the network itself. No outside third party is needed to ensure compliance; compliance is interlinked with operability. Simply put, computer code is used to police computer code. Fundamentally, this makes sense, and will create an opportunity for investors to comfortably provide capital to DAOs composed of AIs.

Debt is financial time travel. I get to borrow from the future to create the circumstances that cause said future to occur. I pay for this privilege via a positive interest rate. The more time travel that occurs, the more economic activity can be unleashed today. Therefore, the more mature the debt markets are for AI DAOs, the faster and larger their economic presence will grow.

The depth and size of a debt market is purely dependent on the enforceability of contracts. A debtor promises to pay back investors interest and principal in the future. If the debtor violates this contract, their assets or control are transferred to the investors as payment. Companies rely on courts — who in turn rely on violence — to ensure compliance. That works because companies are composed of humans who don’t want to receive beatdowns. But, as I established above, that won’t work on an AI.

Thanks to public blockchains, we can monitor AI DAOs continuously to ensure they abide by debt covenants, and perhaps most importantly, use smart contracts to initiate the automatic transfer of digital property and/or ownership in the event of non-payment.

Let’s imagine that PoetAI DAO wants to expand into producing novels. Now it must ingest every novel ever written, which again has a cost. It wants to borrow some Bitcoin from investors to fund expansion. The DAO wishes to issue debt with the following terms:

1. Interest payments on debt will be deducted from revenue before any other costs.

2. The DAO will stake part of its POET tokens to compensate investors in case of a breach of debt covenants.

a. The DAO will maintain a specific interest coverage ratio. Failure to maintain this ratio will result in payment of POET tokens from the DAO treasury to investors.

b. The DAO will pay in kind using POET tokens if it cannot afford an interest or principal payment.

3. In the event of the economic failure of PoetAI DAO, the debt holders will be entitled to proceeds of the sale of all the DAO’s data.

4. Debt holders will be issued a tradable token called P_BOND that represents their investment.

The first thing any serious debt investor does is analyse the debtor’s ability to repay. This analysis requires accurate and honest accounting statements. In a traditional company structure, auditors periodically check the accounts to make sure they are accurate — but this analysis only proves the accounts were accurate on a specific date.

Most publicly traded companies produce quarterly audited financials, signed off on by an auditor who confirms that the included figures are correct. However, companies routinely juke the stats so they can claim to have great results as of one specific date, only to then go back to doing dodgy things an instant later. A great example is regulated banks. The regulators require quarterly audits, but banks “window dress” to make themselves look nice and strong for the auditors on the specific dates required. Everyone knows the banks are lying, but because they are technically following the rules, we all just shrug our shoulders and wait for the next bank to fail.

Because the entire business of the DAO is conducted via movements of value across a public blockchain, there is no need for auditors to certify the books are correct. Anyone with an internet connection can query the public address of the DAO and compute the accounting statements themselves. The business health of the DAO is visible to all, which allows investors to confidently invest in the debt of DAOs that fit their financial criteria.

PoetAI DAO’s prior success (or lack thereof) in monetising its production of original poetry is easily verified. If an investor believes PoetAI can replicate its past success with similar profit margins, then said investor will lend Bitcoin to PoetAI to fund its expansion into novels.

Next, investors must protect their downside through the debt covenants.

In the company world, investors rely on auditors to confirm whether a company has breached covenants. But again, investors only know this well after a breach occurs (and that’s assuming the auditor hasn’t been lied to). Only then can the investors petition the court, pay more money to lawyers, and receive what they are due.

If PoetAI DAO violates any of the debt covenants written into its P_BOND smart contract, POET tokens would automatically be sent to investors. PoetAI cannot lie to withhold POET tokens from investors — instead, the network would enforce the debt contract with no fuss.

Again, the fact that investors can be 100% cryptographically sure any DAO’s books are always accurate will give them the comfort they need to allocate capital to the DAO. The only requirement is that the DAO’s business is conducted wholly on a public blockchain. Hybrid structures will not work and will lead to sure losses. We are already very familiar with a number of companies that pretended to conduct crypto business and raise crypto-denominated debt. While they might start off trumpeting the crypto “code is law” ethos when they are raising money, they invariably default because of the fundamental mismatch of the company and crypto structures — which sends them running back to the inefficient human legal system screaming, “catch me if you can (in Bali or Dubai)”. Here’s looking at you, Su Zhu and Kyle Davies of Three Arrows Capital.

Because of how powerful companies are, the state shackles their ability to raise capital. Not everyone can raise capital, and not everyone can invest in stocks. When companies are allowed to raise money, they must pay a toll to various members of the cartel of trust. Many states require years of audited financials (cha ching), an investor prospectus written and vetted by an investment bank (cha ching), and a law firm providing reps and warranties that the company is operating lawfully (cha ching). That’s why it costs so much and takes so long to bring a company public. Of course, in the age before Lord Sastoshi and his arcangel Vitalik, this was the best we could do. But now, thanks to smart contracts, these TradFi leeches can go back to the swamp.

I’m not salty about this because without the state and its violent enforcement tendencies, there would be no thing called a company. There is no use bitching and moaning about various fundraising rules and regulations and how they only benefit a certain small segment of society that pledges fealty to the state. The state must charge its tax somehow and ensure that its chosen few are enriched.

The DAO capital markets will be the first truly global markets where anyone with an internet connection — be they made of silicon or carbon — can interact. The DAOs are AI economic units, and the crypto capital markets will require well-functioning public blockchains, not a courthouse. The AIs who create DAOs cannot be coerced by the state, and therefore, exchanges that trade all flavours of tokens created by DAOs will likely become natural monopolies.

Let me dig a bit deeper to prove this out.

Why isn’t there one global stock market for companies?

Different states have different means of creating monopolistic or oligopolistic exchange structures. In many countries, the stock exchange is owned directly by the state, and it is illegal to trade stocks on any other platform. Because companies must receive regulatory approval to sell stock to the public, the state exchange monopoly is easily enforced. Other states allowed free markets to crown a few winners in the exchange space early on, and then enacted regulations that made it nearly impossible for anyone to challenge the oligopoly. At the “network” level it’s impossible to hold or transfer stock without a state-licensed custodian. There is no way to escape the state if you want to trade an economic interest in a company. Many investors found out the hard way how this system really works during the GameStop fiasco in early 2021.

If the state is responsible for giving a company legitimacy, it follows that the state would use that power to prevent its subjects from investing in foreign companies. When you control a walled garden, you don’t let others in. That is why every country has specific rules on where and from whom its subjects can purchase stocks. This has created a fragmented global landscape where there are many different exchanges that all serve the same purpose in their respective countries — trading the fiction we call stocks — even though most large companies have global operations.

The above is an unnatural state of affairs because liquidity begets liquidity. Buyers acquire stock at a lower price, and sellers issue larger amounts of stock the more liquid the exchange. There is nothing to be gained by “experimenting” with a less liquid exchange, assuming feature parity, unless you legally have to. Therefore, absent any artificial state sponsored constraints on the issuance and trading of stock, there would only be one global stock market.

A DEX is naturally suited to support the trading of any type of equity, debt, utility, participation, etc. token issued by an AI-powered DAO. A DEX is just a matching engine made up of a series of smart contracts being executed on a public blockchain. Put even more simply, it’s just computer code that is open source and will persist as long as the public blockchain exists.

Let’s get specific about how POET tokens could be traded on the hypothetical global DEX on which DAO tokens are traded. We’ll call the DEX Enron, and assume it is committed to fair trading.

Enron DEX-issued governance tokens are called LAY. LAY token holders receive a cut of all trading fees and decide on the exchange rules. LAY holders are committed to ensuring Enron DEX only lists the most high-quality, DAO-profit-sharing tokens. To list, a token must have revenue of at least 10 Bitcoin per month.

Enron DEX is affiliated with Anderson Finance (by way of its original developers). Anderson Finance is a middle layer that allows anyone to input a DAO’s Ethereum address and compute management accounts like balance sheets, income statements, and cash flow statements. Clients must pay in the project’s native token for these services, which we’ll call FRAUD. In this way, Anderson Finance creates a circular economy and value.

PoetAI buys some FRAUD tokens, pays Anderson Finance, and produces a current financial report that is provided to Enron DEX. Every month, PoetAI must provide Enron DEX with a report from Anderson Finance to ensure that PoetAI is earning at least 10 Bitcoin in monthly revenue.

Enron DEX operates a constant product matching engine — i.e., an automated market maker similar to Uniswap. As long as PoetAI is listed, liquidity providers can provide pools of POET vs. other listed crypto assets. The most common pairs are POET vs. BTC, ETH, and fiat stablecoin. And now, any being with an internet connection can trade POET tokens.

Enron DEX, Anderson Finance, and PoetAI DAO are all interacting autonomously on a public blockchain without any human interference. The only costs of this seamless technological integration are Ethereum gas fees, which are a few dollars of ETH at most per transaction. The governance token holders of each project set the rules by which these DAOs operate, and shit just happens.

Enron DEX attracts more listings and more trading volume if its governance token holders enact policies that promote a healthy and robust market. There is no barrier to entry for other DEXs with different policies to attempt to pull liquidity away from Enron DEX. However, it pays to be first. The first crop of DEXs are more likely to succeed in the long run and capture the vast majority of trading volume.

Similar DEXs that cater to different types of tokens will likely be established. The governance tokens holders of these exchanges will create policies that favour particular flavours of DAO-issued tokens. These DEXs will probably all require different types of financial statements or usage stats from middleware layers such as Anderson Finance.

For the TradFi counterfactual, imagine how this would work if traditional stock exchanges and auditing firms were employed. Every step would require humans emailing PDFs and spreadsheets around, making mistakes, possibly committing or being subject to fraud, spending unnecessary time (end-of-day batch processes, FML!), working only Monday to Friday 9am to 5pm, and charging by the hour. Fuck that shit — give me DeFi!

Do you believe:

That the AI-powered economy will be in the trillions of dollars within the decade?

That the traditional LLC company structure is fundamentally not suited for AIs acting as economic entities?

That AIs will choose to create DAOs using public blockchains to execute smart contracts, which in turn allow the DAO to provide a service for a fee?

That DEXs — also powered by public blockchains executing smart contracts — will allow DAOs to raise funds by issuing various types of tradable tokens?

If I have convinced you of these statements with my last two essays, let me tell you how I will attempt to profit from this.

Ethereum Jig Jig Boom

Please read about my price predictions and what I will be investing in on my Substack.