Scrolling normally through X, until you find a tweet that could easily go unnoticed—but your brain suddenly realizes the magnitude of what was written. It was from Brian Armstrong, CEO of Coinbase, one of the largest exchanges in the United States and also in the global top ten.
They can’t open a bank account, but they can own a crypto wallet.
At first it doesn’t make you think much. But then you read it again, and something about it feels unsettling. The interesting part is not when he talks about cryptocurrencies. It’s when he talks about who participates in the economy.
And that changes the entire game. Because…think about it, our economic system assumes that financial transactions are administered by a person—or a group of people: institutions, banks—operating with some form of human logic. Inside the traditional financial system, we require:
- Some kind of ID
- KYC
- Legal responsibility
- Documentation requested by institutions
In other words, we assume from the beginning that the financial system is built on a human premise. Until artificial intelligence (AI) arrives with a very interesting proposition: that human entities are merely players inside a new ecosystem where the economy can move without necessarily being human.
Think about it for a second. An AI agent could buy data from you, pay for API access, hire computing power, pay for storage, buy information and content, and execute financial operations. What am I saying? AI can participate in markets. The only thing it cannot do is open a savings account.
Armstrong’s pointArmstrong’s point
And this is where things become interesting. A self-custodial wallet does not require permission.
You only need:
- a private key
- an address
And that’s it.
No approval.
No identity verification.
No institutional authorization.
Or, put more directly:
an AI can own a wallet. It can generate its own keys, receive payments, send payments, execute contracts. At that point we are no longer talking about money alone, but about economic infrastructure for autonomous software.
When machines interact with machinesWhen machines interact with machines
If AI begins interacting with other AI systems, a new element appears: markets where machines negotiate with other machines. It’s similar to how AI is already used to improve the code of other AI systems.
An agent could:
- Pay cents for a dataset
- Pay fractions of cents for information queries
- Pay for cloud computing resources
- Access algorithms
- Use distributed storage
These payments are micropayments, so small that traditional financial institutions cannot even process them efficiently. This is where micropayment systems become powerful: they allow micro-transactions between autonomous agents. Opening the door to a completely new type of economy:
M2M — machine-to-machine.
The change is social, not technologicalThe change is social, not technological
The real change here is not technological. It’s social. We already mentioned that the traditional financial system rests on three pillars:
- human identity
- intermediaries
- legal jurisdiction
But when an AI agent operates without a bank, without civil identity, and without state permission, we are suddenly dealing with something new: an economic actor that the traditional system never anticipated.
What Armstrong’s tweet suggestsWhat Armstrong’s tweet suggests
So what does Armstrong’s tweet really imply?
It suggests the possibility that AI agents could become more active participants in markets than humans themselves. And honestly, that doesn’t sound crazy. A single user could deploy:
- trading bots
- assistants purchasing information
- programs running to manage that information
- systems paying for the computing resources needed to run them
Do we realize what this means?
Economic traffic could be generated by software systems. In an M2M economy, machines buy, sell, and negotiate with other machines. Meanwhile, we watch and ask three questions:
- Who controls those agents?
- Who is responsible for their decisions?
- To what extent do they actually represent human interests?
BitcoinBitcoin
Now we arrive at the central point.
The internet has never distinguished between humans and machines.
A server receives information from both in the same way it transmits emails or data packets. The network simply moves data.
If money becomes native to the internet, that money also doesn’t distinguish between humans and software.
A wallet can belong to a person or to a program. ..Or to thousands of them.
A deeper implicationA deeper implication
Bitcoin and other cryptocurrencies are often described as a way to free money from banks. But perhaps we are not yet seeing their most radical implication. A monetary system where humans and machines participate under the same rules. That leads us to a new paradigm.
One that feels slightly uncomfortable. Because the real question may no longer be:
Who controls the money?
But something far more unsettling:
Who—or what—is actually participating in the market?
https://twiiit.com/brian_armstrong/status/2031021867973194172
As an aside, one of the real strengths of LLMs is the ability to deal with lots of boilerplate and mundane task that humans find taxing.
Regarding Bitcoin, I think this may play out advantageously with regards to LN. The one issue that has hampered LN self-custody is its a bit daunting for average person to setup and manage all the minutiae...balancing channels, etc.
However, this is exactly the type of work that LLMs excel at... would be great if we eventually got some specifically trained SLM's...maybe 300M - 500M parameters that were exclusively trained on managing bitcoin / LN nodes.