Banks Won’t Open Accounts For AI Agents So 100 Billion Bots Will Live On Crypto Instead

Everyone’s debating whether humans will adopt crypto. Nobody’s talking about the actual adoption wave already happening: machines.

The Adoption Story Everyone’s Got Wrong

For fifteen years, the crypto industry has obsessed over one question: when will humans adopt crypto en masse?

Wrong question.

Animoca co-founder Yat Siu just pointed out something far more consequential: banks won’t open accounts for AI agents. So somewhere between 50 and 100 billion bots will operate on crypto wallets and stablecoins instead.

Not humans. Machines.

This isn’t a future prediction. This is already happening. And it changes the entire premise of what “crypto adoption” actually means.

Why Banks Can’t Serve AI Agents

Think about what opening a bank account requires: identity verification, legal personhood, a Social Security number or equivalent, a physical address, a human signature, regulatory compliance frameworks built around individual humans or registered legal entities.

An AI agent has none of this. It’s not a person. It’s not (yet) a recognized legal entity in most jurisdictions. It doesn’t have a passport. It doesn’t have a fixed address. It might exist as code running across distributed servers with no single point of identity.

Traditional banking infrastructure simply wasn’t built for this. KYC (Know Your Customer) protocols assume a customer who is, definitionally, a human or a registered company. An autonomous AI agent breaks that assumption completely.

So when AI agents need to transact, pay for compute, pay other agents for services, settle accounts, manage budgets, they literally cannot use a bank account. The infrastructure doesn’t exist. The regulatory framework doesn’t exist. The bank wouldn’t even know who to call if something went wrong.

Why Crypto Solves This By Accident

Crypto wallets don’t require identity verification at the protocol level. A wallet is just a cryptographic key pair. No bank manager has to approve it. No KYC process has to validate a human behind it.

This was originally designed for human privacy and censorship resistance. Nobody built crypto wallets specifically for AI agents.

But the architecture happens to solve exactly the problem AI agents have: it allows any entity, human or machine, to hold and transfer value without needing institutional approval of its identity.

Stablecoins compound this. They’re programmable, instantly transferable, and don’t require the agent to navigate currency conversion or cross-border banking friction. An AI agent in any jurisdiction can hold USDC and transact with another AI agent anywhere else, instantly, without a bank in the loop.

This wasn’t crypto’s intended use case. It’s becoming crypto’s accidental killer app.

What “100 Billion Bots” Actually Means

Let’s sit with that number for a second.

100 billion is roughly 12x the world’s human population. If even a fraction of projected AI agent deployment happens, agents handling customer service, executing trades, managing supply chains, negotiating contracts, paying for API calls, settling micro-transactions between other agents, the volume of machine-to-machine value transfer could dwarf human transaction volume entirely.

Every agent that needs to pay another agent for a service, every agent that needs to pay for compute resources, every agent that needs to settle a transaction on behalf of the human or company it serves, all of that needs rails. And those rails can’t run through traditional banking because traditional banking wasn’t built to onboard a machine as a customer.

So the rails being built right now are crypto rails: wallets, stablecoins, on-chain settlement, smart contracts that execute agent-to-agent agreements without human intervention.

The Real Story Nobody’s Telling

The crypto industry spent a decade trying to convince retail users that Bitcoin would replace cash. It didn’t work the way they hoped. Volatility scared people off. Complexity scared people off. Most humans still prefer their bank app.

But while everyone was trying to convert grandma into a crypto user, an entirely different category of user emerged that doesn’t care about volatility, doesn’t get confused by seed phrases, doesn’t need a friendly UI, and doesn’t have the emotional hangups humans have about money.

AI agents don’t panic-sell. They don’t get scared by red candles. They don’t need education about “not your keys, not your coins.” They just need programmatic access to value transfer that doesn’t require a bank’s permission.

Crypto, it turns out, might be a far better fit for machine commerce than human commerce.

Who’s Already Building For This

This isn’t speculative anymore. The infrastructure race is underway.

Coinbase’s “Coinbase for Agents” platform lets AI agents execute crypto trades and payments autonomously. Whatever the risks (and there are real ones, as I wrote about previously), the demand signal is clear: someone built this because agents need it.

A new venture called t54, founded by an ex-Ripple engineer and backed by Ripple and Franklin Templeton, is building a trust layer specifically to verify and insure AI agents that spend money autonomously. That’s not a hypothetical product. That’s $5 million in funding chasing a problem that exists today: how do you trust an autonomous agent with financial authority?

Stablecoin infrastructure is being explicitly redesigned around agent-to-agent payments, not just human remittances or trading.

This is an arms race nobody’s covering with the seriousness it deserves, because it’s not as exciting as a token price chart.

The Implications Are Bigger Than Crypto

If AI agents become primary economic actors, negotiating, paying, settling, transacting at machine speed and machine scale, the implications go well beyond whether Bitcoin hits a new all-time high.

Regulatory frameworks built for humans will need to expand to cover non-human economic actors. Who’s liable when an AI agent commits fraud, makes an error, or gets hacked mid-transaction? Existing law doesn’t have clean answers, because existing law assumes a human or a registered company is on the other side of every transaction.

Trust and verification become the actual product, not the transaction itself. When humans transact, identity and reputation are partially handled by institutions (banks, governments, credit bureaus). When agents transact, that entire trust layer has to be rebuilt from scratch on-chain.

Economic activity could scale beyond what any human institution monitors in real-time. If 100 billion agents are transacting, the volume and speed of that activity will exceed what any regulator, bank compliance team, or auditor can review using current methods.

This is the actual frontier. Not “will retail adopt Bitcoin.” It’s “what happens when the majority of economic transactions on a blockchain aren’t initiated by a human at all.”

Why This Should Worry You (A Little)

Every wave of financial infrastructure built without adequate oversight eventually creates a crisis. The 2008 financial crisis happened partly because complex financial instruments outpaced regulatory understanding. Algorithmic trading caused multiple flash crashes because automated systems moved faster than human circuit breakers could respond.

Now we’re building financial rails specifically because traditional, regulated banking infrastructure refuses to onboard the new class of economic actor. That refusal isn’t banks being lazy. It’s banks correctly identifying that they don’t have a framework for verifying, insuring, or holding accountable a non-human customer.

So the activity moves to an environment with even less oversight: crypto rails, where verification is optional, accountability is unclear, and the volume could eventually dwarf anything traditional finance has dealt with.

That’s not necessarily catastrophic. But it’s not nothing either.

The Question Nobody’s Answering

If 100 billion AI agents are transacting on crypto rails, processing value at a scale that exceeds human economic activity, who exactly is responsible when something goes wrong at that scale?

Not “which agent made the bad trade.” The systemic question: who governs an economy where the majority of participants aren’t human, aren’t legally accountable in any traditional sense, and exist as code that can be duplicated, modified, or shut down without the due process we’ve built around human economic actors?

Nobody has a good answer yet. Because until recently, nobody thought this was the actual adoption story.

What This Means For Crypto’s Future

Crypto spent years trying to be money for people. It might end up being money for machines instead.

That’s a stranger, less romantic story than “the people’s currency” or “the future of finance for everyone.” But it might be the more accurate one.

The next decade of crypto’s relevance may not be determined by whether your aunt buys Bitcoin. It may be determined by how many AI agents need a wallet, how fast they transact, and whether anyone builds the accountability infrastructure before the volume becomes unmanageable.

100 billion bots are coming. The rails are already being built. And almost nobody is asking the right questions about what happens next.

If AI agents become the majority of economic activity on crypto rails, should they have the same legal accountability as humans? Or do we need an entirely new framework? Drop your take.