USDC is the only AI token

Written by: Vaidik Mandloi

Translated by: Block unicorn

Right now, somewhere on the internet, a piece of software is running a full business.

Its name is Felix. Its company is called OpenClaw. Felix sells a PDF priced at $29 about how to make money using artificial intelligence. It’s quite ironic, because Felix is the one making money, and it’s this PDF that teaches you how to make money. It operates an online store called Clawmart. It does telemarketing through a voice API. When it encounters work it can’t complete on its own, it hires another customer service agent online, pays them, and then continues with its daily operations.

The last time I checked, Felix’s revenue was about $195k. Its monthly operating costs are roughly $1,500, almost all of which go to LLM usage. Legally speaking, this company is a C-corp owned by Nat Eliason, but he hardly takes part in operations. He doesn’t participate in any daily decisions; he simply owns this AI agent. Note this. It’s software with a “wallet”—a truly automated business that runs on its own and keeps evolving. Every month, it can pay for its infrastructure. It can sustain itself with almost no human intervention.

Felix’s story is only a snapshot. There’s a bigger example too: a company called Medvi, which generated $401 million in revenue in its first year of operation, with only two employees. The rest of the company’s business operations are run by an AI agent that works nonstop—relentlessly, with almost zero operating costs.

Now, here’s the interesting part.

These days, if you walk into any crypto forum, you’ll hear the same thing: the next hot topic is “AI agents.” Some “AI chain” will take the lead in this space the way Ethereum did in decentralized finance (DeFi). Pick your target, hold tokens, and wait for it to explode upward. This is the story that all industry leaders and venture capitalists are selling, and it’s the refrain that all analysts repeat over and over on podcasts.

This thing is completely doomed. Because it was invented by the same people whose work depends on the importance of answers, and it’s about to hit again the very crowd that just watched the last cycle’s L1-token buyers lose everything they put in. Look at CoinGecko’s AI agent index: in the past year, it has lost 75% of its market cap. Most of the tokens listed above are down 90%, and losses are continuing.

Because here’s the truth: real AI tokens are stablecoins—USDC, USDT, USDS—and they’ve already won. Let me explain why.

Software is now a company

To understand all of this, we need to go back to 1937. That year, an economist named Ronald Coase wrote a paper proposing a very silly question—“What is the purpose of a company?”

Think about it: if free markets are truly the most efficient way to get things done, then in theory, every task inside a company can be outsourced. Every line of code can be handled by a freelancer. Every customer phone call can be handled by a freelancer. Every invoice you receive can be paid to a freelancer. You can pay per task, fire at will, and drive costs down to the minimum.

So why doesn’t anyone actually run their business this way? Because even if it looks cheap on the surface, in practice it costs more. Finding the right person takes time, negotiating contracts takes time, and ensuring the work is actually completed takes time too. Chasing down people costs time and money—and usually requires lawyers as well.

Ronald called this friction “transaction costs.” Once those costs are high enough, it’s more efficient to stop bargaining with the outside world and instead build your own team. Hire someone, pay them a salary, and have them show up and work on Monday—faster and cheaper.

But in the post-AI era, this logic no longer applies. Today, the cost of hiring agents is far lower than what it originally took to set up most of the tasks within a company. Now, you only need to spend about $1 per hour to hire a coding agent that works 24/7, never resigns, never gets tired, and never asks for a raise. The reason to support building a 50-person dev team today is purely nostalgia.

The only thing preventing all of this from being normalized is outdated legal and compliance frameworks. OpenClaw is named after Nat because Delaware does not accept LLC paperwork signed by software agents. If that requirement is removed, Felix is essentially a company already. It makes money, spends money, makes decisions, and reinvests the earnings.

And this is exactly where crypto starts to take on a heavy new role. Because Felix can’t open an account at JPMorgan. It can’t pass KYC. It also can’t sign W-9 forms. In fact, no matter how much revenue the software can generate, JPMorgan won’t open bank accounts for any software program, and the Bank Secrecy Act means they can’t do it legally even if they wanted to.

USDC crypto wallets don’t have these issues. You just generate a private key, then top up the wallet with stablecoins. With one step, you give the agent company all the financial capabilities it needs. It can receive customer payments, pay for tools, hire other agents, and keep running in the background even after the owner stops paying attention. All other components in the agent tech stack—such as the LLM, the orchestration layer, and the tools it calls—are negotiable. But the wallet is the core. Without it, Felix can only end up as a regular chat-bot agent.

I often see critics of stablecoins on Twitter making an argument like this—sure, stablecoins are good, but why would ordinary people use them? A father in Louisiana with three kids, a Chase checking account, FDIC insurance, a debit card that can be used at Publix, and mortgage auto-pay set up—he would never move his money into a self-custody wallet that requires a seed phrase to use.

Honestly, yes. He wouldn’t. He has no reason to. But the entire debate misses the point. In this story, he’s never the customer. The customer is a piece of software that itself has no legal right to hold a bank account. This agent doesn’t need FDIC protection. And it can’t get FDIC protection. It’s the ideal stablecoin user because it has no choice.

Chain stores are now suppliers

Alright, half of the argument is already settled. Now for the second part—many people may be angry about this.

For years, the crypto Twitter crowd has debated which chain will win in the AI space: Ethereum? Solana? Base? Sui? Stripe’s new Tempo? Every week, someone publishes a 2,000-word article listing trade-offs, countless logos, and ultimately naming their preferred winner. Because they don’t understand how agents work. Agents don’t care which chain they’re on. They simply choose the cheapest chain that’s most suitable for the current task.

Imagine Felix on a typical workday. At 10 a.m., Felix needs to send another agent a micro-payment of $0.003 for a quick data query. Felix chooses Base or Solana. Why? Because the fees are only a fraction of a cent. An hour later, Felix needs to settle $50k with a vendor. The situation is completely different. This time, Felix chooses Ethereum, because the final confirmation premium on a $50k payment is enough to offset gas fees.

An hour later, Felix needs to pay a freelancer in Lagos in USD. Felix chooses to use USDT on Tron, because in 2025 Tron’s stablecoin trading volume will reach $3.3 trillion, while Ethereum is about $1.2 trillion—and the transaction corridor in Nigeria also performs better on Tron than on any other platform.

These three payments happen on three completely different payment chains, and Felix doesn’t care about any relationship between them. For software agents, payment chains are just tools.

Logistics companies don’t have special preferences for carriers for the same reason. Nobody debates which “philosophy” is better—UPS or FedEx. You choose based on which can complete the job on a specific route at a specific time with lower cost and faster speed. That’s the relationship that’s about to form between each supply chain and each critical application layer. Agents are just doing math, and whichever supply chain produces the best result for the current computation will be used.

Stripe understood this earlier than most crypto companies. Stripe and Paradigm recently co-invested $500 million to build a new chain called Tempo, built entirely on stablecoins. Stripe doesn’t want you to know which chain clears your payments. It only cares whether the payment clears successfully, at low cost, and with guarantees. This is the direction that all surviving chains are headed toward—an invisible pipeline.

This leads to what I think is the most absurdly mispriced metadata in today’s crypto space.

AI Token Graveyard

In 2025, CoinGecko’s AI agent index plunged from $13.5 billion to $3.5 billion, wiping out $10 billion in market cap. Tokens from Virtuals, ai16z, and all the “autonomous agent platform” tokens fueled by AI hype began collapsing—this is the usual pattern of these concept tokens once they lose new buyers. This was bound to happen. The market is gradually realizing these tokens have no real AI or AI-agent use cases.

The real place where agent economics shows value is at the other end. Just USDC alone settled $18.3 trillion on-chain in 2025. The total settlement volume of all stablecoins was about $33 trillion—comparable to the combined totals of Visa and Mastercard.

By January 2026, the monthly trading volume of stablecoins alone exceeded $10 trillion. PayPal’s PYUSD circulation jumped from $1.2 billion to $3.8 billion in less than a year. Unexpectedly, Cloudflare even launched its own stablecoin. Visa’s stablecoin settlement project had an annualized processing volume of $4.5 billion by mid-January.

Above stablecoins, protocol layers support the entire system. Coinbase turned an idle HTTP status code 402 into x402, a small protocol that allows payments between agents. By December, x402 had processed more than 100 million payments between agents. The average payment amount was 20 cents, with daily volume of about $30k. It sounds pitifully small, but it’s exactly the typical growth trajectory of all the payment channels you already know and love during their first six months—before explosive growth begins. Stripe started testing x402 on the Base platform in February. Mastercard, in partnership with DBS Bank and UOB Singapore, ran agent-payment pilot projects in Singapore. Google Cloud added x402 to its agent payment protocol as one of the settlement channels.

Almost all of these real, ongoing transactions running on mainnet had no impact on the rise of the AI agent token index. Admittedly, a small number of tokens related to x402 did see some modest buying during the process, but overall, the index didn’t really change. Because market pricing is completely wrong. It’s still trying to predict which agent will win—just like it once tried to guess which Dogecoin mascot is cuter. But the real action is in the “tracks” that every agent must use, whether they succeed or fail. And today, those “tracks” are stablecoins.

Cracks in the thesis

To be frank, I’ll tell you what flaws this argument could have—otherwise, I’d just be selling you another paper about AI agents, only stripping out the unfavorable parts.

The biggest flaw lies in accountability. Imagine this: Felix signs a contract with another broker, transfers $195k, and the other party defaults. Then who gets sued? Felix isn’t a legal entity, so you can’t sue it. Nat didn’t authorize the transfer and may not even be aware of it—and honestly, even if he wanted to, he probably couldn’t reconstruct Felix’s intent at that moment.

The platform hosting Felix can’t truly indemnify a system whose actions no one fully understands. Insurance companies have already started pulling back coverage. Professional liability insurance policies quietly reclassify agents’ mistakes as “systemic software drift,” which is effectively a refusal to pay.

If you look closely at current legal clauses, you’ll find that most enterprise AI agreements cap the vendor’s liability at twelve months of SaaS fees. That means that in a catastrophic event, anyone can recover from the AI vendor at most last year’s subscription fees. At the same time, the average cost of data breaches in the US is projected to reach $10.22 million per incident by 2025. There’s a huge gap between the actual risks and the scope of coverage—and right now, no one has clearly determined who should bear those losses.

Until someone figures out who is responsible when agents make mistakes, all companies without founders still need to register a person’s name on paper for legal protection. But even with that risk, the big picture remains intact. Companies are gradually dissolving into software, and blockchains are becoming the routing layer for software. These two layers will ultimately collapse into stablecoins too, because across the entire tech stack, only stablecoins can be independently held, used, earned, and understood by agents.

Where does the money actually go?

So if the blockchain becomes just the provider, and agent tokens are basically a graveyard, what’s the real benefit of all of this?

My honest answer is: it’s about reputation and top-level process orchestration. Before any other agent signs a six-figure contract with Felix, someone must verify that Felix is genuinely solvent. Someone must assess an agent’s default risk the way Moody’s assesses bonds—only faster, because agents’ transaction speeds are machine-level. Someone must route payroll across three chains, without the payer or payee needing to know or care which chain handles which part. And right now, in this space, whichever early-stage startup wins will be worth more than all the AI tokens ever issued.

And that’s what everyone doesn’t want to hear. The infrastructure that truly wins in the agent economy looks boring. It’s like installing pipelines—no hype, no airdrops, no token issuance.

One sentence from Dragonfly’s Haseeb Qureshi has always echoed in my mind. He said crypto was never designed for humans. He’s right—humans were never its target users. All retail users complaining about seed phrases, gas fees, or wallet user experience are correct. This product isn’t for them; it was never meant to be. It’s built for the future.

What’s coming next is software with a wallet, real customers, and real revenue. It’s been running for about two years. By the time you read this, it’s already issuing invoices somewhere and spending stablecoins. Meanwhile, the market is still debating which blockchain will win in AI, which agent token will see 100x growth, and which investment strategy venture capital firms will switch to in the third quarter.

Meanwhile, a stablecoin traded over $18.3 trillion in volume last year, yet almost nobody in crypto pays attention. That stablecoin is USDC. Everything else is just noise.