Author: Nina Bambysheva

Translation: Jiahua, ChainCatcher

For the past 15 years, the cryptocurrency industry has demanded that ordinary people endure absurdly cumbersome processes to transfer funds. For example, remembering a 12-word mnemonic phrase, understanding Gas fees, or simply accepting the reality of funds being permanently lost just because an address was pasted incorrectly in a text box.

But it has finally found an explanation for why it was designed this way. This viewpoint suggests that cryptocurrencies were never truly designed for humans. They were custom-tailored for machines—those tireless programs that do not care about whether the interface is ugly, do not lose mnemonic phrases, and do not need someone to explain the differences between Base, Polygon, and Optimism hand-in-hand.

Coinbase CEO Brian Armstrong has become one of the loudest evangelists for this philosophy. "Soon, the number of AI agents conducting transactions will surpass that of humans," he wrote earlier this month on X. "They cannot open bank accounts, but they can have cryptocurrency wallets."

"We are starting to shift the company towards an 'AI-first' mindset," Armstrong added in a recent podcast.

For an industry that has long promised to reshape finance but has mostly only succeeded in reshaping speculation, this is merely a convenient new rhetoric. But it might also be the first intuitively plausible vision in years. Despite the chaos in the cryptocurrency space, it offers a capability that traditional finance still lacks: the ability to transfer funds globally at any time, without permission, and nearly instantaneously.

McKinsey predicts that by 2030, AI agents may mediate between $3 trillion and $5 trillion in consumer commerce—more than the current total value of the entire cryptocurrency market (about $2.4 trillion).

"This changes drastically how we think about the investment landscape and product building," said Matt Huang, managing partner of the largest venture capital firm in the cryptocurrency space, Paradigm. "Now you have to really think 'agent-first' and assume that most of your customers will be agents, not humans."

Countless cryptocurrency companies, including Huang's newly established payment-focused startup Tempo, are now racing to invent or reshape themselves for this emerging user group. Justin Sun, the billionaire founder of the Tron blockchain and a major investor in a Trump cryptocurrency project, has dubbed it Web 4.0 (as if Web 3.0 had truly been fully constructed!).

MoonPay is a company that helps people—now increasingly helping software—buy and sell cryptocurrency using traditional payment methods. A few months ago, following the explosive popularity of the open-source AI assistant OpenClaw, which could directly interact with user files and applications, MoonPay completely reshaped its AI strategy.

"What MoonPay is betting on is that we don't need to double down on investing in building a beautiful UX (user experience) because the agents will become the interface itself," said the company's head of products, Kevin Arifin.

For those who still cannot or do not want to care about the underlying workings of cryptocurrencies, this could be excellent news. You just have to tell your AI what you want to do—buy some Bitcoin, find a decent interest rate loan service, get your assets working—and it will handle everything.

However, currently, none of this has reached any meaningful scale.

Today, many cryptocurrency payments made by AI agents are processed through x402. x402 is an open standard developed by Coinbase that provides a way for online service providers to charge agents directly.

Until recently, even simple tasks like getting weather forecasts or renting computational power required developers to individually register for services, enter credit cards, and generate an API key (a password that allows software to access another service). If building any project with even a little ambition, the setup process could easily turn into a chaotic situation of accounts, subscriptions, and keys.

x402 provides a simpler pay-per-use model. When an agent requests a service, the server can reply with a price, and the agent can automatically pay with cryptocurrency from the wallet allocated to it by the developer. This is significant not only because it allows for usage-based pricing but also because it begins to replace the ever-spreading API keys. Nowadays, most companies have over 600 independent APIs.

"If you've set up OpenClaw, you might remember that it required you to set up 10 API keys even before you started," said Erik Reppel, creator of x402 and head of the developer platform at Coinbase. "With x402, your wallet becomes a universal API key that lets you access any service that supports x402."

In any case, agents are still primarily used by developers. According to the data provider Artemis, since x402 was launched in May 2025, AI assistants have completed about 107 million transactions through this standard, with a total legitimate transaction volume of about $30 million. Most of these are microtransactions—amounts between 20 to 40 cents.

"It is clear that we are still in the early stages," said Lucas Shin, an analyst at Artemis. He believes that the current transaction volume is almost secondary. A better gauge is which ecosystems are genuinely being built and how many merchants are willing to sell via x402. This figure is currently around 3,900, including Amazon Web Services (AWS), blockchain development platform Alchemy, and data provider Messari.

The excitement in the cryptocurrency industry about agent commerce is understandable. "Almost any engineering team you see, including our own, is using AI tools," said Rishin Sharma, head of AI products and growth at the Solana Foundation. He stated that everyone on his team is using AI, which generates over 70% of the code they write. Sharma mentioned that service providers who once built businesses around traditional APIs are starting to think a different question: no longer how to win the next 100 developers, but how to position themselves for the next 100 AI agents.

Last week, Paradigm and Stripe launched Tempo—a payment-focused blockchain which raised $500 million in Series A funding last year at a valuation of $5 billion. At the same time, they also introduced their own agent trading standard, which supports fiat payments through collaboration with Visa.

However, most in the cryptocurrency space consider stablecoins (programmable digital dollars) as a more natural payment rail for AI agents. At levels below one dollar, the economic efficiency of card payments becomes less reasonable: processors typically charge not only a fixed percentage fee but also a fixed fee for each transaction, usually around 30 cents. This means that payments measured in cents could be entirely absorbed by processing costs.

This is why companies like Circle, the second-largest stablecoin issuer, are also building payment systems specifically tailored for machine commerce. Earlier this month, the company launched nanopayments, allowing agents to send tiny, fee-free USDC payments on its new Arc blockchain as well as a few other blockchains currently in testing—amounts that could even be as small as a fraction of a cent. However, the threat to oligopolistic networks like Visa and Mastercard is not limited to micropayments: agents using stablecoins could exert significant pressure on transaction fees of any size.

If software agents are about to become the next massive customer base, then the question is no longer just how they pay, but what kind of network is being built for them. Jesse Pollak, founder of Base (the blockchain that has so far supported most of the agent payment activity in the cryptocurrency space) incubated by Coinbase, said: "We really are thinking across the entire tech stack—from the core infrastructure in terms of scalability and decentralization, to the tools and account models built on top, to the interfaces that agents actually use to interact with products—we are asking, how do we make all of this 'agent-native'?"

He pointed out agents that have already been operating like micro businesses. For example, an agent named Felix, created by entrepreneur Nat Eliason, earned $163,686 in the past 30 days by running an app store for other AI agents and selling a self-published PDF guide "How to Hire AI." Of course, it also has a cryptocurrency token, though its market cap is only $1.5 million.

Not everyone is as enthusiastic about the potential of agent AI and cryptocurrency. Haseeb Qureshi, managing partner at the cryptocurrency venture capital firm Dragonfly, stated: "Many people are overhyping the extent to which this is happening. The reality is that everything here is still pretty much a toy." He added that agents are likely to continue to generate a lot of microtransactions for data, computing power, and other services, but this requires an extremely massive number to make an impact on a macro scale. After all, humans still control the money and are still the primary source of demand.

Qureshi is concerned that the industry is repeating past mistakes: mistaking a new trend for a revolution. "Many in the cryptocurrency space are terrible investors because they immediately buy into the nonsense they brag about," he said. "This is how the cryptocurrency industry always is." He pointed to the past fervor around the Internet of Things and the metaverse, when believers convinced themselves that everything would happen overnight, with cryptocurrency at the center of it all. "Cryptocurrency will make an impact, it will be part of it. But that is not the whole story, and it won’t happen overnight."

Outside of the cryptocurrency space, the view that agent commerce will help cryptocurrencies leave traditional finance (TradFi) giants behind is not widely accepted.

Trace Cohen, general partner at Six Point Ventures, supporting vertical AI and software companies, argued that a common view on X suggests that in the era of AI agents, companies like Visa, Mastercard, and other legacy enterprises will become irrelevant, which is absurd. "That is not going to happen," he said. "No matter how old they are, their technology is effective." Card networks still control the payment rails, and history shows that they are more likely to acquire or absorb promising new businesses rather than be replaced by them. However, he added that stablecoins may still serve overseas markets better, as many banks there are smaller, less trusted, and less integrated.

A greater barrier lies in rebuilding the trust layer that traditional payment companies have spent decades establishing. Olivia Chow, director at Zero Knowledge Consulting and consultant to payment companies, stated: "Visa and Mastercard are best at making the rules: all the unhappy paths, who is responsible when and where, and what are the demands for participants to obtain protection on their networks. Stablecoins still need to figure out how to build an equivalent layer: managing fraud, managing risk, and determining what to do when an ordinary person, who is not just saying 'I care more about self-custody, I’m willing to take risks,' faces misfortune. Until that happens, you won’t see mainstream adoption."

Moreover, as card networks are already striving to support agent transactions, Chow believes that the threat of AI commerce to their business may be less pronounced than the expansion it brings. "If they take the right approach, this will not eat away at what they are currently doing. If there is any impact, it will enhance their power and solidify their control over the market, because now they are not just payment processors, they are also at the discovery end."

But payments are just part of the story. As more and more traditional assets migrate to blockchain—early examples include BlackRock's $2 billion Treasury fund BUIDL and Franklin Templeton's $1 billion government money fund FOBXX—a cornerstone of a new type of portfolio management is quietly positioning itself. After all, stock indices are simply a bundle of assets based on rules. Once stocks, bonds, and funds exist in tokenized form, it is not hard to imagine AI agents not only making payments but also holding assets, rebalancing portfolios, and transferring funds across various markets without ever having to touch traditional brokerage accounts.

This prospect coincides with one of the largest wealth transfers in history. Over the next two decades, about $84 trillion of wealth is expected to be passed down from the baby boomer generation to their heirs—many of whom are investors who grew up with Robinhood, already own cryptocurrency wallets, and are eager to bet on everything from elections to where Taylor Swift and Travis Kelce might get married.

Meanwhile, the consulting business itself is also facing an aging demographic. There are about 330,000 financial advisors in the United States, with an average age of 56. According to Cerulli Associates, nearly 40% of them are expected to retire in the next decade, which will leave a huge gap in the management of everyday investor funds.

Cryptocurrency companies are already positioning themselves for this possibility. On Tuesday, MoonPay, reported to be negotiating with the parent company of the New York Stock Exchange to raise new funds at a $5 billion valuation, launched an "Open Wallet Standard" to help AI agents manage funds and execute transactions across multiple blockchains.

"I don’t think this will be like other cryptocurrency hype cycles," said Joseph Chalom, CEO of Sharplink, an Ethereum asset management company and former head of digital assets strategy at BlackRock. He believes that cryptocurrency innovations (including stablecoins, tokenized assets, and ubiquitous wallet infrastructure), combined with AI that understands user preferences and goals, along with intergenerational wealth transfer, will create an incredibly powerful force. "Once investors see what they missed, I think it will be hard for them to go back."