Why is the crypto industry so enthusiastic about AI intelligent agents?

Written by: Nina Bambysheva, Forbes

Translated by: Luffy, Foresight News

Over the past 15 years, the crypto industry has made ordinary users endure extremely cumbersome processes. Just to complete a transfer, users need to remember 12 mnemonic words, understand Gas fees, and accept the reality that a wrong address paste can lead to permanent asset loss.

But now, the industry has finally found a narrative for this architecture: cryptocurrencies were not designed for humans from the start; their true users are machines. Those tireless robots don’t care about poor interfaces, won’t lose mnemonic phrases, and don’t need seasoned traders to explain the differences between Base, Polygon, and Optimism.

Coinbase co-founder and CEO Brian Armstrong is one of the most enthusiastic advocates of this idea. Earlier this month on X, he wrote: “Soon, the number of AI agents making transactions will surpass humans. They can’t open bank accounts, but they can hold crypto wallets.”

He added in a recent podcast: “We’re starting to implement an ‘AI-first’ mindset across the company.”

For an industry that has promised to rebuild finance but has mostly only reshaped speculation over the years, this is a rather clever new narrative. But it might also be the first story that truly makes intuitive sense. Despite the chaos in the crypto space, it offers capabilities that traditional finance still lacks: permissionless, near-instant, global, 24/7 fund flows.

McKinsey predicts that by 2030, AI agents will drive a consumer commerce market worth $3 trillion to $5 trillion, surpassing the current total crypto market cap of about $2.4 trillion.

Matt Huang, managing partner at Paradigm, the largest venture capital firm in crypto, said: “This greatly changes how we think about investment patterns and product development. Now you have to design with an ‘agent-first’ approach, assuming most of your customers will be agents, not humans.”

Countless crypto companies, including Huang’s new payments startup Tempo, are racing to adapt or reshape their products for this emerging user base. Tron founder Justin Sun has already directly called it Web4.0 (as if Web3.0 was truly built).

MoonPay, which originally helped users (now increasingly software) buy and sell crypto via traditional payment methods, has completely overhauled its AI strategy after the rise of open-source AI assistant OpenClaw. MoonPay product lead Kevin Arifin said: “MoonPay’s bet is that we no longer need to invest heavily in beautiful user interfaces because agents will become the new interaction point.”

For ordinary users who simply don’t want to care about the underlying crypto details, this is undoubtedly good news: just tell AI what you want to do—buy some Bitcoin, find a lending service with a good interest rate, let assets generate yields—and it will handle everything.

However, this is still far from large-scale application.

Most of the current AI-driven crypto payments are conducted via Coinbase’s open standard x402, which allows network service providers to charge agents directly.

Recently, even simple tasks like getting weather forecasts or renting computing power require developers to register services one by one, link credit cards, and generate API keys. Slightly more complex projects fall into chaos managing accounts, subscriptions, and keys.

x402 offers a simpler pay-as-you-go model: when an agent requests a service, the server returns a price, and the agent can automatically pay with crypto from the developer’s allocated wallet. This not only enables usage-based billing but also begins to replace the flood of API keys.

Reppel, head of Coinbase’s developer platform and founder of x402, said: “People who have used OpenClaw might remember that you needed to configure 10 API keys before you could start. With x402, the wallet becomes a universal API key that can connect to any service supporting x402.”

So far, the users of these agents are mainly developers. According to Artemis, since x402 launched in May 2025, AI assistants have completed about 107 million transactions through this standard, with a total real transaction volume of around $30 million. Most individual transactions are small, between $0.20 and $0.40.

Artemis analyst Lucas Shin said: “It’s clear we’re still in the early stages.” He believes that transaction volume at this point is almost irrelevant; more important are which ecosystems are truly building, and how many merchants are willing to offer services via x402. Currently, about 3,900 merchants are involved, including Amazon Web Services, blockchain development platform Alchemy, and data provider Messari.

Crypto industry excitement about agent-based commerce is understandable. Rishin Sharma, AI product and growth lead at Solana Foundation, said: “Almost every engineering team we see, including ours, is using AI tools.” He said everyone on his team uses AI, with over 70% of code generated by AI. Service providers that once built businesses around traditional APIs are now thinking about a different question: not how to attract the next hundred developers, but how to prepare for the next hundred agents.

Recently, Paradigm and Stripe launched Tempo, a blockchain-focused payments project that last year raised $500 million at a $5 billion valuation, and introduced its own agent trading standard, while partnering with Visa to support fiat payments.

However, most in crypto believe stablecoins are a more natural payment track for AI agents. Card-based payments are not economical for small transactions: payment providers often charge not only a percentage fee but also a fixed fee of about $0.30 per transaction, meaning tiny transactions can be completely eaten up by fees.

This is why institutions like Circle, the second-largest stablecoin issuer, are developing systems tailored for machine payments. Earlier this month, Circle launched Nano Payments, allowing agents to send ultra-low, fee-free USDC on its new chain Arc and multiple test chains, with amounts as low as less than a cent. But the threat to giant networks like Visa and Mastercard isn’t just micro-payments: AI agents using stablecoins could impose huge fee pressures on any scale of transaction.

If software agents are about to become the next major user group, the question is no longer just how they pay, but what kind of network is built for them. Jesse Pollak, founder of Base, said: “We’re thinking from a full-stack perspective: from the scalable and decentralized underlying infrastructure, to upper-layer tools and account models, to the interfaces of actual agent interactions. We need to ask: how can all this be natively adapted for agents?”

He mentioned that some agents already operate like micro-enterprises. For example, entrepreneur Nat Eliason’s Felix agent earned $163,686 in the past 30 days by operating an AI app store and selling his own guide, “How to Hire AI.” It also issued a crypto token, but its market cap is only $1.5 million.

Not everyone is optimistic about the prospects of combining AI agents with cryptocurrencies. Haseeb Qureshi, managing partner at crypto VC firm Dragonfly, said: “Many people overhype the current state of development. The reality is that everything here is mostly just toys right now.”

He added that while agents might indeed bring small, continuous payments for data, computing, and other services, reaching macro-level impact would require an enormous number of agents. After all, humans still control the funds and are the main demand source.

Qureshi worries that the industry is repeating past mistakes, mistaking new trends for revolutions: “Many in crypto are terrible investors because they immediately believe the stories they make up. That’s how crypto always is.”

He pointed to past booms like IoT and the metaverse, where believers once thought everything would happen overnight, and cryptocurrencies would be at the core of it all. “Crypto will be important, it will be part of the story, but not everything, and not overnight.”

Outside the crypto industry, the idea that “agent commerce will help crypto shake off traditional financial giants” is not widely accepted.

Trace Cohen, general partner at Six Point Ventures, which focuses on vertical AI and software investments, said: “The idea that old systems like Visa and Mastercard will become irrelevant in the AI agent era is ridiculous. That’s impossible. No matter how outdated the technology, it still works.”

He believes card networks will continue to control payment rails, and history shows they’re more likely to acquire or absorb promising new businesses rather than be replaced. But he also admits that stablecoins might have an advantage in overseas markets, where many regions have smaller banks, lower trust, and poorer interoperability.

The bigger obstacle is rebuilding the trust layer that traditional payment companies spent decades establishing. Olivia Chow, director of Zero Knowledge Consulting and a payments industry advisor, said: “Visa and Mastercard excel at setting rules—handling exceptions, responsibilities, and participant access.” She added, “Stablecoins still need to establish mechanisms for fraud prevention, risk management, and clear procedures for when ordinary users encounter issues. These users don’t just say ‘I care about my safety and will take risks.’ Until then, mainstream adoption is unlikely.”

She also believes that since card networks are already supporting agent transactions, AI commerce might not threaten their business—in fact, it could expand their reach. “If they do it right, not only will they not cannibalize existing business, but they will strengthen their position by entering the flow discovery layer, too.”

But payments are only part of the story. As more traditional assets are tokenized, early examples include BlackRock’s $2 billion bond fund BUIDL and Franklin Templeton’s $1 billion government money fund FOBXX. Infrastructure for next-generation asset management is quietly taking shape. After all, stock indices are essentially rule-based asset baskets. Once stocks, bonds, and funds are tokenized, AI agents can not only make payments but also hold assets, rebalance portfolios, and move funds across markets—all without traditional broker accounts.

This vision coincides with one of the largest wealth transfers in human history. Over the next 20 years, approximately $84 trillion will shift from the Baby Boomer generation to their descendants. Many grew up with Robinhood, already own crypto wallets, and are willing to bet on everything from election results to Taylor Swift’s wedding venue.

Meanwhile, the financial advisory industry itself is aging. There are about 330,000 financial advisors in the U.S., with an average age of 56. According to Cerulli Associates, nearly 40% will retire in the next decade, leaving a huge gap in asset management for ordinary investors.

Crypto companies are already preparing for this. On Tuesday, reports emerged that MoonPay, which is in talks with the NYSE parent company for funding and a $5 billion valuation, launched an open wallet standard to help AI agents manage funds and execute trades across multiple blockchains.

Joseph Chalom, former head of digital asset strategy at BlackRock and CEO of Ethereums Treasury company Sharplink, said: “I don’t think this crypto boom will be like previous ones.” He believes that innovations like stablecoins, tokenized assets, and widespread wallet infrastructure, combined with AI understanding user preferences and goals, and the ongoing intergenerational wealth transfer, will create a powerful synergy. “Once investors realize what they’ve missed, it will be hard to go back.”