This article is written by Tiger Research. Headlines are always reporting on AI agents autonomously trading and processing payments by themselves. In fact, the cryptocurrency wallet industry has long been quietly paving the way. Currently, more than ten companies are specifically creating wallets for agents. What do they really want? What are the potential returns?

Key Points

• When AI agents browse the internet, purchase goods, or access information instead of humans, they initiate hundreds or thousands of micro-payments worth just a few cents or even less. The existing credit card payment system cannot support this volume, thus requiring wallets that can automatically split and send funds based on predefined conditions.
• Although there is almost no revenue in the short term, companies like Coinbase and Binance are still ramping up AI wallet infrastructure. The reason is simple: to secure the future user base before large-scale trading by agents occurs. This current phase is about seizing opportunities before real demand explodes.
• Based on data from Coinbase, as the usage of AI agents rises, their revenue could potentially reach up to seven times its current level.
• The accumulated payment records in wallets can clearly indicate whether an AI agent is generating revenue, which opens the door for lending based on future earnings—similar to granting credit based on a small business's card transaction flow.
• However, all this remains at the potential stage rather than a validated fact. AI agents can make mistakes and execute incorrect payments; regulations vary by country and company; and the legal status of agents is still unclear. Thus, the current competitive focus is not on earning today’s money but on positioning in a market that is expected to emerge in a few years.

AI Agents Are Becoming Fully Active

Earlier this year, there was a widely followed experiment on the prediction market Polymarket: giving an AI agent $50 in startup capital to trade independently, on the condition that if it did not make enough to cover the API and server costs, it would "disappear." As a result, this agent successfully traded. Following this, a batch of similar agents started trading in the same manner.

AI agents have not yet entered daily life, but it is clear that they will be used on a large scale in the near future.

Every Transaction by an Agent Begins with a Wallet

Currently, AI agents have not entered everyday payment scenarios. Their most active application remains in trading bots within the cryptocurrency ecosystem—operating independently from traditional payment tracks, focusing on cryptocurrency trading.

In the future, payments will extend into areas that are difficult to imagine today. As we pointed out in previous reports, AI is changing the nature of payments. Once actions are taken by agents rather than humans directly online, the amount of single payments will drastically decrease. An API call or a data query could cost as low as $0.001, and in extreme cases, even just $0.00001.

To surpass the current usage of wallets and realize such small, automatically split payments based on predefined conditions, without human intervention, one must rely on programmable payment systems. This is the backdrop for the emergence of the x402 payment track, with wallets being the foundation for this track's operation.

However, the existing payment tracks are designed around "humans" as the transactional subjects.

Credit cards are issued to specific cardholders, using chargeback mechanisms—issues are disputed and transactions reversed by humans, with each transaction incurring fixed fees of several cents. When a human occasionally spends $20, these are not problems. But once an agent begins processing thousands of payments per second, with each API call costing $0.001 and each data record $0.00001, this payment model becomes economically unfeasible.

The core question is: Is money itself programmable?

Credit cards can automate the input of payment information, but they cannot be programmed to split, stream, or settle payments based on conditions. The track on which wallets operate inherently possesses these capabilities. Storing payment information on a card can at most execute one "human-scale" transaction. Once the economy shifts to direct transactions between machines, wallets become the only potential starting point.

Agents Could Represent a $50 Billion Business

From the related charts, it can be seen that wallet providers encompass a wide range, from exchanges to stablecoin issuers. Why are so many different types of players entering the AI agent wallet infrastructure, which has hardly any visible profits in the short term?

The answer is: they are laying out future revenue and businesses, not today’s. Adding AI functionalities to wallets is not about immediate profits but about building capacity in advance to handle massive transaction volumes when agents operate on a large scale.

The key is that AI agents will ultimately operate around the clock in a non-browser environment, without human intervention. Imagine a user asking an agent to complete a research report. As the agent collects information, every time it pulls paid data from different platforms, it executes a small payment. A user's single command could trigger 20 to 30 payments or even more in an instant.

What seems like a simple operational request to humans transforms into numerous payment transactions once processed by AI agents.

How will this change in the payment environment affect corporate revenues? We can estimate this using data disclosed by Coinbase. The calculation is based on Coinbase’s 9.2 million monthly active trading users (MTU), rather than its total registered users of about 120 million.

By combining three variables: adoption rate, the number of agents per user, and daily invocation frequency, we arrive at the following scenarios:

• Conservative scenario (adoption rate 10%, 1 agent per user, 50 calls per day): annual incremental revenue of about $84 million, an increase of 1.2%.
• Neutral scenario (adoption rate 50%, 2 agents per user, 200 calls per day): additional revenue soars to about $3.36 billion, an increase of 46.8%.
• Aggressive scenario (adoption rate 100%, 3 agents per user, 1000 calls per day): annual revenue of about $50.37 billion, approximately 7 times Coinbase’s current total revenue.

The most striking point in this comparison is the geometric amplification between the three scenarios, rather than a simple additive increase. The revenue gap expands by about 600 times—from $84 million to $50.37 billion—when the adoption rate increases from 10% to 100% (a tenfold increase).

Because the variables of adoption rate, number of agents per user, and daily call volume are multiplicative, any slight increase in one will lead to exponentially higher total volume. Thus, once agents achieve large-scale adoption, resulting in a surge in user numbers, the resulting revenue flow could reach about seven times the current total revenue.

This is also why Coinbase, despite having almost no relevant revenues today, is still heavily promoting AI wallet infrastructure—because it wants to prematurely secure its share of the expected revenues in the agent era.

Moving Towards a New Banking for Agents

The transaction data accumulated through wallet infrastructure is far more than a simple record. It provides the foundation for new business models: the payment history stored in wallets can serve as a credit standard for assessing the financial status and performance of AI agents.

Once this data-driven credit evaluation system is established, wallet providers can naturally extend into the next generation of financial services, such as revenue-based financing (RBF) specifically targeting agents.

Stripe Capital is a typical case. It successfully built a new financial business on top of existing payment data. When Stripe launched its lending service, Stripe Capital, in September 2019, it did not rely on external credit agencies or cumbersome loan documentation, but instead directly utilized the real-time sales data of every merchant in its payment network to assess loan qualifications and amounts.

The Stripe case shows that a company can construct a high-value financial business on top of existing operational data pipelines without the need for additional sales networks or marketing investment.

Wallet providers for agents are likely to follow the same path of expansion. By continuously accumulating income data from agents through wallets, they will have the foundation to provide operational funds via RBF, and as a focused financial platform for agents, generate profits.

However, for this new business line to be truly established, one prerequisite must be met: AI agents must evolve from mere payment executors to being assets capable of generating their own income, earning enough real revenue to repay loans.

This Growth Remains Unverified

The description above regarding Coinbase's potential revenue growth of seven times, and the expansion into RBF, is based on optimistic scenarios assuming widespread adoption of agent payments. Significant barriers still remain to bring these into the real economy.

First, there are major questions regarding the actual purchasing conversion rates and payment reliability of AI agents. Agents can still "hallucinate" errors when placing orders autonomously, leading to incorrect payments; sometimes they are directly intercepted by the issuer's fraud detection systems (FDS). Consequently, the actual completion rate of payments remains low.

Moreover, payment protocols like x402, AP2, and MPP are still in a fragmented state and have yet to converge into a unified standard; AI agents lack legal entity status, and the absence of clear KYC (Know Your Customer) and financial regulations further obstructs market expansion.

Thus, the current goal of wallet providers is not short-term fee income. It took Apple’s App Store 15 years to build a fee market worth $10 billion annually, and WeChat Pay took 7 years to establish a vast mini-program ecosystem. AI wallets are also on a long timeline—they are building an ecosystem, not competing for immediate returns.

Current competition is not about today’s marginal income but about which company can first control the capital flow data once the AI economy matures fully in the next five to ten years.

免责声明：本文章仅代表作者个人观点，不代表本平台的立场和观点。本文章仅供信息分享，不构成对任何人的任何投资建议。用户与作者之间的任何争议，与本平台无关。如网页中刊载的文章或图片涉及侵权，请提供相关的权利证明和身份证明发送邮件到support@aicoin.com，本平台相关工作人员将会进行核查。