Tiger Research: AI Agent Wallet Infrastructure, the Underlying Engine for a 7x Revenue Leap
- Core Thesis: To support the large-scale, high-frequency, micropayment needs of future AI agents, the crypto wallet industry is racing to position itself. Its true goal is to seize the infrastructure of the agent economy era and the potential financial services associated with it, even though there is no short-term revenue.
- Key Elements:
- Existing payment systems (e.g., credit cards) cannot handle the thousands of sub-cent payment requests per second from AI agents due to high fixed processing fees and chargeback mechanisms.
- Companies like Coinbase are proactively building agent wallets that support programmability and automatic disbursement. The core logic is to lock in a future user base, not to compete for current transaction fee revenue.
- Estimates based on Coinbase data show that in an aggressive scenario of widespread AI agent adoption, its annual revenue could reach up to approximately 7 times its current total revenue, with revenue increments scaling geometrically.
- The payment history accumulated by wallets can serve as a credit standard for evaluating AI agent income, laying the foundation for new financial services like future-revenue-based loans (e.g., RBF).
- Currently, the market still faces major obstacles, including erroneous payments by AI agents, fragmentation of payment protocols, and unclear legal status of agents, meaning it is still in a validation phase.
This article is written by Tiger Research. News headlines often report on AI agents trading autonomously and handling payments by themselves. In reality, the crypto wallet industry has been quietly laying the groundwork for some time. Currently, more than ten companies are specializing in building wallets for AI agents. What do they truly want? And how significant is the potential payoff?
Key Takeaways
- When AI agents browse the web, purchase goods, or access information on behalf of humans, they initiate hundreds or thousands of micro-payments worth just a few cents or even less. Existing card-based payment systems simply cannot handle this volume, necessitating wallets capable of automatically splitting and sending funds based on preset conditions.
- Despite seeing almost no revenue in the short term, companies like Coinbase and Binance are doubling down on AI wallet infrastructure. The reason is simple: to capture the future user base before large-scale agent transactions begin. The current phase is about gaining an edge before actual demand explodes.
- Based on Coinbase's data, its revenue could potentially increase up to roughly 7 times its current level following a rise in AI agent usage.
- The payment history accumulated in a wallet can intuitively show whether a specific AI agent is generating income. This opens the door for lending based on future earnings – similar to offering credit to a small business based on its card transaction flow.
- However, all of this remains in the realm of possibility rather than verified reality. AI agents can still make mistakes and execute erroneous payments; regulations vary by country and company; the legal status of AI agents is yet to be clarified. Therefore, the current focus of competition is not on making money today, but on securing a position years in advance for a market expected to take shape.
AI Agents Are Becoming Fully Active
Earlier this year, a widely watched experiment took place on the prediction market Polymarket: an AI agent was given $50 in startup capital and allowed to trade autonomously, with the condition that it would "disappear" if it couldn't earn enough to cover its API and server costs. The result? The agent succeeded in its trading. Subsequently, a wave of similar agents began trading in the same manner.
While AI agents haven't yet entered daily life, it's clear they will be used on a massive scale in the near future.
Every Agent Transaction Starts with a Wallet
Currently, AI agents haven't penetrated everyday payment scenarios. Their most active application remains trading bots within the crypto ecosystem – operating independently of traditional payment rails and focused on cryptocurrency trading.
In the future, payments will extend into areas that are hard to imagine today. As we pointed out in a previous report, AI is changing the very nature of payments. Once agents, rather than humans, act directly on the network, the value of individual payments will drop dramatically. The cost of a single API call or data query could be as low as $0.001, or even $0.00001 in extreme cases.
To evolve beyond current wallet usage and enable such micro-payments that are automatically split based on preset conditions without human intervention, a programmable payment system is essential. This is the context in which the x402 payment rail emerged, with the wallet serving as its foundational operating infrastructure.
However, existing payment rails are designed around "humans" as the transaction entity.
Bank cards are issued to specific cardholders and rely on chargeback mechanisms – where humans dispute and reverse transactions when problems occur – with a fixed fee of tens of cents per transaction. These aren't issues when a person occasionally spends $20. But this payment model becomes economically unviable when an agent starts sending thousands of payments per second, each API call costing $0.001 and each data record costing $0.00001.
The core question is: is the money itself programmable?
Bank cards allow for the automation of payment information entry, but they cannot be programmed to split funds conditionally, execute streaming payments, or settle fund flows instantly. The rails that wallets operate on, however, inherently possess these capabilities. Storing payment information on a card can, at best, execute a "human-scale" transaction on behalf of a person. Once the economy shifts towards machine-to-machine transactions, the wallet becomes the only possible starting point.
AI Agents Could Be a $50 Billion Business
As the table shows, wallet providers range from exchanges to stablecoin issuers. Why are so many different types of players entering the agent wallet infrastructure space, which is unlikely to be profitable in the short term?
The answer: They are positioning for future revenue and business, not today's profits. Adding agent functionality to wallets now isn't about immediate earnings; it's about building the capacity in advance to absorb massive transaction volumes when agents become highly active.
The key point is that AI agents will eventually operate 24/7 in a browserless environment, requiring no human intervention. Imagine a user asking an agent to complete a research report. While gathering information, the agent executes a micro-payment each time it pulls fee-based data from different platforms. A single user instruction could trigger 20, 30, or even more payments in an instant.
A request that seems like a simple operation to a human translates into a high volume of payment transactions once processed by an AI agent.
How would this change in the payment environment affect a company's revenue? We can estimate using data disclosed by Coinbase. The calculation is based on Coinbase's 9.2 million monthly transacting users (MTU), not its total of approximately 120 million registered users.
Combining three variables – adoption rate, number of agents per user, and daily call frequency – yields the following scenarios:
- Conservative Scenario (10% adoption, 1 agent per user, 50 daily calls): Estimated annual incremental revenue of ~$84 million, a 1.2% increase.
- Moderate Scenario (50% adoption, 2 agents per user, 200 daily calls): Additional revenue jumps significantly to ~$3.36 billion, a 46.8% increase.
- Aggressive Scenario (100% adoption, 3 agents per user, 1000 daily calls): Estimated annual revenue of ~$50.37 billion, roughly 7 times Coinbase's current total revenue.
The most striking aspect of this comparison is that the gap between the scenarios expands geometrically, not linearly. Adoption only increases 10 times (from 10% to 100%), yet the revenue gap expands roughly 600 times – from $84 million to $50.37 billion.
Because the three variables – adoption rate, agents per user, and daily calls – are multiplicative, any small increase leads to exponential growth in total volume. Consequently, if agents achieve mass adoption and user numbers surge, the resulting revenue stream could reach up to roughly 7 times current total revenue.
This is precisely why Coinbase is heavily promoting agent wallet infrastructure even with almost no related revenue today – it aims to secure its share of the revenue expected to materialize in the agent era.
Moving Towards the New Bank for Agents
The transaction data accumulated through wallet infrastructure goes far beyond simple record-keeping. It provides the foundation for new business models: the payment history stored in a wallet can serve as a credit standard for evaluating the financial health and performance of an AI agent.
Once this data-driven credit assessment system is established, wallet providers can naturally expand into next-generation financial services, such as Revenue-Based Financing (RBF) specifically designed for AI agents.
Stripe Capital is a classic example. 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 didn't rely on external credit bureaus or require extensive loan documentation. Instead, it directly used the real-time sales data of each merchant within its own payment network to assess loan eligibility and amounts.
The Stripe case demonstrates that a company can build a high-value financial business on top of its existing operational data pipeline without needing an additional sales network or marketing investment.
Agent wallet providers are likely to follow a similar expansion path. By continuously accumulating revenue data for agents through their wallets, they can build a foundation for providing operational capital via RBF and capturing value as a financial platform focused on agents.
However, there's a prerequisite for truly building this new business line: AI agents must evolve from simple payment executors into asset-holding entities that can generate their own income, earning enough real revenue to repay loans.
This Growth Remains Unproven
The descriptions above – of Coinbase's revenue potentially increasing 7 times and expanding into RBF – are optimistic scenarios assuming widespread adoption of agent payments. Significant obstacles remain in translating these ideas into tangible reality within the current economy.
First, there are major questions about the actual purchase conversion rates and payment reliability of AI agents. Agents can still "hallucinate" and make errors when placing orders autonomously, leading to incorrect payments. They are sometimes blocked outright by card issuers' fraud detection systems (FDS). Actual payment completion rates therefore remain relatively low.
Furthermore, payment protocols like x402, AP2, and MPP are still fragmented and haven't converged on a single standard. The lack of a clear legal personality for AI agents, combined with the absence of clear KYC (identity verification) and financial regulatory frameworks, further hinders market expansion.
Therefore, the immediate goal for wallet providers is not short-term fee income. It took Apple's App Store 15 years to build an annual fee market of $10 billion, and WeChat Pay took 7 years to establish its vast mini-program ecosystem. Agent wallets are on a similarly long timeline – they are building an ecosystem, not chasing immediate returns.
The current competition isn't about today's marginal revenue. It's about which company can be the first to control the data flow of funds in the fully formed agent economy of the next five to ten years.
