Coinbase Projects 700% Revenue Surge as Exchanges Race for AI Agent Wallet Infrastructure

Woofun AI reports that the crypto wallet industry is quietly laying the groundwork for an autonomous payment era, with over a dozen companies developing specialized infrastructure for AI agents. This strategic pivot, analyzed by Ekko an and Ryan Yoon of Tiger Research and compiled by AididiaoJP for Foresight News, reveals that exchanges and stablecoin giants are battling for control of the future transaction layer. While headlines focus on independent trading, the underlying competition is for the wallet infrastructure that enables these agents to operate. Key players including Coinbase and Binance are investing heavily, not for immediate returns, but to secure a foothold before actual demand surges. The core question is no longer if AI will pay, but who will hold the keys to its financial identity.

The operational necessity for these new wallets stems from the fundamental shift in payment volume and granularity. When AI agents browse the web, purchase goods, or access information on behalf of humans, they initiate hundreds or even thousands of small payments amounting to just a few cents or less. The existing credit card payment system cannot handle such volumes, making it necessary to have wallets that can automatically split and send funds based on preset conditions. Traditional systems are designed for human-scale transactions, where a user occasionally spends $20 on something. In contrast, an agent executing a research report might trigger 20 to 30—or even more—payments in an instant as it fetches paid data from different platforms. This high-frequency, micro-transaction model requires a programmable payment system that can operate without human intervention, a capability that current card networks lack.

Strategic motivation for this infrastructure build-out is rooted in securing future user footholds rather than short-term profit. Although there is little chance of generating revenue in the short term, companies like Coinbase and Binance are still investing heavily in AI wallet infrastructure. The reason is simple: to secure a foothold among future users before agents start conducting large-scale transactions. This current phase is about seizing the opportunity before actual demand surges. According to Coinbase’s estimates, as the use of AI agents increases, its revenue could rise to around seven times its current level. The payment records stored in wallets can clearly show whether an AI agent is generating income, opening the door to offering loans based on future earnings—similar to granting credit to a small business based on its transaction history.

However, all of this remains at the conceptual stage rather than being proven facts yet.

Current challenges and legal ambiguities pose significant hurdles to widespread adoption. AI agents may make mistakes and initiate incorrect payments; rules vary across different countries and companies; and the legal status of agents is still unclear. Therefore, the focus of competition now is not to earn money today but to secure a position in a market expected to take shape in a few years. Agents may still experience "hallucinations" when placing orders autonomously, leading to incorrect payments. Sometimes, they are also blocked directly by card issuer fraud detection systems (FDS), resulting in a low actual payment completion rate.

Additionally, payment protocols like x402, AP2, and MPP remain fragmented and have not converged toward unified standards. AI agents are not legal entities, lacking clear KYC (Know Your Customer) procedures and financial regulations, which further hinders market expansion.

A case study from earlier this year highlights the growing activity of these agents. A highly publicized experiment took place on the prediction market Polymarket: An AI agent was given $50 as startup capital to trade independently, with the condition that it would "disappear" if it couldn’t earn enough to cover API and server costs. Surprisingly, the agent succeeded in trading. Since then, a number of similar agents have started trading in the same way. Although AI agents aren’t yet part of daily life, it is clear they will be used on a large scale in the near future. Every transaction made by an agent starts with a wallet. Their most active application remains trading bots within the crypto ecosystem—operating independently of traditional payment systems and focusing solely on cryptocurrency trades. In the future, payments will extend into areas that are hard to imagine today.

Technical limitations of current payment systems versus emerging frameworks like x402 are stark. As we noted in previous reports, AI is transforming the nature of payments. Once agents operate directly on the web instead of humans, the amount per transaction will drop sharply. An API call or data query might cost as little as $0.001, or even $0.00001 in extreme cases. To go beyond the current way of using wallets and achieve such small, automatically split payments without human intervention, a programmable payment system is needed. This is precisely the background behind the emergence of the x402 payment framework, with wallets serving as the foundation for its operation.

However, existing payment systems are designed around "humans" as the main participants. Credit cards are issued to specific holders, and a chargeback mechanism is in place—where humans dispute and cancel transactions when problems arise, with fixed fees of several cents per transaction. These aren’t issues when a person occasionally spends $20 on something. But once agents start issuing thousands of payments per second, with each API call costing $0.001 and each data record costing $0.00001, this payment model becomes economically unviable.

The core question is: Can money itself be programmed? Credit cards can automate the entry of payment information but cannot be programmed to split funds, make stream-based payments, or settle transactions instantly. In contrast, the framework operated by wallets inherently possesses these capabilities. Storing payment information on a card can at most execute one "human-scale" transaction on behalf of a person. Once the economy shifts to direct transactions between machines, wallets become the only possible starting point.

Revenue projections illustrate the potential scale of this shift. The calculation is based on Coinbase’s 9.2 million monthly active trading users (MTUs), rather than its total of around 120 million registered users. By considering three variables—adoption rate, number of agents per user, and daily call frequency—we arrive at the following scenarios: Conservative Scenario (10% adoption rate, 1 agent per user, 50 calls per day): Annual revenue increase of approximately $84 million, representing a 1.2% rise. Neutral Scenario (50% adoption rate, 2 agents per user, 200 calls per day): Additional revenue rises significantly to around $3.36 billion, representing a 46.8% increase. Aggressive Scenario (100% adoption rate, 3 agents per user, 1,000 calls per day): Annual revenue reaches around $50.37 billion, roughly seven times Coinbase’s current total revenue.

Woofun AI data shows that these projections rely on multiplicative variables, meaning small improvements in adoption or usage frequency lead to exponential growth in total revenue.

Exponential growth dynamics and strategic positioning explain the intense competition. The most striking aspect of this comparison is that the gap between the three scenarios grows exponentially rather than simply adding up. Even if the adoption rate increases from 10% to 100% (a 10-fold increase), the revenue gap expands by about 600 times—from $84 million to $50.37 billion. This is because the three variables—adoption rate, number of agents per user, and daily call volume—are multiplicative.

Any slight improvement leads to exponential growth in total revenue. Therefore, once agents are widely adopted and user numbers surge, the resulting revenue stream could reach up to seven times current total revenue. This is also why Coinbase, despite having almost no related revenue today, is still vigorously promoting AI agent wallet infrastructure—it aims to secure a share of the revenue expected in the era of agents. Relevant data shows that wallet providers range from exchanges to stablecoin issuers.

Why do so many different types of players invest in AI agent wallet infrastructure, where profits are unlikely in the short term? The answer is that they are laying the foundation for future revenue and business growth, not today’s profits. Adding agent functions to wallets now isn’t aimed at immediate earnings but at building the capacity to handle massive transaction volumes once agents become widely used. The key point is that AI agents will eventually operate 24/7 in a browser-free environment, completely without human intervention.

Future financial services may evolve through models like Revenue-Based Financing (RBF). The transaction data accumulated through wallet infrastructure goes far beyond simple records. It provides the foundation for new business models: The payment history stored in wallets can serve as a credit benchmark to assess the financial health and performance of AI agents.

Once such a data-driven credit evaluation system is established, wallet providers can naturally expand into the next generation of financial services, such as Revenue-Based Financing (RBF) tailored for agents. Stripe Capital is a prime example. It successfully built new financial services on top of its existing payment data. When Stripe launched its lending service, Stripe Capital, in September 2019, it didn’t rely on external credit bureaus or complicated loan applications.

Instead, it used real-time sales data from each merchant in its payment network to evaluate loan eligibility and amounts. Stripe’s example shows that a company can build high-value financial services on top of its existing operational data pipelines, without the need for additional sales networks or marketing efforts. Agent wallet providers are likely to follow a similar path of expansion.

By continuously accumulating income data from agents through wallets, they can provide operating funds via RBF and generate revenue as a finance platform focused on agents.

However, for this new business line to truly take shape, one prerequisite must be met: AI agents must evolve from mere payment executors to asset holders capable of generating their own income, earning sufficient real revenue to repay loans.

Long-term hurdles, fragmentation, and ecosystem building remain the primary obstacles. The descriptions of Coinbase’s potential revenue increase of up to seven times and its expansion into RBF are based on optimistic scenarios assuming widespread adoption of agent payments. There are still significant obstacles to turning these ideas into reality in the economic world. First, there are major doubts about the actual purchase conversion rate and payment reliability of AI agents. Agents may still experience "hallucinations" when placing orders autonomously, leading to incorrect payments. Sometimes, they are also blocked directly by card issuer fraud detection systems (FDS), resulting in a low actual payment completion rate.

Additionally, payment protocols like x402, AP2, and MPP remain fragmented and have not converged toward unified standards. AI agents are not legal entities, lacking clear KYC (Know Your Customer) procedures and financial regulations, which further hinders market expansion. Therefore, the current goal of wallet providers is not short-term fee income. It took Apple’s App Store 15 years to build a $10 billion annual fee market, and WeChat Pay 7 years to develop its extensive mini-program ecosystem. Agent wallets are also on a long-term journey—they are building ecosystems rather than seeking immediate returns. The current competition is not about marginal revenue today but about which company can take control of the transaction data once the agent economy fully matures over the next five to ten years.