AI payments are helping you save money, but making it harder for banks to profit.

Author and source: Xin Finance

Following industry convention, 2026 will be known as the "Year of AI Payments."

Just last June, within a single week, institutions such as Ant Group, Tencent, JD.com, and China UnionPay collectively launched AI-powered payment products, signaling the imminent start of a new payment battle.

This isn't just jumping on the trend.

For the past few decades, all payment innovations have revolved around "how people pay."

From cash to bank cards, from POS terminals to QR codes, from passwords to facial recognition—no matter how the tools change, it’s always people who make the decisions.

But what makes this round of AI payment warfare unique is that it aims to have AI pay on your behalf—the decision-making authority shifts from humans to AI agents, potentially reshaping the entire transaction ecosystem and industry landscape.

That's why no one dares to miss the "train of the times" that is AI, regardless of whether they're truly ready or not.

Yet, amid the lively competition among tech giants, domestic commercial banks have remained largely inconspicuous. Aside from a few banks issuing AI-enabled equity cards, commercial banks have not yet launched any notable products.

Especially as tech giants compete to become users' super assistants that help them save and spend, traditional banking still seems stuck in the old mindset of issuing cards.

The logic of "get an AI membership and token trading rights upon card activation" is far removed from true AI-native payments.

Financial institutions naturally have higher standards and requirements for security, and they neither need nor should blindly follow suit. But as the entire market undergoes drastic changes, the alarm bells should also ring for banks.

Shortly after, McKinsey noted in the report “The End of Inertia: How AI Agents Are Disrupting Retail and Small Business Banking”:

Technological shifts such as AI agents could undermine the foundations of the financial industry, threatening the business models and revenue streams of banks, small and medium-sized enterprises, and credit card companies, with potential revenue impacts reaching billions of dollars.

This concern is not unfounded. After all, during the previous mobile payment battle, banks lost some of their dominance over front-end access points and suffered significant losses as a result.

The future is here, it's just unevenly distributed.

Not many people have actually experienced the AI payment feature.

Based on trials by several major domestic tech companies, due to limitations in integration scenarios and authorization scope, it is still difficult to say that it has brought a transformative experience.

For example, AI payments can help us buy a cup of coffee at a discounted price. However, for most users, even “free items” are not unusual in the subsidy wars on internet platforms.

In contrast, what’s more noteworthy is the way we receive these benefits, which is entirely different from past subsidy models.

In the past, whether it was online platforms issuing coupons to attract users or banks offering fixed credit card benefits, users had to find, claim, and combine discounts on their own.

However, when we give instructions to AI, we don’t need to worry about the details or processes involved. Whether it negotiates directly with merchants or uses store, platform, or payment-based discounts, it will ultimately choose the best option and place the order for me.

From this perspective, while AI payments currently only help us order a cup of coffee, in the long term, they may fully take over our wallets—managing both spending and assets.

In fact, institutions in overseas markets have taken an even bigger step.

Recently, U.S. fintech company Robinhood launched two products: Agentic Trading, which uses AI to assist with investing, and Agentic Credit Card, which uses AI to assist with spending.

In simple terms, Robinhood, through the MCP protocol, enables users to integrate their investments, credit card services, and AI agents, allowing the agents to perform actions directly.

In terms of investing, the platform can develop strategies based on the user’s investment and trading goals within their Agentic Trading account.

For example, analyze the concentration risk and industry exposure of its portfolio, construct an initial portfolio based on its criteria, and backtest mean-reversion strategies.

In terms of credit card services, the AI Agent can also automatically perform tasks according to user requests—searching for the best prices, monitoring inventory status, and automatically completing payments, all while still allowing users to enjoy the 3% cashback on transactions.

Robinhood lists several specific scenarios on its official website:

Sneaker enthusiasts can instruct the AI agent to immediately purchase the desired size whenever a new pair of sneakers is priced below $300.

Food enthusiasts can instruct the AI Agent to immediately book a table at the city’s top restaurants as soon as availability opens up on their preferred date and time;

Small business owners can authorize an agent to handle multiple purchase requests—such as buying ingredients for custom cakes with a total cost not exceeding $50—or even purchase a website domain on their behalf.

Pet owners can ask an AI agent to purchase highly-rated products on Amazon—for example, “Buy me a five-star, durable dog toy under $30”—saving them from endless searching and comparison.

Of course, so far, both investment and credit card accounts operate independently of the main account, and users can disconnect them from AI at any time; however, this product ecosystem still demonstrates, to some extent, the direction in which the market is evolving.

On the surface, these products don't appear to have directly impacted banks, but if banks continue to stand by, the cost could be much higher than imagined.

The Disrupted "Lazy Reward"

According to McKinsey, banks' traditional profit models have heavily relied on users' "inertia premium."

As the foundation of retail operations, net interest income from global retail banking accounts for approximately 30% of total profits, with low-cost checking and savings deposits being the core source of this profit.

Most consumers don’t constantly monitor fluctuations in deposit interest rates, nor do they have the time, tools, or motivation to optimize the returns on every dollar in their accounts. They prefer to handle deposits, withdrawals, bill payments, and wealth management all in one bank, valuing convenience and peace of mind.

Taking the U.S. market as an example, in June 2025, traditional bank checking account interest rates were only 0.07%, standard savings rates were 0.38%, while the highest deposit rates offered by leading online digital banks have surpassed 4%.

Over the past five years, the compound annual growth rate of non-interest-bearing demand deposits in the U.S. banking sector reached 28%, while interest-bearing deposits grew at an average annual rate of only 3%.

User inertia has allowed banks to accumulate vast amounts of low-cost funds, but AI agents that operate 24/7 will completely transform this dynamic.

It can monitor all account balances in real time, compare deposit interest rates and investment returns across institutions, automatically transfer idle funds into high-yield products, and only return them to the checking account before the bill due date.

McKinsey estimates that annual revenue related to deposits in the European market alone exceeds $100 billion. Even if only 10% to 20% of consumers use AI agents to automatically manage their funds, banks' net interest margins could shrink by 30 to 50 basis points.

This is not an exaggeration.

McKinsey noted that the emergence of Yu’E Bao at the time clearly demonstrated the industry impact of widespread automated fund optimization. AI agents, however, far surpass Yu’E Bao in capability—they can continuously allocate funds across multiple banks, dynamically distributing assets based on yield, settlement speed, and user risk thresholds.

Now let’s look at the credit card business. In 2024, the global credit card industry generated total revenue of $234 billion, consisting of installment interest, interchange fees, service charges and penalties, as well as unredeemed reward points.

This profit model is also built on user inertia.

Research shows that over 20% of cardholders do not redeem any points throughout the year; according to data from the U.S. Consumer Financial Protection Bureau, 3%-5% of points are forfeited annually due to account closure or expiration, with all such gains going entirely to the issuing institutions.

In addition, cardholders forgetting to switch to high-cash-back cards or missing low-interest installment windows are also part of the bank’s inertia-driven benefits.

And when tireless AI agents take over consumption decisions, the situation will be completely different.

Real-time matching of the best cashback card for purchases, automatic application for credit cards with superior rewards, automatic transfer of balances to low-interest installment plans, and automatic redemption of points before they expire—all tasks requiring user effort are handled automatically by AI.

Essentially, when the selection of financial products is no longer determined by consumption habits but by return-driven logic, the profits previously captured by banks will continuously shift to customers.

3 Pay "New Campaign"

Looking back at the history of the payments industry, it has been an "endless war."

Whenever everyone believes the market has become saturated, the landscape has solidified, and the story has reached its end, it always tears open new openings and restarts a fresh round of competition.

Just as the mobile payment market reached saturation, transaction volumes declined, and growth opportunities were fully exploited—leading many to believe that payments had no new stories left to tell—AI arrived.

As the core infrastructure of all commercial ecosystems, each payment war背后 is a shift in user habits and a restructuring of the industry landscape.

In this wave of AI-powered payments, AI is rewriting the fundamental logic of commerce.

As the reasoning capabilities of large models and the task decomposition abilities of agents continue to mature, AI is no longer just a chatbot that answers questions—it has become an independent agent that understands goals, breaks down steps, invokes tools, assesses risks, and completes transaction cycles.

This is a trading system entirely different from the past.

Currently, different market participants have varying entry strategies and core challenges to address, but ultimately, they are all paving the way for the future—seeking their place within the new ecosystem and vying for industry leadership.

Internet giants, with their control over users, scenarios, and service ecosystems, will become the super entry point of the AI era once they become users' "super assistants."

Card networks moved earlier, addressing how AI payments can integrate with existing clearing networks, comply with regulatory requirements, and build industry trust, providing the foundational support for the smooth operation of the new transaction system.

What about banks?

In addition to profit margins being compressed, a more pressing issue is:

If users increasingly rely on AI to handle the entire process—from thinking and deciding to executing trades—does it really matter which bank is serving me?