Alipay envisions an era moving toward agents.

In the internet era, Alipay, which "builds roads and bridges," is now coming to clear the bottlenecks of the AI era.

On May 26th, Alipay entered the booming intelligent agent track with the AI payment "family bucket." AI collection and payment, AI wallet, Token Pay services all debuted, forming a payment system centered around agents.

The "infrastructure fanatic" takes action—Alipay is betting on the growth of AI and agents, entering thousands of households, ultimately unable to bypass payment and settlement links. Li Jiajia, co-president of Ant Group's Digital Payment Business Group, told Wallstreetcn, "No business can function without payment as its foundation." Throughout history, the rise and fall of commercial societies has proven this point.

Right now, everyone is talking about agents, but the number of people actually using them daily worldwide is still sparse, far from entering households. According to statistics from Sider.ai's founder, even in software engineering—the field with the highest agent penetration—only about 600,000 to 1 million globally use agents deeply each day, about 3%–4% of the world's 27 million software developers; within the global 1 billion white-collar workers, the penetration rate is only about 0.01%.

This is not just a payment issue. The spread of agents is still limited by capability stability, task success rates, costs, user habits, and trustworthiness. Yet a new business bottleneck is emerging: as agents truly enter transaction, fulfillment, and settlement links, the current payment systems may not fully support these new roles.

A huge cognitive vacuum lies ahead, and this is exactly the position Ant wants to preempt.

For the past two years, the AI industry’s most lively part has been the models. Almost every few months, there’s a leap in technology. But from a business perspective, it’s clear AI still hasn’t entered the stage of large-scale commercialization.

Most AI today is more of a capability showcase. It can already create value but hasn’t truly participated in value exchange—which is the crucial step in the business world.

Ant told Wallstreetcn that all value flows cannot work without payment. “I can volunteer once or twice, but if there is a return, it becomes a flywheel, and the scale can further expand.”

This is actually reminiscent of e-commerce twenty years ago.

Before escrow transactions appeared, buyers didn’t trust sellers would ship, and sellers didn’t trust buyers would pay. The entire transaction chain lacked a trustworthy intermediary, so everything stayed limited to small, low-frequency, familiar circles. When this bottleneck was resolved, strangers started daring to transact online, and e-commerce truly began to expand.

Today, the trust issue facing agents is even more complex. Users have to trust agents not to spend money recklessly, merchants have to trust machine orders are genuine, developers have to trust they’ll get paid, and model vendors must trust token consumption can be clearly measured and settled.

Without payment and settlement, even the strongest AI remains at the tool level; only when value starts to flow can AI truly become an industry. Especially with the agent market, the emergence of massive micro-transactions is driving the need for new payment infrastructures.

Unlike traditional e-commerce, transactions between AI agents may be high-frequency, small amount, and automated. Circle’s data shows that the average payment per AI agent transaction is only $0.31; high fees and delayed settlement in traditional payment networks make them uneconomic in some microtransaction scenarios.

McKinsey predicts that by 2030, global B2C sales driven by AI agents will reach $3 trillion to $5 trillion. This blue ocean market is calling for a payment solution more suited to the machine economy.

Clearly, this is a market rich in imagination. Payment firms, stablecoin institutions, international card organizations, and even hardware manufacturers—top players in the payment ecosystem are entering, all wanting a share.

For Alipay, it must also seize this new gateway to the future. Its strategy is to solve the trust issue first.

A person close to Ant told Wallstreetcn: "Security has many layers. AI's hallucinations exist in the scenario part, but payment has no AI; payment demands extreme certainty, doesn't allow 'generative' uncertainty."

He mentioned this is why Ant now emphasizes the ACT protocol. And to ensure this information can't be tampered with, Ant is considering introducing blockchain capabilities. Zhu Lin, general manager of Ant Group’s AI Payment, also told Wallstreetcn, "Ant Chain may later play this role to solve the trust issue."

However, the real complexity in the AI era lies not only in payment, but in multilateral collaboration.

"Basic security and risk control on Alipay's side is definitely fine, but more will need extension to upstream and downstream, to different multilateral networks, merchants, brands, agents, consumers—all need to be in a protocol to mutually understand, or rather, so the machines of all parties can mutually read each other," the person said.

This means what Ant now wants to do is not just a payment tool but a commercial protocol for the AI era. Under the iceberg of the protocol is Ant’s ambition to build an ecosystem.

In Ant's view, what the AI industry really lacks today is not just models but ecosystems. Only when the developers behind agents truly make money can the whole system form a flywheel.

Li Jiajia told Wallstreetcn bluntly: "At present, Alipay's focus is on making agentic business work in China's large economic environment. Otherwise, all business can only write external coding, only sell tokens, and won't produce enough agents."

He mentioned that now Ant's consideration is "not business, not revenue, not competition, but how to accelerate the speed. We are the leader in this field, our mission is to speed things up."

Li Jiajia said, "Only when providers behind agents generate business value through paid models will more people be willing to create such agents; only when more people are willing to develop such agents will the agent ecosystem thrive. So we first have to solve the payment issue for agents."

This is also the background for the emergence of Token Pay.

Li Jiajia mentioned: "The way to do Token Pay is to let all model vendors focus more on refinement and improvement of their models. They are not good at selling tokens to consumers, so we provide a comprehensive Token Pay solution to lower their sales and operational costs."

What really needs to be solved is not just the act of "selling tokens." For model vendors and agent developers, consumer-oriented payment, subscription, risk control, refunds, splitting bills, invoicing, and merchant systems are all a complete set of commercial infrastructure. What Alipay wants to provide is precisely the packaging capability of this infrastructure.

As agents truly enter transaction systems, the business world itself will also begin to change.

An industry insider told Wallstreetcn: "Previously, user and merchant needs were not aligned. If the price wasn't psychologically right for the user, the order might get lost." But with agents, this is changing.

"Many people have their own expectations for products. If the original price is ten thousand, I'd be willing to buy at seven thousand. Then the user sets a task with the agent: notify me to buy when it hits seven thousand." This is essentially a reverse group-buying model.

From Ant's perspective, this is one of the most important changes with agents. When agents become a new role between people, and between people and transactions/services, previously scattered and hidden needs will be released more definitively.

Of course, most agent payments today are still in the early experimental stage. The so-called machine economy is far from entering ordinary people's daily life. Alipay's move now is more like a bet on a possible future: when agents transition from dialogue tools to transaction roles, the commercial world needs a new protocol for trust, authorization, settlement, and fulfillment.

This is the path Alipay wants to take toward the agent era.

It may not be able to determine the speed at which agents will spread, but it hopes to build the bridge before agents truly start to create, distribute, and circulate value. Otherwise, AI remains just a promise, hard to cross over into a self-running business system.

Risk Warning and DisclaimerThe market has risks, investment needs caution. This article does not constitute personal investment advice, nor does it take into account the special investment objectives, financial situation or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investment based on this, the responsibility is borne by yourself.