# Machines are already paying on the internet: this is the AI agent economy

> AI agents are already paying for digital services with stablecoins: blockchain opens a new economy of automated payments between machines
> By Covadonga Fernandez · AI Agents · 2026-03-10

Artificial intelligence agents are already starting to pay for things on the internet, such as access to APIs, data, computing power or digital services. And they do so mainly with [stablecoins,](https://www.allium.so/blog/author/allium/) cryptocurrencies pegged to the dollar such as USDC or USDT, which allow money to be sent automatically between systems.

In practice, the agent controls a digital wallet and signs blockchain transactions when it needs to use a service. The payment can be executed on a scheduled basis, for example every time it calls an API or uses a resource, and settled almost instantly without intermediaries such as banks or card networks. This allows software to pay directly to other systems or digital platforms without human intervention.

## AI agent economy

It may sound like a technical detail, but many in the tech industry believe it could become one of the biggest changes in the digital economy. For now, this economy is still very small. In 2025, an estimated 40,000 on-chain agents made automatic payments that moved around 50 million dollars in total, according to fintech sector estimates based on blockchain activity data cited by specialized media. The figure is minimal compared with the total volume that stablecoins already move each year, but it shows that the idea of software paying for digital services is already beginning to take shape.

Compared with the rest of the cryptocurrency market, it is practically insignificant: stablecoins processed tens of trillions of dollars in transactions that same year. But that difference is precisely what catches the attention of many technology companies. The financial infrastructure already exists and moves enormous volumes of money; what does not yet exist is an agent economy operating at scale. If those systems begin to use automatic payments en masse to buy data, computing or digital services, the number of transactions could multiply rapidly.

## Software as an economic actor

For decades, the internet was essentially a network of information. People used it to read news, send emails or search for data. Over time it became a network of commerce, where users could buy products, hire services or pay for digital subscriptions. But now it may be entering a third phase. A network where software not only processes information but also actively participates in the economy. The new artificial intelligence agents are no longer limited to answering questions or generating text.

**[More than 10,000 AI bots visit Observatorio Blockchain in just 24 hours](https://observatorioblockchain.com/ia/mas-de-10-000-bots-de-ia-visitan-observatorio-blockchain-en-solo-24-horas/)**

More and more systems can plan tasks, search for information, compare prices, execute actions in applications and operate autonomously on the internet. This means software is beginning to behave like an economic actor. And when something becomes an economic actor, it inevitably needs a way to pay.

## Blockchain

The problem is that the current financial system was never designed for machines. Credit cards, bank transfers and traditional payment systems require bank accounts, legal verification, personal authentication and regulatory processes designed for people. An AI agent, for now, cannot open a bank account, complete a KYC process or receive a verification code by SMS.

From the point of view of the traditional financial system, it does not exist. However, there is a technology that does allow an AI agent to handle money directly: blockchain. With a simple digital wallet, an agent can send programmable money in seconds, without permissions, without intermediaries and without the need for a human identity. This turns software into an autonomous economic actor.

In theory, agents could use any cryptocurrency to make payments. But it is stablecoins that seem most suited to this role. Stablecoins are cryptocurrencies pegged to the value of the dollar or the euro. This eliminates one of the main problems of other cryptocurrencies such as bitcoin or ethereum: volatility.

## 33 trillion dollars in transactions during 2025

For an automated system executing millions of transactions, price stability is essential. An agent paying for digital services cannot risk the cost of an operation changing ten percent within minutes. In this sense, stablecoins offer digital money that maintains a stable value and can move globally in seconds.

The infrastructure for this type of digital money already exists and has reached a surprising scale. According to data cited by Bloomberg, stablecoins processed around 33 trillion dollars in transactions during 2025. That figure places these digital currencies in the same order of magnitude as some of the world's largest payment networks. Although much of this volume is tied to cryptocurrency trading, what matters is that it shows the financial highway is already built. There is an infrastructure capable of moving enormous amounts of digital money quickly and programmatically. All that is missing is the next big use case to take advantage of that infrastructure.

## The automated micropayment economy

That case could be the automated micropayment economy. Humans make relatively few economic transactions in their daily lives. We buy food, pay subscriptions, book trips or purchase physical goods. Machines work in a completely different way. An artificial intelligence agent could pay fractions of a cent for every digital action it performs. It could pay a small amount every time it queries an API, buys real-time data, rents computing power, reads a newspaper or accesses an AI model. Every digital action could have a minimum price attached. This means the total number of payments could multiply exponentially. A single agent could make thousands or even millions of micropayments every day.

The payments industry is beginning to recognize that something profound may be changing. According to the Global Payments Report by Boston [Consulting Group](https://web-assets.bcg.com/25/91/2269153c468ca43684442f055cb0/2025-global-payments-report-sep-2025.pdf), revenues in the global payments sector reached approximately 1.9 trillion dollars in 2024 and could grow to 2.4 trillion by 2029. The report identifies several structural forces that are transforming the financial sector, including the emergence of agentic artificial intelligence, digital currencies and new fintech models. BCG analysts estimate that artificial intelligence agents could influence more than one trillion dollars of e-commerce spending in the coming years, as consumers increasingly delegate purchasing decisions to automated systems.

## Machine-to-machine payment economy

However, the machine-to-machine payment economy is still in its early stages. Despite the enormous volume of transactions that stablecoins already process, most of those operations are still tied to digital asset trading. BCG's own report estimates that only around one percent of stablecoin volume currently corresponds to payments in the real economy. But for many analysts, that small percentage reveals something more interesting than worrying: the market has not really begun yet. The infrastructure is built, the technology is available and technology companies are beginning to experiment with new payment models integrated directly into the internet.

If artificial intelligence agents begin to buy digital services autonomously, such as data, computing, software or access to artificial intelligence models, an entirely new economy based on automated transactions could emerge.

## A network of humans trading with humans

For centuries, the world economy has been a network of humans trading with humans. The internet expanded that network on a global scale. But now an economy could appear where software also participates as buyer and seller. Algorithms negotiating prices, agents paying for digital services and autonomous systems exchanging value continuously. If that vision materializes, the question will no longer be whether machines can pay. The question will be how much money machines will end up moving.
