Over the past 2–3 years, the digital financial world has changed not only in terms of assets or products but is undergoing a true transformation at the "infrastructure" level.
The key factors are not just blockchain or cryptocurrencies but the rise of AI Agents that are beginning to independently make decisions and conduct financial transactions. Sam Broner's article “Tourists in the Bazaar” from a16z crypto poses a simple yet deeply systemic question: When AI starts paying on behalf of humans, what kind of payment system will truly support this? The answer is not the credit cards we use today.
What is happening is not just the addition of new payment channels but a change in the nature of the "payer" from human beings to automated software. When the payer changes, the rules of the financial system must change accordingly.
Typically, when we think of AI Agents, we imagine assistants answering questions, helping write emails, or analyzing data. But by 2026, AI Agents’ roles have expanded far beyond that. Many systems can analyze costs, evaluate value, select providers, and automatically "pay" to purchase resources or services from other systems—whether it’s buying GPU time for model processing, paying API fees to access databases, or settling cloud service fees by the second.
On some platforms, reports show that transactions generated by bots and agents constitute a very high proportion compared to individual users. This reflects a rapid increase in “non-human players” in finance. In many cases, AI Agents operate more in a Business-to-Business manner rather than as individual consumers. Broner likens credit cards to being designed for “tourists in the market” who buy items one at a time, pay, and leave. AI Agents don’t act like tourists but more like “locals” with long-term relationships with suppliers, contracts, credit lines, and continuous repeat purchases.
Structurally, credit cards have three key limitations in the AI world: first, they do not support very low-level micropayments, such as fractions of a cent, which are essential for streaming payment models based on actual usage. Second, transaction fees are relatively high compared to the value of small payments, and with large volumes, costs accumulate significantly. Lastly, fraud prevention and verification systems are designed for human behaviors, not for machine-to-machine transactions occurring continuously on a per-second basis.
As payment system demands change, stablecoins are increasingly seen as the "infrastructure" for the AI world, not just coins for crypto trading anymore. Currently, the total stablecoin market capitalization is roughly between $150 billion and $200 billion, occasionally approaching $300 billion. This figure has multiplied several times since the early 2010s, reflecting stablecoins’ expansion from speculative tools to practical use in digital financial systems.
Regarding transaction volume, many reports estimate that the total value of stablecoin transfers on-chain amounts to “tens of trillions of dollars per year.” Although figures vary depending on counting methods and include bot and internal transfers, the big picture is clear: stablecoins have become one of the largest value transfer rails in the digital world, especially in cross-border transactions, payments between technology companies, and settlement between crypto platforms, which have grown significantly over the past two years. Stablecoins suit the AI world because they are “programmable money” that can set advance payment conditions, such as paying based on actual usage, paying upon certain events, or paying in real-time—aligning with AI Agents’ operational models. They also cross borders easily, reduce intermediaries and costs, support micropayments well—especially on low-fee networks—and connect directly with APIs, enabling automatic, single-step service requests and payments.
The good news for Thailand is that it is not outside the trend. The Bank of Thailand launched the Programmable Payment Sandbox in 2024 to test new payment systems, including concepts of programmable payments and potential infrastructure to support stablecoins in the future. Several commercial banks are participating in trials to evaluate applications in real contexts.
However, challenges remain at two levels. The first is regulatory clarity: responsibilities between the Bank of Thailand and the Securities and Exchange Commission regarding digital assets and stablecoins need further definition, especially if commercial baht-pegged stablecoins are issued. The second is that Thailand’s AI Agent ecosystem is still in its early stages—in terms of developers, startups, cloud infrastructure, and business sector understanding. If Thailand wants to be a serious player in the AI + Stablecoin wave, investment in talent and open sandbox environments will be critical.
For investors, this trend provides several important strategic insights. First, AI Agent + Stablecoin is no longer a distant future vision but is being developed at the infrastructure level worldwide—in technology, finance, and payment networks. Next, due to the nature of payment systems, once an organization or platform moves its processes onto a new rail, reverting to old systems tends to be costly and complex. Early entrants into the ecosystem typically gain advantages in network effects and business relationships. At the same time, for Thai investors, closely monitoring the development of stablecoin sandboxes is advisable.