Will the biggest user of stablecoins be agentic AI?

Even if you aren’t a digital currency holder, an agentic AI might soon buy something using stablecoin on your behalf. In fact, as Galaxy Digital Founder and CEO Michael Novogratz told Bloomberg TV in a recent interview, “in the not-so-distant future, the biggest user of stablecoins is going to be AI.” To some, that might sound like the plot out of an Isaac Asimov novel, especially since stablecoin is a type of cryptocurrency—a dirty word for some skeptics. But, as it turns out, that future is already here.

Last month, Coinbase announced that it is giving AI agents some of the same financial tools their human customers use, asserting that “every agent deserves a wallet.” Payments MCP, as the new feature is called, will allow LLMs such as Claude and Gemini to make payments on customers’ behalf, rather than the current technology of simply recommending purchases. (MCP, or Model Context Protocol, allows AI agents to be aware of the context of their interactions with the other components, e.g., users or LLMs.)

The AI agents’ currency will be stablecoins—digital currency pegged to a specific currency (such as the USD or EUR) or asset. It’s not enough for AI agents to “just analyze or recommend,” per the Coinbase release; rather, they should “act in the global economy.”

This development is actually a boon in many ways, according to Laurent Lambert, IBM’s Head of Blockchain for Australia and New Zealand. “I’m not surprised by Novogratz’s assertion,” he told IBM Think. “[AI agent commerce] has always been one of the premises of blockchain.”

You might well wonder why AI agents would use stablecoin, rather than just using fiat.

For one thing, stablecoin reduces transaction friction, Lambert said. Under the traditional setup, a transaction requires reconciling the ledgers; if it’s an overseas transaction, the transaction becomes even more involved because the banks must reconciliate at an international level. “It’s very cumbersome,” he said. “Think of a bank as a house that’s completely locked up, and every time you want to go in there and ask for something, someone needs to open the door. Usually, a human has to validate your ID and consent, and it takes days.”

Digital money, by contrast, has its ledger on the blockchain. “This allows you to do instant payments without having to clear the ledgers,” Lambert said. “The ledgers are balancing themselves using blockchain for security, for validating the transaction and making sure everyone agrees with it.”

“Obviously, blockchain brings a lot of things to the table, but most importantly, it’s the ability to exchange value on the internet,” he said. “At the moment, AI agents are still kind of boxed into the way we program them. But ultimately, they will start connecting to each other.” And when they do, the coin of the realm will be stablecoin.

In his Bloomberg interview, Novogratz illustrated his point with an example of an AI agent ordering groceries online on a consumer’s behalf, using stablecoin as currency. But the potential for this technology goes beyond your AI-powered fridge noticing you’re out of cheese and getting some for you. From an enterprise point of view, the potential is vast, Lambert said.

Friction is what drives up the cost of transactions. It’s one of the reasons why, for example, it can cost an additional USD 25-50 to wire funds from a US bank to a bank in the eurozone, once the sending and receiving bank have each collected their wire transfer fees. Friction is what makes micropayments—a longstanding theme of “future of money” discussions—untenable under the traditional monetary system.

By contrast, explained Lambert, a scenario where AI agents are transacting in stablecoins makes micropayments much easier. “Let’s say you just need a little set of data, like weather data,” he said. “You can just call an API and that API can be paid one cent instantly. The fiat system is just not geared for that,” as the cost of the transaction would greatly exceed one cent.

Lambert illustrated what it would look like if these agentic transactions were to become a part of everyday life, using Tesla as a hypothetical use case. “You have to charge a Tesla every day,” he said. “Imagine a world in the near future where your Tesla is parking at a shopping center, and it automatically connects to a charging plate. You could pay for this through a transaction on the blockchain using a smart contract.”

A smart contract is a computer protocol containing instructions for when and how you’ve pre-determined a transaction can occur. In fact, according to Lambert, “Smart contracts are the most important innovation to come out of crypto.” In the Tesla example, he said, “Since you can program the money, the transaction will automatically happen. The funds will be deducted from your wallet and credited to the shopping center that owns the charging station.”

Worried about an agent making a purchase you didn’t ask for?  With a smart contract, the transaction cannot occur until the pre-determined criteria are met. In this scenario, the Tesla isn’t making rogue purchases; it’s following the precise terms of the smart contract, the pre-programmed rules saying that under X circumstances, the Tesla can do Y.

“Smart contracts are why stablecoin is going to start being powerful,” said Lambert, adding that he thinks this will be especially true in the world of finance, be it between banks and customers or inter-bank. “Imagine millions of transactions a minute,” he said.

Agentic AI fits in well with this technology, explained Lambert. “Blockchain and stablecoins provide a digitally aligned financial backbone,” he said. “These AI agents won’t just record and verify transactions; they’ll be able to autonomously make decisions, negotiate terms and optimize processes in real time.”

According to Lambert, agentic stablecoin use creates a far more efficient and cost-effective way to operate at scale. Lambert points out that major financial players are already moving in this direction—JP Morgan, for example, has launched its own stablecoin. Bank of America intends to do the same, CEO Brian Moynihan said in a July post-earnings call. As is Citi, as Biswarup Chatterjee, Global Head of Partnerships and Innovation, confirmed to CNBC in October.

“The main challenges will revolve around regulatory compliance and integrating these capabilities within existing legal frameworks,” Lambert said.

These days, security is always top of mind, especially when it comes to AI. But where blockchain transactions are concerned, said Lambert, the real risk doesn’t come from the apparatus itself. “Hacking at the blockchain level is very unlikely. The blockchain provides a certain level of security through the cryptography,” he said. “Sure, there are some immature blockchains. But look at Bitcoin, for example; it’s been around for a long time”—16 years, in fact—“and no one has hacked it.” Rather, the burden of security, to some extent, is up to the individual. As Lambert explained, “The challenge is at the wallet level,” where threat actors might get access to your digital assets or even steal your identity.

But in future scenarios regarding stablecoin or any digital asset, are these hackers always going to be human? Or will the AI agents themselves become hackers? “You’re imagining a scenario where the AI agents are sentient,” he said jokingly. “There’s always going to be a risk. Obviously, when there is code, there will be gaps. But blockchain technology and security have become proven.”