The competition for the settlement unit of AI Agent payments has officially begun, with contestants evolving from "stablecoin vs stablecoin" to "stablecoin vs bank deposits."

Written by: EX.IO Research Institute

An AI Agent independently discovered a digital service, invoked it, and then completed a payment—less than $2 per transaction, real-time settlement, each transaction passing through the bank's real-time risk control.

This sounds like a narrative native to crypto's stablecoin. But it is not. It is a technological validation jointly announced by a major bank in Hong Kong and a well-known tech company on October 9, 2026, using "bank deposits" itself as the settlement unit—tokenized bank deposits.

In short: the "settlement unit competition" of the AI economy has seen the entrance of bank-related players, who want to seize the opportunity to write the definition into their balance sheets before stablecoin issuers.

A "technological validation" is also a statement of direction

On October 9, major banks in Hong Kong and a well-known tech company jointly announced that they had completed a micro-payment technological validation test for AI Agents.

The test involved three components:

• The Tokenised Deposit Service (TDS) of major banks in Hong Kong: providing tokenized bank deposits and settlement capabilities, including real-time risk control checks through the MCP interface;
• The Anvita Flow network of a well-known tech company: enabling AI Agents to discover services, invoke services, and manage micro-payment settlements;
• The Jovay Testnet: a Layer-2 blockchain testing environment, serving as the underlying transaction layer.

In the test, an AI Agent demonstrated a complete closed loop of "discovering and invoking a digital service → initiating a micro-payment (commonly defined as under $2) → real-time settlement." The official statements were deliberately restrained: this "is merely a technological exploration, does not represent a commercial launch, nor is it a live product aimed at customers."

But behind this restrained rhetoric is a clear statement of direction. The question it seeks to answer is a singular one: whose "money" should the expenditure of the AI Agent ultimately be?

A struggle for the right to define

In the past two years, the narrative "AI Agents using stablecoins to buy computational power, data, and services" has been the hottest topic in the crypto industry. Multiple asset management and research institutions have echoed the same sentiment—for instance, the research institution Citrini Research proposed in its October 2026 report "Breaking The Wall" that AI Agents are "programmatic, 7x24, cross-application," hence the "money" should ultimately also become programmatic, 7x24, cross-application.

The question is: who will be the "programmatic money"?

The answer from the crypto-native perspective is stablecoins (USDC, USDT). The answer from the banking perspective is tokenized deposits—real bank deposits from within bank accounts represented in tokenized form on the blockchain.

These two seem similar, but their legal identities are entirely different:

• Stablecoins: liabilities of the issuer, essentially a "blockchain dollar promissory note," competing in reserves, compliance, and regulatory acceptance;
• Tokenized deposits: liabilities of the bank itself, representing "bank deposits" on the blockchain, backed by bank credit, and part of a regulated funding system.

The strategic significance of this test by major banks in Hong Kong and a well-known tech company lies precisely in using tokenized deposits rather than stablecoins to settle AI Agent micro-payments. This is an attempt for "bank funds" to seize the initiative over "stablecoins" to define the settlement unit for the AI economy.

For banks, this is not a technology showcase but a defensive move—if the payment layer of the AI Agent economy is ultimately occupied by stablecoins, banks will lose their intermediary status in "agent-to-agent" transactions, reducing them to mere KYC/fund channel providers. Tokenized deposits are the means by which banks can assert "I am still in the settlement layer" in the era of AI.

The global competition has officially begun

Placing this test within a timeline allows us to truly understand its significance:

• March 2026: Spain's Santander Bank partnered with Mastercard to complete the first end-to-end real-time payment test initiated by an AI Agent in Europe (based on Mastercard Agent Pay);
• May 2026: The Swiss Digital Asset Bank Sygnum completed on-chain transaction tests for AI Agents under regulatory oversight (clients must approve transactions individually);
• July 2026: Spain's CaixaBank partnered with Visa to complete the first credit card transaction initiated by an AI Agent (based on Visa Intelligent Commerce);
• October 2026: Major banks in Hong Kong × a well-known tech company completed technological validation of AI Agent micro-payments using tokenized deposits.

The previous cases were focused on whether "AI Agents can initiate payments." The collaboration of major banks in Hong Kong and a well-known tech company takes it a step further: it has replaced the "settlement unit" itself with tokenized deposits, in a combination of "globally systemically important banks (G-SIB) + one of Asia's largest fintech groups."

This means that the participants on this track have upgraded from "banks + card organizations" to a complete stack of "bank balance sheets + Asian payment/blockchain infrastructure."

What banks truly want to secure

On the surface, this is a validation of "AI + blockchain + regulated bank infrastructure." However, looking deeper, banks have three motivations worth exploring separately:

1. Defend their status as deposit and payment intermediaries. If the AI Agent economy defaults to using stablecoins for settlement, banks will lose the "flow of funds between agents" line, becoming purely KYC/fund pipelines. Tokenized deposits are a means for banks to retain the "essential path of funds."
2. Transform "risk control" into a differentiated selling point. Major banks in Hong Kong explicitly emphasized "real-time risk control checks via the MCP interface" in the test. This communicates to the market that autonomous payments by AI Agents require "the speed of software + the control of financial institutions," while "control" is precisely what stablecoin issuers and DeFi protocols cannot provide. Compliance, risk control, and accountability are banks' strongest assets in the age of AI payments.
3. Establish the definition of "regulated settlement units" before regulatory frameworks take shape. Global regulations (EU MiCA, US GENIUS Act, Hong Kong Stablecoin Regulation) are currently drawing boundaries for stablecoins. Banking-related tokenized deposits naturally fall on the "regulated funds" side and do not need to reapply for stablecoin licenses. Whoever is first to write "AI Agent settlement = tokenized deposits" as industry default will lock in the next generation's institutional settlement layer.

The long-term significance of the test: from "technical issues" to "institutional issues"

For the industry, the signals released by this test are more important than the actual technical actions completed:

• The competition for the settlement unit of AI Agent payments has officially begun, evolving from "stablecoin vs stablecoin" to "stablecoin vs bank deposits";
• The way banks participate is evolving: from "watching/testing" to "defining rails with balance sheets + risk control interfaces directly";
• "Regulated, accountable, redeemable" funds are becoming the premium point in the payment layer of the AI economy—which is precisely the structural advantage of tokenized deposits over unbacked stablecoins.

But it is also essential to view it rationally: this is merely a technological validation, not a commercial launch. There are still many unresolved issues between "AI Agent discovering services and paying $2" and "the AI Agent economy truly scaling," including identity verification, authorization boundaries, liability attribution, cross-border regulations, anti-money laundering, and more. It opens up "possibilities," not "results."

EX.IO Research Institute believes that the long-term significance of this test by major banks in Hong Kong and a well-known tech company lies not in "whether AI Agents can pay," but in upgrading the struggle for settlement units in the AI economy from "technical issues" to "institutional issues"—who's funds, whose track, and whose risk control ultimately become the default settlement layer for autonomous trading by AI agents.

For market participants, a key indicator to continue monitoring is: which will accelerate the real payment scenarios for the scaled AI Agent first—bank-related tokenized deposits or stablecoins—and how major global regulatory jurisdictions (especially Hong Kong, the EU, and the US) will delineate institutional boundaries between these two forms of "programmable currency."

The data and viewpoints in the article are sourced from Citrini Research's "Breaking The Wall" (October 2026), announcements and reports from Santander, Mastercard, Sygnum, CaixaBank, Visa, and related banks and tech companies as of October 12, 2026, and should be referenced in their original publication.