Hong Kong Bank and Tech Firm Test AI Agent Micro-Payments Using Tokenized Deposits

EX.IO Research Institute | October 13, 2026

An AI agent independently discovered a digital service, invoked it, and completed the payment—less than $2 per transaction, settled in real time, with every transaction passing through the bank’s real-time risk control.

This sounds like a crypto-native stablecoin narrative. But it isn’t. It’s a technology validation announced on October 9, 2026, by Hong Kong’s leading banks in collaboration with a well-known tech company, using “bank deposits” themselves as the settlement unit—tokenized deposits.

In the "battle for the settlement unit" of the AI economy, a banking-sector player has entered—and aims to lock in the definition within its own balance sheet before stablecoin issuers can.

A "technical validation," and more importantly, a statement of direction

On October 9, Hong Kong’s major banks partnered with leading technology companies to successfully complete a pilot test of AI Agent micropayment technology.

Test stacked three blocks:

Tokenised Deposit Service (TDS) from Hong Kong’s major banks: Offers tokenised bank deposits and settlement capabilities, including real-time risk control checks via the MCP interface;
Anvita Flow network from a leading tech company: enables AI agents to discover services, invoke services, and manage micropayment settlements;
Jovay Testnet: A Layer-2 blockchain test environment serving as the underlying transaction layer.

During testing, an AI agent demonstrated a complete闭环: discovering and invoking a digital service → initiating a micropayment (typically defined as under $2) → real-time settlement. The official statement was deliberately restrained: this "is merely a technical exploration and does not represent commercial deployment or a live product for customers."

But behind the restrained wording is a clear stance on direction. There is only one question it seeks to answer: Whose money should the funds spent by AI Agents truly be?

A battle over definition rights

Over the past two years, the narrative that “AI agents use stablecoins to buy compute, data, and services” has been the hottest in the crypto industry. Multiple asset management and research firms have been saying the same thing—for example, in its October 2026 report, “Breaking The Wall,” research firm Citrini Research proposed that AI agents are “programmatic, 7×24, and cross-application,” so “money” should ultimately become programmatic, 7×24, and cross-application as well.

The question is: who will be this "programmed money"?

The crypto-native answer is stablecoins (USDC, USDT). The banking sector’s answer is tokenized deposits—bringing real deposits from bank accounts onto the blockchain in tokenized form.

They may look similar, but their legal identities are completely different:

Stablecoin: a liability of the issuer, essentially a "blockchain-based IOU for USD," requiring competition in reserves, compliance, and regulatory acceptance;
Tokenized deposits: liabilities of the bank itself, representing a blockchain-based mapping of "bank deposits," backed by bank credit and part of the regulated funds system.

The strategic significance of this test by Hong Kong's major banks and leading tech companies lies in using tokenized deposits, rather than stablecoins, to settle micropayments for AI agents. It is an attempt by "bank funds" to define the settlement unit for the AI economy ahead of "stablecoins."

For banks, this isn’t a technology showcase—it’s a defense. If the payment layer of the AI agent economy is ultimately dominated by stablecoins, banks’ payment fees, deposit bases, and customer relationships will be rendered irrelevant in a world where agents transact directly with agents. Tokenized deposits are banks’ way of asserting, “We’re still in the settlement layer,” in the age of AI.

The global competition has officially begun

Place this test on the timeline to truly understand its position:

March 2026: Santander Bank of Spain, in collaboration with Mastercard, completes Europe’s first end-to-end real-time payment test initiated by an AI Agent (based on Mastercard Agent Pay);
May 2026: Switzerland’s digital asset bank Sygnum completes a regulated bank AI Agent on-chain transaction test (requiring individual customer approval and signature for each transaction);
July 2026: Spain’s CaixaBank, in partnership with Visa, completes the first credit card transaction initiated by an AI Agent (powered by Visa Intelligent Commerce);
October 2026: Hong Kong’s major banks × renowned tech companies complete a technical validation of AI agent micropayments using tokenized deposits.

The top few are focused on whether AI agents can initiate payments. Leading banks in Hong Kong and prominent tech companies have gone a step further: they’ve replaced the settlement unit itself with tokenized deposits, combining a globally systemically important bank (G-SIB) with one of Asia’s largest fintech groups.

This means the participants in this pipeline have upgraded from “banks + card networks” to a full stack of “bank balance sheets + Asian payment/blockchain infrastructure.”

What banks truly want to lock down

On the surface, this is a technical integration validation of "AI + blockchain + regulated banking infrastructure." Looking deeper, banks have three motivations worth examining individually:

Defend the position as deposit and payment intermediaries. If the AI agent economy defaults to settling in stablecoins, banks will lose the stream of funds between agents and be reduced to mere KYC and fiat on/off-ramps. Tokenized deposits are how banks preserve their role as the essential pathway for fund flows.
Turn "risk control" into a differentiated selling point. Hong Kong’s major banks explicitly emphasize during testing that they provide real-time risk control checks via the MCP interface. This sends a clear message to the market: autonomous payments by AI agents require "software speed + financial institution control"—and it’s precisely this "control" that stablecoin issuers and DeFi protocols cannot provide. Compliance, risk control, and accountability are banks’ most valuable assets in the age of AI-powered payments.
Secure the definition of "regulated settlement unit" before regulations solidify. Global regulators (EU MiCA, U.S. GENIUS Act, Hong Kong stablecoin rules) are drawing boundaries around stablecoins. Bank-issued tokenized deposits naturally fall under "regulated funds" and do not require a separate stablecoin license. Whoever establishes "AI Agent settlement = tokenized deposits" as the industry default will lock in the next-generation institutional settlement layer.

The long-term significance of testing: from a "technical issue" to an "institutional issue"

For the industry, the signal this test sends is more important than the actual technical actions it accomplished:

The battle over the settlement unit for AI agent payments has officially begun, shifting from "stablecoin vs. stablecoin" to "stablecoin vs. bank deposits";
The way banks participate is evolving: from "watching/piloting" to "directly defining rails using their balance sheets and risk control interfaces";
Regulated, accountable, and redeemable funds are becoming a premium feature in the AI economy's payment layer—this is precisely the structural advantage of tokenized deposits over unbacked stablecoins.

But it’s important to keep things in perspective: this is merely a technical proof of concept, not a commercial launch. Between an AI agent discovering a service and paying $2, and a truly scaled AI agent economy, there remain unresolved challenges—identity verification, authorization boundaries, liability assignment, cross-border regulation, anti-money laundering, and more. What this opens is “possibility,” not “outcome.”

EX.IO Research believes that the long-term significance of this test by Hong Kong’s major banks and leading tech companies lies not in whether “AI agents can make payments,” but in how it has elevated the debate over the settlement unit for the AI economy from a “technical issue” to an “institutional one”—whose funds, whose infrastructure, and whose risk controls should serve as the default settlement layer for autonomous AI transactions.

For market participants, key indicators to monitor continuously are: which of bank-issued tokenized deposits or stablecoins will first successfully enable large-scale, real-world payment scenarios for AI agents, and how major global regulatory jurisdictions—particularly Hong Kong, the EU, and the U.S.—will define institutional boundaries between these two forms of programmable money.

Source

The data and opinions in this text are sourced from Citrini Research’s “Breaking The Wall” (October 2026), Santander, Mastercard, Sygnum, CaixaBank, Visa, and related banks and technology companies’ public announcements and reports, as of October 12, 2026, based on their original publications.

Important Notice

This report, prepared by EX.IO Research Institute based on publicly available information, is a forward-looking analysis under hypothetical scenarios. It is provided for general informational purposes only and does not constitute any investment advice, offer, or solicitation. Virtual assets and tokenized products involve high risks; their prices may rise or fall, and investors may lose their entire principal. These products are intended solely for qualified professional investors. Please carefully review the product documentation and risk disclosures, and consult a licensed professional before investing.