HSBC, Ant Digital test AI-agent payments using tokenized deposits

cointelegraphPublished on 2026-10-10Last updated on 2026-10-10

Abstract

HSBC and Ant Digital Technologies have conducted a test enabling AI agents to access digital services and make micropayments (typically under $2) using tokenized bank deposits, with real-time settlement on a blockchain testnet. The trial combined HSBC's Tokenised Deposit Service, Ant Digital's Anvita Flow network for AI agent coordination, and the Jovay Testnet. It focused on technical verification, not a commercial launch. This aligns with broader banking experiments, as seen with Santander, Sygnum, and CaixaBank, exploring AI-agent-initiated payments. However, some argue traditional banking infrastructure may be ill-suited for AI-driven finance, prompting projects like Augustus Bank to build new systems. Investment research suggests autonomous AI agents could boost demand for programmable, always-on financial infrastructure like blockchain networks.

British banking giant HSBC and Ant Digital Technologies have tested a system that lets AI agents access digital services and make micropayments using tokenized bank deposits, with transactions settled in real time on a blockchain testnet.

According to Friday’s announcement, the test combined HSBC’s Tokenised Deposit Service with Ant Digital’s Anvita Flow network, which enables AI agents to find and use services, and Jovay Testnet, a layer-2 blockchain testing environment.

HSBC provided settlement capabilities and real-time risk checks, while Ant Digital’s network coordinated service access and payments. The demonstration showed an AI agent selecting a digital service and completing a payment. The companies described the transactions as micropayments, typically defined as less than $2.

The companies said the test was limited to technical verification and did not represent a commercial launch or live customer offering.

Related: Sui tunnels hit 40.6M TPS in live AI agent test

Banks experiment with AI-agent payments

HSBC is among several banks testing how AI agents could initiate financial transactions on customers’ behalf.

In March, Spanish banking giant Santander completed an AI-agent-initiated payment using Mastercard’s Agent Pay infrastructure in a controlled test involving the bank’s live payment systems.

Swiss digital asset bank Sygnum followed in May with a test of AI-agent-driven transactions on a blockchain mainnet, with customers required to approve and sign each transaction. Spanish banking group CaixaBank also completed an AI-agent-initiated card transaction using Visa Intelligent Commerce and existing merchant payment systems.

However, not everyone believes established banks can adapt their existing infrastructure to AI-driven finance. In a May interview with Cointelegraph, Augustus Bank CEO Ferdinand Dabitz argued that traditional clearing banks rely on decades-old systems designed for human operations rather than automated, around-the-clock transactions.

Augustus is developing a US bank built around stablecoins and AI-driven operations, betting that purpose-built infrastructure can replace parts of traditional banking systems.

Related: Nasdaq Ventures, Deutsche Bank back Elliptic in $120M funding round

AI agents could drive blockchain adoption

Investment research firm Citrini Research recently explored blockchain’s potential role in AI-driven commerce. In an Oct. 8 report titled Breaking The Wall, the firm argued that autonomous AI agents could increase demand for programmable financial infrastructure.

The firm said traditional financial systems, designed primarily for human users, may need to adapt as AI agents increasingly handle transactions across applications. It argued that blockchain networks could provide the always-on infrastructure needed to move money and financial assets programmatically.

“AI agents move programmatically, 24/7, across applications, and it’s only logical that money and financial assets eventually will, too,” Citrini wrote.

Mafazine: France is world’s most dangerous place for Bitcoiners: 90 attacks in 7 months

Related Questions

QWhat did HSBC and Ant Digital Technologies test together, and what technology was used for settlement?

AThey tested a system that allows AI agents to access digital services and make micropayments using tokenized bank deposits. The transactions were settled in real-time on a blockchain testnet.

QWhat are the specific systems or networks involved in the test conducted by HSBC and Ant Digital?

AThe test combined HSBC's Tokenised Deposit Service, Ant Digital's Anvita Flow network (which enables AI agents to find and use services), and the Jovay Testnet, a layer-2 blockchain testing environment.

QWhat is the main argument made by Citrini Research regarding AI agents and blockchain technology in their report 'Breaking The Wall'?

ACitrini Research argued that autonomous AI agents could increase demand for programmable financial infrastructure. They stated that blockchain networks could provide the always-on infrastructure needed to move money and financial assets programmatically, as AI agents operate 24/7 across applications.

QWhich other banks have tested AI-agent-initiated payments, and what were some key details of their tests?

ASeveral other banks have tested AI-agent payments: Santander completed a test using Mastercard's Agent Pay infrastructure with its live payment systems; Sygnum tested AI-agent-driven transactions on a blockchain mainnet, requiring customer approval for each transaction; and CaixaBank completed an AI-agent-initiated card transaction using Visa Intelligent Commerce and existing merchant systems.

QWhat criticism did Augustus Bank's CEO raise about traditional banks regarding AI-driven finance?

AAugustus Bank CEO Ferdinand Dabitz argued that traditional clearing banks rely on decades-old systems designed for human operations, which may not be suitable for automated, around-the-clock transactions required by AI-driven finance. He believes purpose-built infrastructure is needed.

Related Reads

OpenAI launches AI payments in ChatGPT with Visa, Mastercard and Amex

OpenAI launched a feature letting AI agents make purchases for users directly within ChatGPT through existing payment networks, in partnership with American Express, Mastercard, and Visa. The wallet uses biometric verification like facial recognition or fingerprints to authorize transactions, and payment networks generate tokenized card credentials for each approved purchase, keeping users' actual card details from merchants and AI agents. The service is rolling out to ChatGPT Pro subscribers in the US with eligible US-issued cards, with plans to eventually support purchases within preset spending limits and other restrictions without requiring approval for each transaction. The launch reflects AI companies moving beyond product recommendations into executing transactions, potentially changing how consumers interact with online merchants. OpenAI is working with payment processors and fraud prevention firms to support AI purchases through existing checkout systems. It joins a growing field: Google introduced its Agent Payments Protocol in 2025, Visa and Mastercard have launched AI transaction initiatives, and crypto companies like Coinbase (x402 protocol), Circle (Agent Stack), and Crossmint are building payment infrastructure for AI agents. Global agentic commerce transaction value is expected to reach $1.5 trillion by 2030, up from an estimated $8 billion in 2026, according to Juniper Research.

cointelegraph1h ago

OpenAI launches AI payments in ChatGPT with Visa, Mastercard and Amex

cointelegraph1h ago

Kraken parent Payward gains market share as slower trading weighs on earnings

Kraken parent Payward nearly doubled its share of global crypto spot trading to 6.5% in September from 3.3% a year earlier, even as weaker trading activity weighed on earnings. The market-share gain reflects a larger portion of spot trading relative to competitors rather than increased activity; the company handled $353 billion in transactions in Q3, down 30% year over year. Adjusted EBITDA fell to $60 million from $178.6 million a year earlier, and total platform margin fell 12% to $487 million. Payward has spent two years expanding beyond its core exchange through acquisitions in derivatives, proprietary trading, tokenized assets and payments, including NinjaTrader, Breakout, Bitnomial and Reap. Derivatives contracts with per-contract fees rose 84% to 66 million, and margin from subscriptions and other paid services rose 59% to $125 million. Results improved from Q2, with total transaction volume up 22% and platform margin up 12% to $487 million from $436 million. Margin from customer-held assets, including staking, rose 26% to $128 million, accounting for roughly half the quarterly increase. Adjusted EBITDA climbed to $60 million from $23 million, though it remained down 66% year over year. Asset- and access-based services accounted for 52% of total platform margin, up from 46%. Payward confidentially filed for a US IPO in November 2025 but paused plans in March; the earliest potential listing window under consideration was Q2 2027.

cointelegraph2h ago

Kraken parent Payward gains market share as slower trading weighs on earnings

cointelegraph2h ago

Metaplanet swings to $170M loss as Bitcoin sale hits earnings

Metaplanet, the world's second-largest publicly traded corporate Bitcoin holder, reported a net loss of 27.16 billion yen (about $170 million) in the first nine months of 2026, compared with a profit of 13.53 billion yen ($84.6 million) a year earlier. The loss was driven by a 19.17 billion yen ($119.8 million) loss on the sale of 10,000 BTC and an 8.04 billion yen ($50.3 million) loss from a decline in the value of its Bitcoin holdings, which outweighed gains in its operating business. During the quarter, Metaplanet sold 10,000 BTC for 124.7 billion yen, temporarily held the proceeds in cash, stablecoins and other liquid assets, then spent 149.9 billion yen to buy 11,000 BTC at an average price 9.3% higher per coin. The company said the sale was intended to demonstrate to rating agencies and bond investors that it could convert Bitcoin into cash to meet financing obligations. Holdings ended September at 44,000 BTC, a net increase of 1,000 coins. Metaplanet borrowed against its Bitcoin during the quarter, raising its credit line debt to $649.5 million at the end of September from $414 million at the end of June. It funded purchases from the second quarter onward with borrowed funds, bond proceeds, exercised stock acquisition rights and revenue from its Bitcoin options business. Interest expense rose 56% quarter over quarter to 1.36 billion yen ($8.5 million), and short-term borrowings and bonds payable totaled 110.5 billion yen ($691 million). For the nine months through September, revenue rose 30.5% to 5.9 billion yen ($36.9 million) and operating profit increased 26.5% to 3.48 billion yen ($21.8 million). The company cut its full-year revenue forecast to 6.5 billion yen ($40.6 million) from 16 billion yen ($100 million). The results come as Bitcoin treasury companies face pressure from falling prices and index removals; MSCI confirmed Strategy, Metaplanet and Upexi will be removed from its equity indexes at the Nov. 30 market close. Bitcoin fell below $70,000 on Friday, trading around $68,600, down about 9% over the previous week.

cointelegraph5h ago

Metaplanet swings to $170M loss as Bitcoin sale hits earnings

cointelegraph5h ago

Trading

Spot
活动图片