Seoul is running one of the most technically advanced live tests of a model that other central banks are watching closely: a two-tier digital currency system where commercial banks issue consumer-facing tokens backed 1:1 by Korean won, while the central bank controls a wholesale settlement layer underneath.

Phase 2 moves beyond controlled simulations. Participants are now spending real digital tokens, receiving initial government subsidy disbursements through them (beginning with travel allowances to government employees, with broader subsidy categories to follow), and testing payments made autonomously by AI systems, a capability that, as far as publicly disclosed, no other G20 central bank has formally included within a CBDC-linked framework in live testing.

What Changed in Phase 2

Phase 1, which ran through 2024 and 2025, confirmed that deposit tokens could be issued, transferred, and redeemed across bank systems without technical failure. Peer-to-peer transfers proved difficult, and the pilot relied on simulated rather than live transactions. Phase 2 addresses both gaps. The expanded rollout adds peer-to-peer transfers, biometric authentication, automatic top-ups from linked bank accounts, interest payments on token balances, support for recurring payments via CMS, and cash receipt issuance.

The legal foundation for the programme was established in 2024, when enabling legislation classified deposit tokens as equivalent to bank deposits, giving the model its regulatory basis before Phase 2 began.

Two banks, Kyongnam Bank and iM Bank, joined the original seven participants. The nine-bank consortium now includes KB Kookmin, Shinhan, Woori, Hana, IBK, NH Nonghyup, Busan Bank, Kyongnam Bank, and iM Bank, representing Korea's major commercial banks.

The cost efficiency numbers are the most concrete data point in the pilot so far. According to CryptoRank, administrative costs for distributing value through the deposit token system run approximately 0.2% of the amount distributed, compared with roughly 1.5% under conventional government payment infrastructure. CryptoRank presents these as projected or targeted figures rather than confirmed operational results from Phase 2 live transactions. Settlement is instant and around the clock rather than limited to business hours.

Kim Dong-sub, the Bank of Korea's Digital Currency Planning Head, described the commercial opportunity for participating banks in March: "Participating banks are actively securing diverse use cases, such as large businesses and small merchants with high public relevance and significant payment fee burdens, focusing on the potential for drastically reduced fees when using digital currency for payments."

The Subsidy Play and the AI Pilot

The most significant policy dimension of Phase 2 is the use of deposit tokens for government subsidy disbursements. South Korea has targeted approximately 110 trillion won (around $79.4 billion USD) in government subsidies for deposit token distribution across categories including agricultural support, small business grants, social welfare payments, and electric vehicle incentives.

Initial testing began with government employees receiving travel allowances as deposit tokens. EV charging subsidies were targeted for the first live disbursement in the first half of 2026.

The AI agent payment component is less visible but technically notable. The Bank of Korea is testing scenarios where AI systems purchase goods and services autonomously, without human initiation of each transaction. As far as publicly disclosed, no comparable G20 central bank has formally included machine-to-machine payments within a CBDC-linked framework in live testing.

What the Architecture Means Outside Korea

South Korea's deposit token model sits between two more common approaches. China's e-CNY was originally structured as a direct retail issuance by the People's Bank of China, though in January 2026 the PBoC reclassified e-CNY as deposit liabilities, a significant structural shift that moves it closer to the two-tier model Korea has operated from the outset.

Private stablecoins such as USDC are issued by non-bank entities and, unlike the Korean deposit token model, do not carry a central bank settlement backstop, though some stablecoin issuers have pursued banking licenses in certain jurisdictions. The Korean approach preserves the existing commercial banking system while adding programmable settlement infrastructure on top of it. The Bank of Korea controls the rails; the banks control the customer relationship.

This matters to countries in South Asia and Sub-Saharan Africa that are building or revising CBDC frameworks. Nigeria's eNaira has struggled with adoption due to limited merchant integration and weak user incentives. South Korea's strategy of routing government payments through deposit tokens from the start addresses that problem directly. India's Reserve Bank, which has been cautious about private stablecoin integration, may find the BOK model useful as a reference point for how to add programmability to payments without ceding monetary control to private issuers. South Africa's Financial Sector Conduct Authority, which is actively developing its own stablecoin regulatory position, represents another jurisdiction for which the BOK framework offers a relevant point of comparison.

One important caveat applies. South Korea has near-universal banking penetration and reliable internet infrastructure. Ghana's eCedi, which runs on cards and devices without requiring a live connection, is better suited to markets with significant unbanked populations or inconsistent connectivity. The Korean architecture is not a universal template.

Cross-Border Dimension and Stablecoin Policy

The Bank of Korea is also participating in Project Agorá, a BIS-coordinated initiative involving seven other central banks: the Federal Reserve Bank of New York, the Bank of England, the Bank of Japan, the Bank of France (representing the Eurosystem), the Bank of Mexico, and the Swiss National Bank.

A May 2026 BIS report confirmed that the project demonstrated around-the-clock, multi-currency atomic settlement of wholesale cross-border payments using tokenised reserves and commercial bank deposits. Real-value testing is the next step.

For South and Southeast Asian and West African workers sending remittances to and from Korea, including those from Nepal, the Philippines, and Sri Lanka, the practical outcome of that work could eventually reduce transfer costs on those corridors.

Governor Shin Hyun-song, who began his four-year term in April 2026, made CBDCs and deposit tokens a centerpiece of his inaugural address and confirmed continued participation in Project Agorá.

Notably, he made no mention of stablecoins despite previously acknowledging they could play a supplementary and competitive role in digital payments.

The BOK's current framework positions regulated commercial banks as the only appropriate issuers of won-denominated digital instruments, a stance that functions as an implicit policy signal to jurisdictions still debating whether to permit private stablecoin issuance. That signal carries particular weight domestically: South Korea's Digital Asset Basic Act remains delayed due to unresolved disputes over who should have authority to issue Korean won-pegged stablecoins, making the BOK's deposit token push a de facto policy position in an unresolved legislative debate.

The BOK has not announced a timeline for any permanent deployment as of publication. The combination of live transaction data, government subsidy integration, and cross-border coordination through Project Agorá means the next 12 months will produce an unusually detailed operational record for a two-tier digital currency architecture, one that few if any central banks have yet matched in scope or ambition.