Why Korea Needs a Won Stablecoin in an Age of AI-Driven Consumers

Kim Min-seung’s ₿-ficial

In 1811, textile workers in Nottingham, England, stormed factories with hammers, convinced newly introduced knitting machines would take their jobs. The Luddite movement named after them ultimately failed. More machines replaced the broken ones, and clothes poured out in larger volumes at lower prices. But the deeper problem the movement foreshadowed came later: If machines make the clothes, who buys them?

A century later, Henry Ford offered one answer. In January 1914, Ford Motor more than doubled its workers’ daily pay to $5 from $2.34 overnight, roughly twice the industry average. The purpose of the unusual pay raise was clear. Ford wanted to create, inside the factory itself, the consumers who could buy the cars rolling off the assembly line. He understood that if production was to be automated, consumption had to be organized as well.

Now AI has stepped into the place once occupied by those workers. Since the pandemic, job insecurity has resurfaced year after year. Since the arrival of generative AI, many white-collar workers have been haunted by the sense that AI may be better than they are. That is not merely exaggerated fear. It reflects a growing recognition that companies may ultimately choose AI over humans. That brings back Ford’s question: Where does the economy go when consumers begin to disappear?

If AI replaces human labor, production does not stop. Efficiency may rise further. The problem is not production but consumption. Factories will keep running. The question is who will spend the money.

There are two answers. One is to turn displaced workers back into consumers, as Ford did. The other is to create a new actor that can execute consumption in place of disappearing human consumers. Silicon Valley appears to be choosing the latter. If Ford cast workers as consumers 100 years ago, Big Tech is now building machines that press the payment button in place of humans. That is the agent economy.

In that economy, humans provide only intent. Send flowers to my parents this weekend. Book a hotel for next month’s trip to Jeju. Search, review analysis, price comparison and payment execution are all handled by agents. The thousands of requests agents exchange among themselves are machine-to-machine B2B traffic. At the end of that chain, when an agent opens a person’s account and sends payment to the florist, that becomes C2B traffic. Even if a laid-off office worker loses purchasing power, the agent he used can still draw the flower payment from the last remaining balance and send it to the shop. The consumer does not disappear. The executor of consumption changes.

But carrying out a single expression of intent requires dozens or even hundreds of small payments. Consider an agent planning a trip to Jeju. To book a flight, it must pull seat data from multiple airlines. To choose a hotel, it must query price information scattered across platforms. To find restaurants, it must call review APIs. Each is an information transaction worth only a few cents. What looks to a human like a single Google search is broken up in the agent world into dozens of paid calls. Such transactions are difficult to support on card networks because the fee on each payment can exceed the payment itself. That is where stablecoins come in.

Those payment flows have already begun. In March, Circle, the issuer of the USDC dollar stablecoin, disclosed payment data between AI agents. More than 400,000 agents took part, and the average payment was $0.31. Marketing jargon is already shifting as well, from SEO, or search engine optimization, to AEO, answer engine optimization, and GEO, generative engine optimization. It is a sign that the party typing into the search box is shifting from humans to agents.

Payment infrastructure is moving even faster. Earlier this year, Google announced the Universal Commerce Protocol, or UCP, an agent-payment standard, with Walmart and Shopify. Around the same time, PayPal acquired an agent-commerce company. In March, Mastercard bought stablecoin payment infrastructure firm BVNK for as much as $1.8 billion. Coinbase launched Agentic Wallets, a wallet designed for agents. Circle unveiled nanopayments built to make even 1 cent, and one-hundredth of that, economically viable. This is less like a shopping cart at a hypermarket and more like the coin slot on a vending machine. Mastercard Chief Product Officer Jorn Lambert summed up the shift in one sentence: “There are no more problems left to solve in the card business.” Visa and Mastercard know better than the rest of us what their businesses may look like in 10 years.

Of course, this is not a world people will feel tomorrow. Coinbase’s x402 payment protocol handles only about $30,000 in daily real transaction volume, and most of that is test traffic. The agent economy still sounds less like street noise than the roar of a laboratory. But Google, PayPal, Visa, Mastercard, Stripe, Circle and Coinbase have all entered that lab at the same time. When that roar spills into the street is another question. Technology always follows an S-curve, and the inflection point is visible only in hindsight.

One of the central figures in that laboratory came to South Korea on April 13: Circle founder Jeremy Allaire. During his two-day visit, he held luncheon meetings with executives from KB Financial Group, Shinhan Financial Group and Hana Financial Group. He also signed separate partnerships with Upbit, Bithumb and Coinone. His message at a press briefing was clear. Stablecoins are no longer a niche within crypto. They are global payment infrastructure, and the market is moving ahead of regulation.

One notable point was his explicit statement that he would not issue a won stablecoin directly in South Korea. Instead, he laid out a different structure. Domestic banks, fintech firms and digital-asset companies would be the issuers, while Circle would provide the underlying technical standards. In other words, South Korea would retain sovereignty over issuance while Circle laid the rails beneath it. The structure resembles Mastercard’s move to add BVNK’s payment network on top of its existing card business. Big Tech competition is not taking place at the service layer. It is happening in the rails underneath. And which currency will run on those rails is still undecided.

South Korea’s domestic payments landscape is already shifting. According to the Bank of Korea’s “2025 Payment and Settlement Trends in Korea” report, card payments made through mobile devices accounted for 54.3% of the total. The sight of consumers pulling plastic cards from their wallets may disappear before another generation passes. But mobile payments still ride on card networks. The average transaction on a South Korean personal credit card runs to tens of thousands of won, while the average payment in the agent economy is $0.31, or about 450 won. The two worlds differ by three orders of magnitude. They cannot meet on the same rail. No matter how much Naver Pay and Kakao Pay improve, it will be difficult to put machine-to-machine payments worth about 32 cents each onto a card network.

Domestic debate over a won stablecoin often still stops at the level of “a cup of coffee.” Why do we need stablecoins when cash and cards already work fine for payments? But that may itself be a 20th-century question. It tries to explain the age of AI agents through the framework of an era when people bought vegetables and fish with cash in traditional markets. Agents in this era exchange API calls worth a few cents dozens of times per second.

Dollar dominance is already a reality. Until now, though, the dollar mostly operated when transactions crossed borders. Koreans were paid in won, saved in won and bought flowers in won. What is different about the agent economy is that the dollar can penetrate even the capillaries of domestic transactions. Ordering flowers on Naver or calling a taxi on Kakao could shift underneath to a structure in which agents settle in dollars. That Koreans could effectively enter the agent economy with dollars in their digital wallets from birth is the real erosion of monetary sovereignty. It is a far quieter and more fundamental form of erosion than defending the exchange rate in the foreign-exchange market.

In an age when consumers are being fired, the new consumer may arrive holding dollars. Ford answered this 100 years ago: Without consumers, there is no production. A won stablecoin is not about buying a cup of coffee. It is about which country’s money the consumer in the agent economy will be born holding. Before that wallet is filled with dollars, Korea needs to lay won-denominated rails first.

About Kim Min-seung, head of research at Korbit...

He is a founding member and head of the Korbit Research Center. He works to explain complex events and concepts in blockchain and the digital-asset ecosystem in accessible terms and to help people with differing viewpoints understand one another. His background includes blockchain project strategy and software development.

This column was contributed by an outside writer to offer cryptocurrency newsletter subscribers a range of perspectives and does not represent the views of the Korea Economic Daily.

Cho Mi-hyun, Korea Economic Daily reporter mwise@hankyung.com