From Printing Money to Building Roads: The Stablecoin War Enters the Era of Interface Competition

There are two types of businesses that are the most profitable: one is printing money, and the other is building roads. In the past, everyone thought printing money was the best, but by October 2026, they realized that those who print money have already paid tolls to those who build roads.

Written by: Fu Gui

October 6

On October 6, 2026, three significant events occurred in the stablecoin space.

On that day, Arbitrum One officially launched USDG natively, joining the Global Dollar Network led by Paxos. ArbitrumDAO simultaneously proposed a governance proposal to add another 100 million ARB to the USDG incentive program, which is currently in the voting stage. The stablecoin scale on the Arbitrum chain is approximately $3.78 billion, and the interest generated from this money lying in wallets every year has previously gone entirely into the pockets of Circle and Tether. The official selling point of the Global Dollar Network is straightforward—"Stop funding their profits," meaning no longer support others' profits. This step by Arbitrum is not just about supporting another stablecoin; it is the first attempt to turn on-chain stablecoin deposits into its own source of income.

On the same day, OKX announced the completion of a new round of strategic financing, with a pre-investment valuation of $25 billion. The list of investors includes Circle, QRT, Ripple, and SC Ventures, a subsidiary of Standard Chartered Bank. OKX stated that it is not short of money but is looking for long-term partners. This translates to the issuer actively spending money to buy tickets for channels, with three issuers coming together to support one exchange, fearing that their tokens will be placed at the back of the order book.

Also during this time window, Henri Stern, who manages stablecoins at Stripe, announced that by the end of the year, they aim to roll out stablecoin cards in over 100 countries, allowing users to spend their stablecoins directly at Visa's 175 million merchants. This card processed $1.2 billion last month, tripling year-on-year. Stripe does not issue stablecoins; it only facilitates the delivery of stablecoins to places where they can be spent.

Chains, exchanges, and payment companies are all reaching out in the same direction on the same day. They are all talking about one thing: the rules of the stablecoin world have changed.

The Most Profitable Issuer

Tether is the most profitable stablecoin issuer in the crypto space, and this statement hardly needs qualification.

In 2025, Tether's net profit exceeded $10 billion, with about 300 employees, resulting in an average profit of about $33 million per employee, nearly a hundred times that of Goldman Sachs. The logic behind its profitability is quite simple: users mint 1 USDT with 1 dollar, and Tether uses that 1 dollar to buy U.S. Treasury bonds, earning about 5% interest per year, while the token holders receive no interest at all; the interest margin goes entirely to Tether. By the end of 2025, Tether directly held over $122 billion in U.S. Treasury bonds, and with repurchases and indirect exposure through money market funds, its total U.S. Treasury position was about $141.6 billion. If Tether were treated as a country, it would be close to the 18th largest holder of U.S. debt globally. It also hoarded 127.5 tons of gold and 96,185 bitcoins, with year-end market values of approximately $17 billion in gold and about $8.4 billion in bitcoin, with an unrealized profit of about $3.5 billion from bitcoin.

According to protocol revenue tracked by DeFiLlama, Tether's 30-day revenue is about $510 million, ranking first; Circle ranks second with about $210 million, less than half of Tether's; and further down, Hyperliquid and Pump.fun are less than one-eighth of Tether's.

Tether earns so much not because it mints USDT well, but because it distributes USDT widely. In convenience stores in Central America, rural markets in Africa, and phone top-up points in Southeast Asia, you won't find Coinbase or Circle's compliant channels, but you will find people using USDT. Research in 2026 showed that about 93% of USDT on Tron exists in ordinary wallets, not on exchanges; the Andersen Institute even estimated that about 95% is outside centralized exchanges. Its advantage lies not in the four words "decentralized," but in highly dispersed distribution; users, scenarios, and liquidity are not locked in a single super entrance. No single channel can choke it off, and thus no one can take away its profits.

The Other Side of Profit

Circle is also profitable, with total revenue of $2.747 billion in fiscal year 2025, a year-on-year increase of 64%, but it is becoming increasingly difficult to earn.

The difficulty does not stem from earning less, but from receiving less of what it earns. Circle's core profitability metric is called RLDC Margin, which is the proportion of total revenue remaining after deducting distribution costs, approximately 39% for the entire year of 2025. This doesn't sound low, but it means that over 60% of revenue must be shared with the channels that help distribute its tokens. In 2025, Circle's distribution, trading, and other costs reached $1.664 billion, a year-on-year increase of about 64%; among these, the distribution cost paid to Coinbase alone was about $1.4 billion, a year-on-year increase of about 51%. The biggest bleeding point is Coinbase—according to the agreement between the two parties, the reserve interest generated by USDC on the Coinbase platform goes mostly to Coinbase, leaving only a small portion for Circle, and the channel's asking price rises year after year.

Even more awkward is the interest rate. Over 90% of Circle's revenue comes from reserve interest, and every time the Federal Reserve lowers rates, its revenue drops. The data from the second quarter of 2026 illustrates the issue best: the on-chain transaction volume of USDC surged 151% year-on-year, but reserve income only grew by 5%. The reason is clear: the average circulation only increased by 25%, while the reserve yield dropped by 66 basis points. The most noteworthy aspect is not the 151% surge in transaction volume, but that transaction volume and reserve income have begun to decouple. Circle earns based on "how many dollars are long-term parked in the system," not "how many times these dollars move in a year." Tokens circulating among users does not mean that more money is accumulating in Circle's accounts to earn interest.

There is also the long-term pressure of interest rate cuts. In 2025, the Federal Reserve entered a rate-cutting cycle, and Circle's adjusted EBITDA for the year was $582 million, but due to $424 million in equity incentive expenses related to its IPO, it reported a net loss of $70.1 million. The capital markets are more concerned about competition: on September 1, 21 of the world's largest financial institutions, including Goldman Sachs, Citigroup, and Bank of America, announced plans to form a joint venture to issue their own stablecoins, causing Circle's stock to drop 6% that day. On June 30, Stripe, Visa, Mastercard, BlackRock, and Google came together to create OpenUSD, leading to a single-day drop of about 17% in Circle's stock; on August 3, Morgan Stanley cut its target price from $106 to $38, citing reasons including the slowdown in USDC growth, competitive pressure from OpenUSD, and the fragility of the reserve income model.

Circle's problem is not that it does not comply well or that its reserves are not transparent; it is that it relies too much on others to distribute its tokens. Once the channels raise their asking prices, it has no room to negotiate.

The Last Mile

Saying that channels are important is not an empty phrase. In the stablecoin space, there is a term called "the last mile," which means that a successful on-chain settlement does not equate to a successful commercial payment. Money can technically fly from Venezuela to Vietnam in one second on-chain; the difficulty lies in reliably converting that USDT into Vietnamese dong according to local regulatory requirements, depositing it into the correct bank account, and ensuring the accounts match.

This last mile involves four steps: cross-border transfer has been resolved, currency exchange is basically controllable, but the real bottleneck is the integration of local payment channels and compliance reconciliation. A service provider wanting to operate in ten markets must maintain dozens of independent banks and mobile payment integrations, with each interface potentially experiencing independent failures, and each country's KYC fields and reporting requirements being entirely different. This is just the issue of withdrawals. Deposits also have their troubles: users may pay in dollars but select the wrong chain, may transfer USDT while the merchant receives USDC, or may forget to fill in the memo. Binance disclosed in 2023 that it processes about 4,000 recharge retrieval requests each month, recovering a total of 7 million USDT; Kraken charges up to $200 per retrieval for wrongly sent tokens.

What is even more discouraging are the real numbers. After cleaning up $62 trillion in stablecoin transfers on public chains, BCG and Allium, after eliminating robot volume, internal transfers, and trading speculation, estimate that the actual economic activity is about $4.2 trillion, and only $350 billion to $550 billion can be identified as payments for goods and services, accounting for less than 2%. A survey by the European Central Bank showed that only 0.2% of online merchants in the Eurozone accept crypto assets, and the acceptance rate in physical retail has remained below 1% for two years. Mastercard's Chief Product Officer bluntly stated that about 90% of stablecoin transaction volume is related to cryptocurrency trading, and stablecoins "lack a clear value proposition" in regular personal-to-merchant payments. Field visits in Yiwu were even more straightforward: most merchants had never heard of stablecoins, and a few who had used them found that traditional settlements could receive 6% to 13% export tax rebates, while using USDT would instead cost them that money. What is solved on-chain is "how money runs," but it does not equate to "how goods are delivered."

A number of companies attempting to fix this last mile have already encountered problems. BitPay's card has stopped accepting new user applications since June 2023 and has not resumed in three years, with a Trustpilot rating of 1.2 out of 5, and customer service only available via email tickets, with weekend working hours reduced to 4 hours. From 2025 to 2026, Dupay ceased operations due to "compliance issues and fundamental obstacles to capital flow," OneKey U card was deactivated, Binance terminated card services in Europe and Latin America, and Kulipa in Paris went bankrupt overnight, affecting over 20 wallet clients including Solflare and Ready—its issuing bank Monavate was fined €270,000 by the Lithuanian central bank, and once its license was suspended, the entire product immediately died. Cambodia's Huayong Payment had its on-chain balance drained to only 990,000 USDT, with its license revoked a year earlier. In Poland, the regulator revoked Quicko's license in January 2026, and on February 3, three cards belonging to different brands, CEX.IO, Trustee Plus, and IN1, simultaneously became invalid.

These cases have different specific causes of death, but they expose the same structural weakness: front-end brands control users, while the real determinants of product survival are the back-end issuing banks, licenses, settlement, and compliance nodes. Most crypto card projects are not Principal Members of Visa; they rely on a BIN sponsorship model to "rent" a bank membership, where the sponsor handles compliance and settlement, while the project manages the brand and customers. The names of the sponsors are hardly visible in the cardholder agreements, but if the sponsor collapses, the card becomes useless. The collapse of Wirecard in 2020 took down the initial Visa cards of Crypto.com and Binance, while Metropolitan Bank's exit from the crypto space in 2023 crippled BitPay.

When it comes to Visa, there is a common misconception. Many people think that card organizations are just toll-collecting intermediaries, but that is not the case. What Visa truly sells is a complete set of road-building infrastructure: transaction authorization, network routing, rule systems, anti-fraud engines, inter-institutional clearing, chargeback dispute handling, and a global acceptance network. For a $100 transaction, the merchant pays a merchant discount to the acquiring institution, the issuing bank takes most of it through interchange, and the network fee collected by Visa is just a small fraction—but the premise for Visa taking that fraction is that it has built a trust system over fifty years that allows billions of cards and millions of merchants to transact safely every day.

Now, Visa is accelerating down this extended road. By September 2026, there will be over 160 stablecoin-linked card programs running on the Visa network, with an annualized run-rate for stablecoin settlements exceeding $20 billion, a year-on-year increase of over 15 times. The CFO did not list this figure separately during the earnings call because it is still just 0.12% of Visa's total annual volume of several trillion dollars. But the trend is clear: the crypto industry has not eliminated Visa; rather, it is paying tolls on the new stablecoin highway that Visa is building.

This last mile is fundamentally not a technical issue, but rather a systemic issue, an infrastructure issue, and a trust issue. The blockchain solves "distance," but "delivery" requires local banking relationships, foreign exchange liquidity, compliance capabilities, and dispute resolution—these are things that Visa and banks have built over fifty years, and the crypto industry has yet to find a low-cost replication path. Whoever can pave this last mile is the true road builder. It is precisely because this last mile is difficult to pave that those platforms that have already built roads to their territories dare to charge a price to the money printers.

The Resistance of Platforms

The one that truly broke through this layer of glass was Hyperliquid.

In September 2025, Hyperliquid did something no one had done before: it publicly tendered the issuance rights of its ecosystem stablecoin USDH. Companies like Paxos, Frax, Ethena, Sky, and Agora came to bid, and the last team called Native Markets won. At that time, Hyperliquid's monthly trading volume was close to $400 billion, with $5.7 billion in stablecoins on-chain, generating hundreds of millions of dollars in interest annually, all of which previously went to Circle and Tether. Dragonfly partner Haseeb Qureshi publicly stated that there were irregularities in the bidding, suggesting that Native Markets seemed to be pre-selected. But whether it was pre-selected or not is not important; what matters is that Hyperliquid made its stance clear: with so much money flowing through this platform every day, the qualification to issue coins is determined by the platform.

USDH operated for 8 months, peaking at a circulation of only $100 million, which is just a fraction compared to the over $5 billion of USDC on the Hyperliquid platform. Users did not recognize it; no matter how good the mechanism is, it is useless. On May 14, 2026, Hyperliquid, Circle, Coinbase, and Native Markets all announced simultaneously that USDH would no longer be issued, and USDC would once again become the platform's sole settlement asset. Native Markets transferred the purchasing rights of the USDH brand asset to Coinbase and exited.

It looks like Circle won, but a closer look at the terms reveals who really won. The new agreement, called AQAv2, has three core points: first, about 90% of the adjusted USDC reserve earnings flow to the Hyperliquid ecosystem, higher than the 50% promised by Native Markets during the USDH period; second, Circle and Coinbase each pledge 500,000 HYPE; if the earnings are insufficient to cover costs, the pledged coins can be forfeited; third, Coinbase is responsible for treasury deployment, while Circle handles technical deployment and cross-chain operations, with both parties contributing labor while the bulk of the money stays with the platform. Earnings began to accumulate on August 26, with the first payment of $14.58 million arriving on October 3. The key to this story is not that Hyperliquid failed to issue its own coin—it first created competition with USDH, and in the end, Circle and Coinbase, in order to remain in this distribution channel, not only pledged real HYPE but also relinquished 90% of the earnings. Hyperliquid did not win the war of "issuing its own stablecoin," but it won the war of "whoever issues must pay rent to the platform."

Polymarket took a different route; it does not negotiate profit-sharing with you but directly pockets the settlement layer. In April 2026, Polymarket upgraded, changing the platform collateral from USDC.e to its own pUSD. pUSD is backed 1:1 by USDC, and the smart contract enforces this, making it look like a wrapped token. However, this layer of wrapping has a different meaning. Users can deposit any type of coin, whether it's ETH, DAI, WBTC on Ethereum, USDT, WETH on Polygon, SOL, USDe on Solana, or even USDT on Bitcoin and Tron—over twenty types of tokens can be deposited, and the system automatically bridges and converts them into pUSD, settling within the platform using pUSD. Users still see dollars, but Polymarket gains control over the internal settlement layer, while the underlying reserves remain anchored to USDC. Polymarket earns interest on these reserve assets, bringing in at least $50 million a year. It did not replace the issuer but turned the issuer into an underlying asset, transforming its own scene's settlement layer into a true value capture layer.

Trading platforms and prediction markets are doing this, and even meme platforms are doing the same. Pump.fun's treasury has accumulated nearly $2 billion, and co-founder Noah Tweedale publicly stated in September 2026 that the team is discussing launching its own stablecoin, bluntly stating that decentralization is not important; who controls the end user is key. Pump.fun first switched its trading pairs from SOL to USDC to solve the income settlement issue, and the next step is to issue its own coin, keeping the reserve earnings entirely within the platform. Four.Meme on the BNB Chain did not wait to issue its own coin but directly designated UXUY's UUSD as the official launch token, embedding UUSD as the default quote asset in the launch process. Solana's LetsBonk collaborated with World Liberty Financial, related to the Trump family, where WLFI provided real money incentives, and developers were paid for pairing launch tokens with USD1. On January 3, 2026, LetsBonk's daily token creation surged to 8,800, with market share skyrocketing from 3% to 30%. Even those issuing memes understand that whoever embeds stablecoins into their scenes holds the faucet of money.

Three Generations of Evolution

Looking back, the stablecoin issue has quietly gone through three generations.

The first generation is called Issuer-owned, where the issuer owns everything. Tether and Circle are typical representatives, where the issuer simultaneously controls the brand, minting rights, reserve earnings, and channel relationships, with channels merely acting as delivery agents, earning a small fee. At that time, issuance was a privilege; whoever obtained compliance status could sit back and earn interest on U.S. Treasury bonds. Tether leveraged its first-mover advantage and low-cost transfers on Tron to saturate emerging markets, while Circle secured U.S. institutions through Coinbase and compliance status, each occupying their own territory, with channels having no say.

The second generation is called Issuer-as-Infrastructure, where the issuer devolves into infrastructure. Paxos is the most typical representative of this generation; it has issued BUSD for Binance, PYUSD for PayPal, and now USDG for Global Dollar Network, effectively becoming a stablecoin Foxconn. Paxos's current product line is simply divided into two: one is White Label Stablecoins, providing OEM services for brand owners; the other is GDN and USDG, actively sharing reserve earnings with network participants. The issuer is left with only licenses, custody, minting technology, compliance capabilities, and APIs; the brand belongs to others, the users belong to others, and by the USDG generation, even the dominant power of profit distribution has been relinquished. The issuer has devolved from a decision-maker to a compliance outsourcing vendor, doing whatever the channel says. Paxos is not unable to survive; it has realized one thing: licenses are becoming less valuable, while user entry points are what truly matter.

The third generation is called Platform-owned Settlement, where the platform controls the settlement layer. This is what is currently happening. OpenUSD has brought in over 140 institutions, with the underlying ledger and reserves being shared, where Visa, Stripe, and Shopify each manage their own scenarios, and revenues are split based on traffic, even equity is to be shared with partners contributing to network growth. PayPal is even more aggressive, creating a PYUSDx platform where any business can issue its own branded stablecoin without touching reserves or compliance; PayPal and MoonPay have turned issuance into a configurable service, simply requiring an API connection. At this point, the platform controls user entry, quoted assets, scene settlements, transaction routing, and reserve earnings sharing, while the stablecoin itself increasingly resembles a commodity.

But the evolution of the three generations is not simply a matter of "the issuer lost, the platform won." The issuers are also busy; they are vertically integrating both upstream and downstream in the value chain. Circle is promoting the Arc chain, aiming to control the infrastructure; Stripe is working on Bridge for stablecoin exchanges, acquiring Privy for wallets, and rolling out stablecoin cards, extending from payments to issuance and infrastructure; Visa has built a Stablecoin Platform, integrating stablecoin settlements into its global network; Tether is also expanding into payments, its own blockchain, wallets, and physical infrastructure. Everyone is building roads, and everyone fears being pushed back to the position of merely printing money like Foxconn.

The essence of the three generations of evolution can be summarized in one sentence: the threshold for printing money is getting lower, while the threshold for building roads is getting higher.

The Power of Interfaces

Now, looking at the market share of USDT and USDC, together they account for over 80%, but focusing on this number has become meaningless.

Because the issuance rights are depreciating while the distribution rights are appreciating. Cross-chain bridges and smart routing make exchanges between stablecoins nearly costless, making it increasingly unimportant whether users hold USDT, USDC, or pUSD. What matters is who can make users think of using their channel first when spending money, and who can integrate their settlement layer into transactions, payments, cross-border, and bank clearing scenarios. Tether's moat is not how many U.S. Treasury bonds it holds, but its presence in millions of wallets and offline points in emerging markets, which cannot be replaced by anyone; its distribution is scattered, wild, and lacks a single gate that can be closed, which is its fundamental difference from Circle. Circle is desperately investing in OKX, ceding 90% of profits at Hyperliquid, and spending its own money to build the Arc chain because it knows that merely having compliance and reserves is useless; if the coins don't reach the users, they are just worthless paper.

In the future, there will be no super stablecoin that dominates the market; there will be countless scenarios with their own settlement currencies. Exchanges will have their own, prediction markets will have theirs, meme platforms will have theirs, banks will have theirs, and card networks will have theirs. They may share the same reserves and compliance at the base level, but on the surface, each will follow its own channels, with profits distributed according to the traffic they bring. a16z once said that the real opportunity for stablecoins is not in disrupting Visa's 150 million merchants, but in serving those merchants that Visa's network has never covered—individual developers and small tool vendors without websites, legal entities, or credit histories, who cannot obtain credit card processing qualifications but can receive USDT with just a wallet address. Machine trading between AI agents is more direct; the x402 protocol embeds stablecoin payments into HTTP requests, making card processing fees just a "tax" that can be optimized away for AI.

These are the new roads.

This war is shifting from "who owns the dollars" to "who controls the interface for dollars entering the real economy." Those who print money are increasingly resembling suppliers of raw materials at the bottom level, while those who control the interfaces are the decision-makers. Whoever gets the coins into the users' hands, whoever truly paves the last mile, will be the rule-makers. Market share is a thing of the past; interface networks are the future.