👛 Wallets and super-apps around the world

Pass-through wallets and stored-value wallets

A payment wallet is an app that shows payers their payment instruments and initiates the transaction on an underlying rail. That rail is a card, an instant credit transfer, or an issuance of e-money. The app is an interface layer: the payment method is still the rail it triggers. Two families split the global market. The first never holds the payer's funds; the second does. Whether it holds them determines the cost, the governing law, who bears the risk, and how long the merchant waits to be paid.

The Staged Digital Wallet Operator is a contractual category Visa created. It covers an operator that funds a balance by card, then pays the merchant out of that balance. The card networks formalized the distinction once this architecture started costing them revenue. The payment splits into two transactions: the load is a card transaction in the wallet's name, and the final payment is not. Visa's rules shut these operators out of its most favorable interchange rates and charge them dedicated fees. For a merchant accepting a stored-value wallet, interchange drops out of the cost structure and is replaced by a single fee negotiated with the operator.

Wallets built on an account-to-account rail form a third family. They won out wherever an instant payment system existed before the wallet. TWINT in Switzerland, Swish in Sweden, BLIK in Poland, Bizum in Spain, MB WAY in Portugal, and Wero in the euro area all follow this model. The app initiates a credit transfer without storing value, and the balance stays on the bank's books. This architecture eliminates interchange and counterparty risk. It also eliminates the chargeback: once a credit transfer is executed, the buyer has no dispute process to turn to.

Apple Pay and Google Wallet as tokenization layers

Apple Pay (Apple Inc., 2014) and Google Wallet (Google LLC, 2011, after two rebrands) are presentation layers on top of the card, not payment methods. The wallet shows the instrument to the payer and sends substitute data at the time of payment. At enrollment, the network issues a DPAN (Device Primary Account Number), which replaces the real card number on the device. The original PAN never leaves the issuer. Each payment generates a transaction-specific cryptogram, so the DPAN is useless anywhere else.

PHYSICAL CARD (contactless)
  DE2  PAN            : 4970 10## #### 1234    cardholder's real number
  DE22 entry mode     : 07                     EMV contactless
  DE55 EMV data       : ARQC generated by the chip

PASS-THROUGH WALLET (Apple Pay / Google Wallet)
  DE2  PAN            : 5301 88## #### 7742    DPAN - device token
  DE22 entry mode     : 07                     EMV contactless (same)
  DE55 EMV data       : ARQC generated by the Secure Element / HCE
  token field         : wallet token requestor ID + ID&V assurance level

  -> the issuer retrieves the real PAN through the Token Service Provider
  -> interchange, scheme fees, and chargebacks are UNCHANGED
  -> a DPAN stolen off the device cannot be replayed: the cryptogram is missing

Apple Pay makes its money by billing card issuers, not merchants. The trade press has long reported a fee of around 0.15% on US credit transactions and a few cents on debit, though Apple has never confirmed it publicly. Google Wallet does not charge issuers this way. That difference explains why European banks have so little bargaining power: they pay to be in a wallet they do not control.

The US banking industry produced two responses: Zelle and Paze. The standalone Zelle app shut down on April 1, 2025, because fewer than 2% of users went through it (Early Warning Services, 2025). The service now lives only inside the banks' own apps. Paze, launched in 2023 by the same operator, claims more than 200 million eligible cards across nine issuers in 2026, yet merchant adoption remains slow, as the operator itself admits. The number of pre-provisioned cards measures eligibility; the share of merchants accepting the service measures how widely it is actually used.

Alipay and WeChat Pay: from ecosystems to regulated infrastructure

Alipay (支付宝, Ant Group, 2004) and WeChat Pay (微信支付, operated under Tencent's Tenpay 财付通 license, 2005) are full-fledged payment accounts, with credit, savings, insurance, and merchant mini-programs. Rather than sitting on top of a card as a presentation layer, both apps hold their users' funds. An OECD competition background note from June 2025 puts their shares of China's mobile payment market at about 54% and 42%. Other operators split what is left. No other market is so concentrated in the hands of two private companies.

China's regulatory crackdown on payments has followed three tracks, each with concrete effects on integration terms. The first, in place since June 2018, requires online payments by non-bank institutions to go through NetsUnion Clearing Corporation (网联清算有限公司, founded in 2017) rather than direct connections to banks. Super-app flows became visible to the central bank and subject to its rules. The second track is the regulation on the supervision of non-bank payment institutions, in force since May 1, 2024. Its implementing rules tie customer reserve funds to the operator's net assets. Since 2025, Alipay, Tenpay, and NetsUnion have been under the central bank's direct AML/CFT supervision.

The third track is QR code interoperability, which the central bank imposed on the three largest operators. For a decade, a Chinese merchant displayed two QR codes, each readable only by its own app, and signed two contracts. Starting in 2021, on orders from the PBoC, Ant, Tencent, and China UnionPay made their offline barcodes scannable by rival apps. In parallel, UnionPay is building its own wallet, Cloud QuickPass (云闪付, 2015), backed by the banks and by public-sector purchasing. The Alipay–WeChat Pay duopoly is intact, but Cloud QuickPass gives the state a payment channel it controls.

Embedded checkout credit means signing up for a financing product during the purchase, without leaving the payment flow. It was a hallmark of the Chinese model, and it is being shut down. Huabei (花呗, Chongqing Ant Consumer Finance, 2014) has one of the world's largest installment loan books, and users currently sign up with a single tap at checkout. From September 30, 2026, consumer credit products of this kind can no longer be offered within the payment flow. Users will have to sign up separately, outside the payment page. Conversion rates measured in China before that date therefore reflect a checkout flow the new rules will no longer allow.

Hong Kong and Macao are separate jurisdictions from mainland China, with their own currencies, regulators, and licenses. WeChat Pay HK (TenPay Payment Technology (HK), 2016) and AlipayHK settle in Hong Kong dollars under a local stored value facility license and connect to FPS, the instant payment rail. Macau Pass / MPay (澳門通, 1999) dominates contactless prepaid in Macao. BoC Pay (Bank of China Hong Kong, 2018) is designed for cross-border use in the Greater Bay Area, through UnionPay QR and HKD/CNY conversion.

Southeast Asia: private super-apps vs. public rails

A super-app bundles several consumer services around a single payment account: ride-hailing, delivery, shopping, and credit. After China, Southeast Asia has produced the most advanced examples, but on a different timeline. In China, the wallet came before the public rail. In Southeast Asia, the two were built side by side, and central banks set their standards while Grab, GoTo, and Sea were raising money. A merchant in the region therefore works with two layers stacked on top of each other: a mandatory national QR code, and competing wallets that customers use to scan it.

Beneath these apps lies a layer of public infrastructure: national QR standards and instant payment rails. QRIS (Bank Indonesia with ASPI, 2019) imposes a single QR standard on every Indonesian operator. By 2024 it already had 50.50 million users and 32.71 million enrolled merchants, and transactions grew 226.5% year over year (Bank Indonesia). DuitNow QR (PayNet, 2019) plays the same role in Malaysia, running on the DuitNow rail, which carried 4.5 billion transactions worth US$330 billion in 2025 (PayNet). PromptPay (National ITMX under a Bank of Thailand mandate, 2017) is the most heavily used rail in ASEAN, with 27.4 billion transactions in 2025. PayNow (Association of Banks in Singapore, 2017) handles more than 45% of Singapore's account-to-account transfers. InstaPay fills that role in the Philippines, under three-tier governance: the BSP sets the framework, the PPMI association governs, and private switches operate the rail.

The MDR, or merchant discount rate, is the percentage the merchant pays on each transaction. In Indonesia, it is set by a regulated schedule rather than negotiated between the operator and the merchant. Since March 15, 2025, Bank Indonesia has set the QRIS MDR by merchant category. Micro-businesses pay 0% on transactions up to IDR 500,000 and 0.3% above that, while small, medium-sized, and large businesses pay 0.7%. The rate drops to 0.6% for education, 0.4% for gas stations, and 0% for public services. The merchant bears the cost and cannot pass it on to the consumer.

Japan, Korea, India: apps and rails compete for the value

Japan's code payment market consolidated after the world's most expensive QR war, which produced a clear winner. PayPay (PayPay Corporation, SoftBank/LY Corporation group, 2018) claimed 70 million registered users as of July 15, 2025, about 55% of the population. Industry analyses from 2025 give it nearly two-thirds of Japan's code payment market. The remaining players are tied to telecom carriers or loyalty point ecosystems: au PAY (KDDI, 2019), d払い (NTT Docomo, 2018), Rakuten Pay (2016), and Merpay (Mercari group, 2019). None comes close in coverage, and a Japanese merchant that accepts only one code accepts PayPay.

In Korea, mobile payment adoption was shaped by the terminal base, which for years was built for the magnetic stripe. Samsung Pay (2015) made Magnetic Secure Transmission mainstream there: it emulated the stripe and worked without NFC. When the feature was dropped from recent models, the country switched to contactless EMV. The leading wallets are tied to digital service platforms. Kakao Pay (Kakao Pay Corp., 2014) is built into the KakaoTalk messaging app. Naver Pay (Naver Financial, 2015) runs checkout for the country's leading web portal and its marketplace. Toss (Viva Republica, 2015) started with free peer-to-peer transfers and grew into a bank and a PSP. These operators are supervised by the Financial Supervisory Service, which fined Kakao Pay 15 billion won in April 2025 for transferring the data of 40 million users to Alipay Singapore Holdings without consent.

India is the case of a free public rail whose usage has been captured by private apps. UPI is run by the NPCI, and wallets connect to it as Third-Party Application Providers. PhonePe processed 10,483.73 million UPI transactions in June 2026 out of a total of 22,716.07 million, about 46% of volume (NPCI). Together with Google Pay, the two apps account for close to 79%. Paytm (One97 Communications, 2010) has built its acceptance footprint on QR codes, the Soundbox, and Android terminals. BHIM, the government's app, has never gained a meaningful share. The NPCI runs the rail but does not own the relationship with end users, which belongs to the third-party apps.

UPI has two extensions, designed for technical constraints the main rail does not handle. UPI Lite and UPI Lite X (NPCI, 2022) store a balance in the app, debited without a PIN below a low limit, to take micropayments off bank servers. Lite X adds offline payments over NFC. Both variants address the per-transaction cost of a free rail. UPI 123PAY (2022) brings payments to feature phones, through interactive voice response, missed calls, or sound waves. The NPCI caps any single TPAP at 30% of volume. The compliance deadline has been pushed back to December 31, 2026.

Latin America and Africa: the wallet as a substitute for a bank account

In Latin America and Africa, the wallet did not replace the card: it replaced the bank account, for people who do not have one. It is where they keep money, how they send it, and how they get paid. Mercado Pago (MercadoLibre, 2004) is at once a wallet, an acquirer, an issuer, a lender, and a marketplace, operating in Argentina, Brazil, Mexico, Chile, Colombia, Peru, and Uruguay. In Argentina, it serves as the de facto payment infrastructure for the informal economy. Playing all these roles makes it a private, cross-border scheme, and integrating it involves far more than adding a checkout button.

Latin American banks have built their own wallets, with mixed results from country to country. In Argentina, MODO (Play Digital S.A., 2020) brings together accounts from more than 30 banks in a single app, running on Transferencias 3.0. It is one of the few bank-consortium wallets to have gained a foothold in Latin America. In Peru, Yape (Banco de Crédito del Perú, 2017) and Plin (a consortium of BBVA, Interbank, Scotiabank, and BanBif, 2020) formed a duopoly that the BCRP, Peru's central bank, forced to interoperate in 2023. It is the best-documented case of wallet market regulation in the region. In Colombia, Nequi (spun off from Bancolombia, and since granted its own banking license) and Daviplata (Banco Davivienda, 2011) built near-closed ecosystems, which became the central argument for creating the Bre-B public rail. In Panama, Yappy (Banco General, 2019) became the de facto national standard because there is no public retail payment rail.

Mobile money is a payment account service built on an e-money license and distributed through a network of local agents. It is the dominant rail in sub-Saharan Africa, and it is neither a bank rail nor a public one. M-PESA (Safaricom, 2007) is still the global benchmark for the model. MTN MoMo is the largest network on the continent by active users, in more than 13 markets. Its revenue mix has shifted: advanced services (credit, insurance, investments, merchant payments) made up 34.1% of MoMo revenue in 2025, up from 29.9% in 2024 (MTN Group, annual results published March 16, 2026). Person-to-person transfers are no longer the growth engine.

Europe: wallets built on the bank account, not the card

European wallets were built on domestic credit transfer rails rather than on cards. Europe has produced no super-app, because its transfer rails were already cheap, fast, and universal before wallets arrived. The European wallets that won out initiate a credit transfer or trigger a direct debit, without holding a balance at the operator. Merchants therefore face a cost structure that has nothing to do with interchange: they negotiate the price with a national operator rather than an international network.

The history of Europe's national wallets shows two patterns, one about fragmentation and one about consolidation. When several wallets compete in one market, the network effect takes longer to kick in, because each app signs up only a fraction of payers and merchants. Finland long had three competing wallets and still lags well behind the other Nordic countries, with about 2.8 million MobilePay users (Vipps MobilePay 2024 annual report). In Switzerland, by contrast, the 2017 merger of Paymit and TWINT created critical mass. Vipps and MobilePay did the same in 2022. In the European markets studied, only one national wallet holds the top spot over time.

How a wallet makes money

A wallet's business model relies on several revenue streams, and no wallet survives on payments alone: the operators that tried have disappeared. Payments bring frequent usage and purchase data, which the operator then monetizes through other products. Six revenue streams combine, in proportions that shift as the operator matures. Newer wallets earn most of their revenue from the merchant discount rate. Established operators earn most of their margin from credit and adjacent financial services.

User acquisition costs have shaped several wallet markets. PayPay took the lead in Japan with a massive cashback campaign funded by SoftBank, combined with an extended fee waiver for small merchants. Incumbents fought back, at great expense. In 2025, PayPay, Rakuten Pay, au PAY, and d払い waged a cashback war estimated at about $1 billion in rewards. Origami Pay did not survive it. A newcomer that buys adoption with cashback therefore pays for its own campaign and for the bidding war that incumbents launch in response.

Accepting wallet payments: what to integrate and what breaks

Integrating a wallet has a technical side and a contractual side, and the contractual side carries most of the cost and risk. It covers the signing entity, the payout delay, refund handling, reconciliation, and what happens to a disputed transaction. Seven questions cover most of this exposure, and they should be asked before signing, not after the first dispute.

• Which legal entity signs, and under what license? In Singapore, GrabPay contracts are signed by Grablink Pte. Ltd. under a MAS license; other markets are handled by other entities. Check the regulator's register, never the product page.
• Does the wallet hold the funds? If it does, it becomes the merchant's counterparty, and its insolvency hits the merchant directly. GCash is not a bank: balances are not covered by Philippine deposit insurance.
• Which rail actually settles? The same logo can mean an irrevocable credit transfer in one country and a disputable card payment in another. MobilePay runs account-to-account in stores in Denmark, but on cards for e-commerce.
• Who sets the fee? The operator or the regulator. In Indonesia, it is Bank Indonesia; in India, the fee is zero by law, and a bill now under consideration could change that.
• How long is the payout delay, and in what currency? This is a treasury item and a currency risk, not a technical detail.
• How is a transaction refunded? An irrevocable credit transfer cannot be refunded: it is offset by a transfer in the opposite direction, with its own fees and its own risk of failure.
• What is the dispute process, and who decides? A stored-value wallet applies its own terms and conditions, not scheme rules. The timelines, the burden of proof, and the outcome are completely different from a card chargeback.

Before announcing regional coverage, list the signing entities, their regulators, and their license numbers, one line per country. The exercise takes a day and almost always shows that a single brand hides five contracts, three licensing regimes, and two payout models. That list reflects the company's real exposure; the number of logos on the checkout page says nothing about it.