The Most Used Stablecoins—and How to Measure Adoption

USDT and USDC lead an identifiable sample of stablecoin payment activity, but a useful answer to “most used” needs a metric and a date.¹ A token can lead in balances held, another in transfer value, and another within a particular application. Those results can coexist. For a builder, the useful comparison is the one that resembles the customer behavior the product needs to support.

A dated leaderboard for payment activity

Artemis’s October 2025 update provides a concrete comparison. Its payment survey reported $10.2 billion of monthly payments in August 2025. The accompanying token-share slide put USDT first at 79% and USDC second at 21%, with USDC up from 14% in February.¹

These are rounded shares from the report’s payment survey, whose update involved more than 30 partners. They describe the surveyed activity, rather than a census of every payment worldwide. The study’s monthly series and its published update dates also differ: August is the observation month; October is the report edition.¹²

This article uses that dated comparison because it can be traced to a published report. It does not present those percentages as September 2026 market shares. Nor does the absence of other tokens from this table establish that nobody uses them. A provider sample can reflect the corridors, customers, and infrastructure its participants serve.

Separate the questions hidden inside adoption

An adoption dashboard should distinguish balances, movement, participation, and completed commercial activity. Each answers a different question.

A treasury product might care about persistent balances. A trading application might prioritize executable liquidity and trading activity. A payroll service needs evidence about recipients, payout completion, and withdrawal options. Calling all three “adoption” conceals those differences.

Transaction count and value should also travel together. Ten thousand small transfers can exceed one large transfer by count while moving less money. Average size helps, but a median and a size distribution reveal whether a few unusually large transfers dominate the total.

Read the methodology before comparing volume

Visa’s onchain dashboard separates adjusted and unadjusted activity. Its published methodology includes a filter that counts the largest stablecoin amount within a transaction, addressing repeated internal movements. It also uses address labels and behavioral thresholds to identify activity such as automated routing. Its “retail sized” category applies a threshold below $250 to adjusted transactions; that size label alone does not prove a purchase occurred.³

Methodologies change. On September 18, 2026, Visa disclosed a refresh that expanded its address-label coverage and excluded more automated activity. Adjusted volume declined after the refresh, while adjusted transaction count decreased by less than 2%.⁴ Comparing a screenshot captured before that revision with a later export can therefore mix measurement versions.

Artemis made a different change in July 2026: its Terminal and API transfer-volume metric moved from an adjusted calculation to gross transfers, excluding mints and burns. Field names stayed the same. Its older Artemis-filtered and P2P metric flavors were being retired.⁵

For an integration, preserve the methodology version beside the value. A familiar API field name is insufficient evidence that two observations remain comparable. Investigate a discontinuity before describing it as customer growth.

Wallet counts need their own definitions

Addresses are observable; people usually require additional evidence. One person can control several addresses, while a custodial platform can represent many customers through a smaller set. An address that once received tokens also need not represent an active customer today.

Issuer definitions can narrow the question. Circle’s 2026 Internet Financial System Report defines “meaningful wallets” as onchain wallets holding more than $10 in USDC at period end.⁶ That measures a balance condition. It does not establish that every qualifying wallet transacted during the period, or that each wallet belongs to a different person.

A useful product dashboard might show funded addresses, active senders, active recipients, and returning customers separately. For customer retention, use the product’s own account records where available and permitted. For market comparisons, keep the label “addresses” unless the research explains how it resolved addresses into users. Changing the label from addresses to people can create a large apparent audience without new evidence.

Compare the token and the network together

“We support USDC” is incomplete as an operational requirement. Record the blockchain and token deployment, then establish whether the asset is native, bridged, or another representation. Allium’s stablecoin data model makes that distinction explicit through deployment and product registries; its documentation describes how to select a single canonical product or include related variants.⁷

For market analysis, decide that grouping policy before calculating shares. Otherwise, one provider’s aggregated token family may be compared against another provider’s native-token series. Networks missing from a dataset can also change the apparent ranking.

For a payment product, compare the combinations customers can actually receive and use. A globally prominent token can still be inconvenient if the recipient’s wallet or withdrawal provider does not support the chosen network. Product availability, settlement experience, and conversion costs should be checked at that level. A high aggregate transfer total cannot answer those implementation questions.

Connect blockchain movement to a payment outcome

Consider a hypothetical $1,000 supplier payment. A platform receives the funds, moves them through an operational wallet, and forwards them to an off-ramp. Three transfers of $1,000 produce $3,000 of gross movement while paying one $1,000 invoice. Additional trades or cross-chain operations can make the accounting more involved. This illustration is why deduplication and business-event reconciliation matter.

Card spending introduces another distinction. Artemis’s January 2026 card study describes arrangements in which a customer spends a crypto or stablecoin balance but the merchant receives ordinary fiat through card infrastructure.⁸ That is meaningful adoption, yet it is different from a merchant accepting the token directly.

Issuer reporting also requires careful labels. Circle reported $14.8 trillion of USDC onchain transaction volume for Q2 2026. It separately reported a $14.7 billion annualized transaction-volume rate for Circle Payments Network, based on the trailing 30 days at quarter end.⁹ Those figures cover different systems and windows. Neither converts the broader onchain total into a measured merchant-sales figure.

Build a comparison that can guide a decision

Start by documenting the measurement rules: the business question, time interval, included tokens, networks, unit, and exclusions. Save the source and retrieval date. If a metric is annualized, retain the underlying observation window; a run rate is different from an amount actually processed over a full year.

Keep a copy of each published comparison and record changes to its coverage. If you add a network halfway through a series, calculate an overlapping period under both definitions. That simple comparison helps distinguish a broader measurement boundary from a change in the activity customers actually generated. Apply the same discipline when a provider revises historical data.

Then compare like with like. Keep trading, transfers, attributed payments, and customer balances in separate series. Reconcile your own payment records to blockchain transaction identifiers so that retries, treasury movements, refunds, and completed invoices receive appropriate treatment.

Finally, evaluate repeat behavior and outcomes. Did recipients successfully receive the expected asset? Could they use or withdraw it? Did they return? USDT’s lead in the cited payments sample and USDC’s substantial participation provide a starting point. The evidence that determines a product decision comes from matching those broad signals to a specific customer, network, and completed financial task.

Sources