Coinbase’s stablecoin report counts $891 billion in a quarter: what the filtered figure measures, and who paid for it

Coinbase published a stablecoin report on 20 October 2026 that puts “adjusted” stablecoin transfers at $891 billion in the third quarter, more than $400 million an hour and about four times the $216 billion of the first quarter of 2023. The study, titled “The Always-On Dollar”, was produced by The Block Pro Research and commissioned by Coinbase, according to CryptoSlate. It also says 73% of a separately screened transfer volume moved outside US banking hours, and that seven of the ten largest US banks have tested stablecoins or tokenized deposits for settlement.

Those headline numbers travelled fast on 20 and 21 October. What travelled less is how they were built. The $891 billion is a filtered figure, the 73% comes from another dataset and another period, and the word “adjusted” has no common definition: Visa, which uses the same data provider, published an adjusted figure for June alone that is twice Coinbase’s whole quarter. And Coinbase, which paid for the study, earns money every time more USDC sits on its platform or circulates elsewhere.

One note on sourcing. The report sits on Coinbase’s blog, but the page refused our requests on 21 October (HTTP 403). The figures below come from four outlets that read it and agree with each other: CryptoSlate, The Block, Crypto Briefing and Journal du Coin. Where they do not give a detail, we say so.

What the Coinbase stablecoin report says, figure by figure

The report’s thesis is in its title: dollars on a blockchain do not keep office hours, and businesses that hold them can move money at night and at weekends. The Block quotes the study as saying that, for businesses, this availability “can also make working capital more productive.” The figures behind that case, as the four outlets report them:

The hourly figure is easy to check. A quarter of 92 days has 2,208 hours, and $891 billion divided by 2,208 gives about $403 million, which matches the “over $400 million” in the coverage.

Neither the outlets nor the summaries we could read name the chains or the stablecoins counted, beyond a Base and USDC example. The Block mentions a chart by chain in the report. We could not see it.

How the $891 billion is built, and what it leaves out

According to CryptoSlate, the quarterly figure uses data from Allium, a blockchain data company, and removes several kinds of activity before counting: transfers involving exchanges, wallet.">decentralized finance, bridges and other infrastructure, automated transactions, and issuance and redemption (the minting and burning of tokens when dollars enter or leave). What remains is transfer activity between other addresses.

For a study about business payments, the filter makes sense. Raw stablecoin volume is inflated by trading bots, by liquidity pools passing the same dollars back and forth and by exchanges moving coins between their own wallets.

It still does not give a payments figure. A transfer between two wallets can be a salary, an invoice or a remittance. It can also be an arbitrage trader moving funds to a new venue, a margin deposit at a broker that is not tagged as an exchange, or a company moving its own treasury from one custodian to another. Journal du Coin put it in one line: a transfer volume is not a payment volume. CryptoSlate makes the same point, noting that the total measures transfers rather than verified merchant sales or distinct people.

None of the coverage we read gives the raw volume before filtering. Without it, readers cannot see how much was removed, or whether the share removed changed between 2023 and 2026. That matters for the growth claim: if the filters caught less automated activity in 2026 than in 2023, part of the fourfold rise could come from the method.

The 73% figure is a different measure

The after-hours share is the report’s most quoted line, and it does not describe the $891 billion. CryptoSlate explains that it rests on Dune Analytics data, covers the 12 months to 31 July 2026 rather than the third quarter, and uses a lighter filter: it removes transfers under $1, self-transfers and mints and burns, but keeps automated transactions. Bots do not sleep, so a measure that includes them will lean towards night and weekend hours almost by construction.

So nobody can write that 73% of $891 billion, about $650 billion, moved outside banking hours. The two numbers come from different data, different filters and different months. US banking hours are also a narrow window: a weekday afternoon in Asia or a Saturday in Europe counts as “outside” them.

Visa’s own adjusted June figure is twice Coinbase’s quarter

Visa runs its own stablecoin dashboard, Visa Onchain Analytics, which the site says was developed with Allium Labs, the same data source the report uses for its quarterly series. In early July the dashboard showed a record adjusted volume of $1.79 trillion for June 2026 alone, up 63% on May’s $1.1 trillion and 125% on a year earlier, Cointelegraph reported on 6 July. USDC accounted for about 67% of that June volume ($1.21 trillion) and USDT for about 32% ($576 billion). By chain, Base carried $565 billion, Ethereum $562 billion and Tron $320 billion.

Set side by side, one month of Visa’s “adjusted” volume is about twice a full quarter of Coinbase’s “adjusted” volume. On a monthly basis, Coinbase’s series averages about $297 billion, roughly a sixth of Visa’s June. Visa’s filter, as Cointelegraph describes it, removes high-frequency trading bots, exchange treasury rebalancing and repeated smart contract transactions. The Coinbase study goes further and also strips all activity involving exchanges, DeFi and bridges.

Both filters can be defended, and neither gives the stablecoin volume. “Adjusted” means adjusted by whoever publishes the figure, so a number quoted without its filter says little. We did not find a Visa figure for September or for the third quarter, so the two series cannot yet be compared over the same period.

Stablecoin supply stayed near $305 billion: the same dollars moved faster

The most useful brake on the growth story sits inside the study itself. According to Crypto Briefing, stablecoin supply rose just 0.4% in the first nine months of 2026, to $305 billion. Transfers grew; the money being transferred barely did.

Our own check on DefiLlama at about 16:55 UTC on 21 October agrees. Total stablecoins in circulation stood near $304.8 billion, against $304.1 billion on 30 June and $306.3 billion on 30 September, so the third quarter was flat. USDT made up about $182.4 billion of that and USDC about $72.3 billion.

Divide $891 billion by about $305 billion and each dollar of stablecoin supply changed hands roughly 2.9 times in the quarter on the report’s filtered count. That is a story about usage and velocity, not about new money entering the system. It is also consistent with a market where a few large holders move big amounts often, which an hourly average cannot distinguish from many small payments.

For readers new to the subject, our guide to what stablecoins are and how they keep their peg covers issuance, reserves and redemption, the activity this report deliberately leaves out.

Who paid for the report: Coinbase’s stake in USDC

Coinbase is not a neutral observer of stablecoin growth. Its second-quarter 2026 earnings deck, filed with the SEC in early August, reports $292 million of stablecoin revenue for the quarter, and $320 million when revenue on corporate balances is added. Average USDC in circulation was $77 billion, of which $20 billion sat in Coinbase products and $57 billion outside them. The deck says Coinbase captured about 50% of all USDC economics over the past year, and that its agreement with Circle renews automatically on the same terms.

Put plainly: the more USDC is held and used, on Coinbase or anywhere else, the more interest income Coinbase shares with Circle. A report arguing that stablecoins are becoming everyday business money, and that banks are joining in, supports Coinbase’s revenue line and its policy arguments in Washington at the same time. Our piece on how Circle splits USDC revenue with its distributors explains how that sharing works.

None of that makes the figures wrong, but the choices behind them matter more when the sponsor gains from the result. Crypto Briefing described Coinbase as a company with a “clear interest in stablecoin adoption” and told readers to treat the rosier projections with caution. The Block’s own news story is another layer: The Block Pro Research, which produced the study, belongs to the same brand, and its article presented the report as a joint study with Coinbase without a commissioned label. The coverage we read found no formal funding disclosure in the report summary, and the price Coinbase paid, if any beyond commissioning, is not public.

Seven of the ten largest US banks have tested stablecoins, but which ones?

The bank claim is the second pillar of the report: seven of the ten largest US banks have tested stablecoins or tokenized deposits for settlement, and the other three have signalled plans, per The Block and Journal du Coin. The coverage does not name the banks, and does not say what “tested” means: a live settlement, an internal pilot, a proof of concept with a vendor or a single transaction.

Grouping stablecoins with tokenized deposits also matters. A tokenized deposit is a bank liability on a blockchain, held by the bank’s own customers. A payment stablecoin is a separate token issued by a licensed issuer and backed by reserves. A count that merges the two says less about stablecoin adoption in particular than the headline suggests.

The timing gives the claim weight anyway. On the same day the coverage spread, 21 October, the Office of the Comptroller of the Currency issued its final rule for payment stablecoin issuers under the GENIUS Act, which takes effect on 18 January 2027; Coindaro covers that rule in a separate piece. Our page on the GENIUS Act’s start date sets out what changes that day.

The law bars issuers from paying yield on stablecoins, but platforms can still offer rewards, and that is where Coinbase’s interest and the banks’ meet head on. On 10 September 2026, 77 state banking associations, backed by the ABA and ICBA, wrote to Senate leaders asking for a ban on balance-based rewards, according to Crypto Briefing. The banks’ worry is deposits leaving for tokens that pay; Coinbase, which earns on USDC balances, sits on the other side of that argument. The Treasury side of the rulemaking is still open too, as our report on the GENIUS Act comment period showed.

The $1.5 trillion projection and the other forward numbers

Four of the report’s other numbers are estimates or comparisons rather than measured transfers.

• The $1.5 trillion a quarter by the end of 2027 is the past growth rate, about 50% a year, extended forward. It holds only if growth holds, as the outlets that reported it said.
• The $155 billion of extra annual global GDP is a scenario in which 80% of international transactions move to stablecoin rails. It is a hypothesis, not a forecast.
• The $3 trillion to $5 trillion of business-to-consumer commerce mediated by AI agents by 2030 is a McKinsey estimate that the report cites, via The Block. It is not the report’s own work.
• The 86% saving on a $200 USDC transfer on Base compares one chain and one coin with traditional rails, as reported by The Block and Crypto Briefing; the coverage gives no detail on which rails or fees.

Stablecoin supply that barely moved for nine months gives no sign of an acceleration, and a slower year would move the 2027 figure a long way.

What a European reader should keep in mind

Everything in the study is measured in US dollars against US banking hours, so its 73% share does not translate to European payment hours. In the European Union, dollar stablecoins such as USDC are regulated under MiCA as e-money tokens, with their own reserve and redemption rules.

Tax is the other local angle. In France, MPs have voted, at first reading of the 2027 budget only, to tax swaps from crypto into stablecoins such as USDC and EURC from 2027; our article on the French stablecoin swap tax sets out what the amendment says and what still has to happen before it becomes law.

What is still unknown, and the next dates

Several points would change how the $891 billion should be read, and none is in the coverage available on 21 October:

• The raw volume before filtering, and the share of activity removed in each quarter since 2023.
• The chains and stablecoins counted, and the weight of Base and USDC in the total.
• The names of the seven banks and what each one tested.
• A Visa adjusted figure for the third quarter, which would allow a like-for-like comparison.

The dates to watch are regulatory. The GENIUS Act and the OCC’s rule take effect on 18 January 2027; the OCC says applications from issuers will open on a date it will announce separately. Rules on anti-money-laundering and sanctions are a separate rulemaking with the Treasury. Coinbase’s third-quarter results, when they come, will show whether its stablecoin revenue moved with the transfer volume its report describes.