A rise in USDT issuance on Ethereum did not produce a sustained increase in the amount held by smart contracts, according to a Bank for International Settlements (BIS) working paper published on 15 September 2026.
The research also found that smart-contract accounts held about 1% of USDT on Tron through most of the historical period examined. The findings question the assumption that an expanding stablecoin supply automatically represents greater capital entering decentralised finance (DeFi).
The study tracks the location of tokens rather than the reasons individual holders own them. Its holder-balance chart ends before 2026, meaning the percentages shown should not be interpreted as measurements for September 2026, despite current dashboards displaying large USDT balances on both networks.
On Ethereum, smart contracts held more than 20% of USDT during parts of 2021 and 2022. That proportion remained between about 15% and 20% until late 2024, before falling to roughly 10% to 15% as the amount of USDT in circulation increased.
The decline represents a smaller share of the total supply being held in contracts, rather than proof that the absolute contract balance was continually falling. BIS researchers said the growth in issuance did not lead to a sustained rise in the amount held by smart contracts.
A proportion can decrease if newly issued tokens accumulate in other types of accounts while contract holdings remain close to previous levels. The Ethereum data showed a substantial increase in USDT held by non-contract accounts, without a comparable long-term rise in smart-contract balances. As a result, the percentage alone cannot establish that dollars were withdrawn from DeFi.
Figure 10 in the paper places Ethereum’s smart-contract USDT balance at approximately $10 billion to $15 billion towards the end of the plotted period. On Tron, the figure was around $1 billion or less. Ethereum’s dollar balance fluctuated in the low tens of billions even as its share declined, while contract-held USDT on Tron remained a small part of a much larger supply.
The researchers reconstructed holdings using transfer-event logs from Ethereum and Tron. They identified smart contracts through contract deployments, treated other addresses as externally owned accounts and checked total supply against minting, burning and blacklist-destruction events.
This method differs from protocol-level total value locked, which measures assets assigned to particular DeFi applications and can count the same tokens more than once. The BIS approach follows USDT across addresses, including holdings outside DeFi applications, making it useful for comparing account types but not for determining why tokens are held.
Smart contracts can contain USDT for bridges, wrappers or custodians rather than lending or trading. Externally owned accounts may be used for payments, savings, remittances or exchange custody. Therefore, Tron’s roughly 1% contract-held share does not show that the rest was used for payments, while Ethereum’s falling proportion does not prove that DeFi activity contracted.
On 28 September, DefiLlama showed USDT’s market capitalisation across chains at about $183.7 billion, including approximately $73.3 billion on Ethereum and $92.5 billion on Tron. Those figures are a later third-party supply snapshot, not an update of the BIS breakdown. Current claims about DeFi usage would require updated balances for identified contracts, with bridges and custody separated where possible.
