Wrapped Bitcoin: lending contracts hold 45%, up from 30%

The two-year increase hides a reversal. Lending balances fell during spring 2026 while wallet balances rose, before lending recovered. Following that episode helps explain what the holder chart reveals about where balances sit and what requires a record of transfers.

Where the added 57k sits#

Lending contracts' balance rose 67k BTC against a 57k increase in the total, while two buckets shrank: unlabelled contracts, the residual with no protocol label, and bridge escrow. Idle wallets gained 10k BTC and still lost four points of share. DEX pools and custody-labelled contracts ended close to where they began.

Coinbase's loan product, launched in January 2025, provides one route into these contracts: it converts a customer's Bitcoin to cbBTC and posts it to a Morpho market on Base. The collateral series counts 38k BTC at Morpho Blue on Base, but does not identify how much came through Coinbase's interface. Other borrowers can use the same markets.

The series is not monotonic. Between 18 February and 6 May 2026 lending balances fell by 27k BTC and idle balances rose by 17k over the same interval. The monthly issuer flows show cbBTC's one large net-redemption month, 11k BTC, in April 2026. Together these observations are consistent with positions being withdrawn to wallets and some tokens being redeemed. They do not trace that sequence: different holders could have contributed to each change. By September lending was back at its February level and idle was 10k BTC above its own.

What a lending balance is#

A balance at a lending market contract means the token was supplied to that market, either as collateral or to earn interest, and the balance shows only that it was supplied. The collateral series covers Aave v3, Morpho Blue and Spark and finds 120k BTC pledged. Set against 146k held by lending contracts of every labelled protocol, that leaves 26k BTC which is either supplied without being pledged or held at a lending protocol outside the covered three.

Where the labels could be wrong#

The buckets rest on contract labels, and the labels are today's, applied to every past day. A contract labelled "lending" now counts as lending on every day it held a balance, including days before it took on that role. For Aave v3 and Morpho Blue contracts, which have had one role since deployment, the label holds on every day. A contract that changed purpose is misdated for the days before the change, and finding those contracts needs the label history rather than the balances.

The lending bucket rests on the firmest labels, curated by hand or taken from the protocol's own contract registry. The custody bucket is weaker: part of its labels come from external aggregators rather than from hand review. "Idle" means an externally owned account with no protocol label, so a custodian's hot wallet, an exchange's cold storage or a market maker's inventory address counts as idle until someone labels it. The chart cannot distinguish those balances from private holdings using the absence of a label. The category name "idle" describes the classification, not the holder's intent.

Unlabelled contracts hold 27k BTC in this snapshot. Identifying some as lending contracts would change the reported split without any tokens moving. That is another reason to separate a change in classification from a change in behaviour.

Following the spring reversal#

The spring lending decline and wallet increase make withdrawals a plausible explanation, but the balance series does not identify the accounts or transfers involved. Nor does today's 26k gap between lending balances and covered collateral identify the positions that moved months earlier.

A historical comparison with the collateral series could test whether covered pledged balances fell during the episode. Linking that fall to the wallet increase and April redemptions would still require transfers. The holder series establishes a changing distribution of balances; it leaves the route between those balances unresolved.