A firmware flaw that sat undetected in Coldcard hardware wallets for nearly five years has now been linked to one of the largest hardware-wallet exploits on record. The root cause traces back to March 2021, when a firmware integration error routed seed generation to a deterministic software pseudorandom number generator instead of the device's dedicated STM32 hardware RNG. A build-time guard checked only whether a configuration macro was defined, not whether it was actually enabled — so affected devices silently fell back to a software generator seeded from the chip's serial number and internal timer, values an attacker with enough information can reconstruct.
The practical effect was a collapse in key strength: seeds generated on vulnerable Coldcard models and firmware versions during that window carried as little as 40 bits of entropy instead of the intended 128, weak enough to brute-force without ever physically touching the device.
Four Waves, Tens of Millions Drained
Attackers began exploiting the flaw on July 30, 2026, draining roughly 1,816 BTC — about $116 million at the time — from more than 5,200 addresses across four separate waves. The most aggressive of those waves hit 1,196 addresses in just 41 minutes, pulling out 1,082.65 BTC worth roughly $70.2 million. Anyone who generated a wallet seed on a Coldcard device between March 2021 and the eventual patch is now advised to treat that seed as compromised and migrate funds to a newly generated one.
Visible on the Chain, Even Before the Full Story Emerged
On-chain analytics firm Glassnode had already flagged unusual dormant-Bitcoin movement in its weekly report, noting that wallet activity tied to the theft was roughly 200 times larger than typical dormant-BTC movement patterns — even as Bitcoin itself stayed range-bound at historically low volatility while global equities and gold pushed to record highs. Notably, Glassnode observed that relatively little of the stolen BTC was sent directly to exchanges, a pattern consistent with attackers routing funds through mixers or over-the-counter channels rather than cashing out through easily traceable, regulated venues.
A Reminder About Where Trust Actually Sits
The incident lands as an uncomfortable case study for the self-custody model hardware wallets are built to support: the flaw wasn't a phishing attack or a user error, but a silent defect in the trusted device itself, invisible to anyone using it exactly as intended for five years. For an industry that markets hardware wallets as the gold standard of Bitcoin security, a vulnerability of this scale and duration is likely to keep fueling debate over how much blind trust self-custody hardware should be given without independent, ongoing audits of its randomness guarantees.
