Coldcard's $70 Million Bitcoin Drain: Coinkite Takes the Hit, but the Real Risk Is Still Expanding

Coldcard's first-wave losses reached about $70 million

The first takeaway is straightforward: roughly $70 million in Bitcoin has already been pulled from Coldcard users, and the speed of the drain suggests active exploitation rather than a slow, isolated leak. Galaxy Research said 1,082.65 bitcoin was moved from 1,196 addresses, with most of the funds transferred in a 40-minute window on July 30.

Weak randomness turned cold wallets guessable

In plain terms, weak randomness turned "impossible to guess" seeds into guessable ones. Coinkite traced the issue to affected firmware builds, and the exposure reached back to seeds created starting in March 2021. That is what turned a wallet flaw into a broad custody risk.

What made the incident more dangerous was the expansion of affected hardware. Coinkite first flagged Mk3 devices with firmware 4.0.1 or later, then widened the warning to include some Mk4, Mk5, and Coldcard Q versions. Reports from Block also cautioned that future attacks could target any Coldcard address generated with the vulnerable firmware, meaning losses were not necessarily confined to the first wave.

The damage estimate rose as researchers tracked more flows

The headline number was only part of the story. One estimate tracked 1,082.65 bitcoin from 1,196 addresses moved in a 40-minute window. A separate read looked at 594 BTC from around 500 single-signature wallets moved in 25 minutes. Both point to rapid, automated sweeping rather than a closed or contained exploit.

The loss estimate itself changed as the picture clarified. Early reporting said $38 million is already gone. Later analysis pointed to over 1,080 BTC worth more than $70M. That rise does not prove new wallets were still being drained indefinitely, but it does show the incident was still expanding beyond initial estimates.

At the technical level, this was an entropy failure, not a routine theft. For Mk3 devices, reports said the flaw reduced entropy from 128 bits to about 40 bits. For newer devices, the issue was 128 bits to 72 bits. Either reduction weakens the security model, which helps explain why the risk could persist after the first visible sweep.

The next risk sits outside the original Mk3 narrative

The immediate damage is already visible, but the part of the story investors are still pricing imperfectly is the wider device set. Coinkite's alert later expanded to some Mk4, Mk5, and Coldcard Q versions, which broadens the exposure zone beyond the original Mk3-focused story.

What may still be underpriced

What may be underpriced is not another recap of the initial drain. It is the chance that newer Coldcard families start showing up in fresh outgoing flows, turning a contained headline event into a broader address pool that is still being scanned. The same logic applies to user migration: Coinkite did not just recommend updating firmware; it also said users should move funds to a newly generated wallet, because previously generated seeds remain exposed.

What to watch next

• Fresh outgoing transfers tied to the broader device set: some Mk4, Mk5, and Coldcard Q versions
• New address clusters moving outside the original burst pattern
• Evidence that affected users are migrating to newly generated wallets, not only installing updated firmware

If on-chain activity stops showing fresh hits from the newer device group and migration behavior clearly reduces the exposed pool, the event likely shifts from active loss expansion to reputational cleanup.