Coldcard wallet breach steals $30M in crypto

TLDR

A serious firmware bug in Coldcard Bitcoin hardware wallets allowed an attacker to steal tens of millions of dollars by guessing vulnerable wallet seeds.

1. An exploit of weak seed generation on certain Coldcard firmware drained around 594 BTC (~$38M) from roughly 500 wallets, with $30M taken in the first 10 minutes.
2. The bug reduced seed entropy dramatically, making private keys guessable; older Mk3 devices were worst affected, and later Mk4, Mk5 and Q also had weakened randomness.
3. Coldcards maker Coinkite has issued firmware fixes and urges users with seeds from vulnerable versions to treat them as compromised and migrate to fresh, well-entropy wallets.

Confidence: high because multiple independent investigations and on chain data align.

Deep Dive

1. Scale And Speed Of The Theft

Blockchain analysis shows the attacker stole roughly $30 million worth of bitcoin in the first 10 minutes, prioritizing the largest wallets, in a sweep later totaling about $38 million across 500 wallets, according to Chainalysis-backed reporting.

Further work by Galaxy Research linked around 1,196 addresses and up to 1,083 BTC, nearly $70 million, emptied over approximately 41 minutes, indicating an extended operation beyond the initial burst highlighted in the Coldcard exploit overview.

Many affected wallets were long dormant, holding significant balances, which made them attractive targets once their seeds were known to be predictable.

2. The Firmware Bug And Seed Weakness

Coldcards firmware bug sat in production for years, causing some devices to use a software random number fallback instead of the hardware random number generator when creating seed phrases, as detailed in the firmware bug analysis.

On Mk3 units, this reduced effective seed entropy from the intended 128 bits to about 40 bits, and to around 72 bits on later Mk4, Mk5 and Q devices before patches, shrinking the key space enough that a determined attacker could precompute many possible seeds and derive corresponding wallet addresses.

Because Bitcoin addresses are visible on chain, the attacker could match precomputed addresses to real balances, then broadcast withdrawals in rapid batches once ready.

3. Who Is Exposed And The Bigger Self-Custody Lesson

Coinkite advises that any Coldcard user who generated a seed on affected firmware, without strong extra entropy such as extensive dice rolls, a robust BIP39 passphrase or multisig, should assume compromise and move funds to a new wallet created on patched firmware.

Other Coinkite products like Tapsigner, Opendime and Satscard are reported as unaffected, and competitors Ledger and Trezor have stated their random number generation does not share this flaw, but the incident still raises questions about how thoroughly hardware wallets are reviewed, as noted in broader self custody commentary.

Regulators and large custodians are watching closely, and the stolen bitcoin remains mostly consolidated, so further on chain movements or additional sweeps of still vulnerable wallets are possible if users delay migration.

Conclusion

The Coldcard incident shows that even respected hardware wallets can harbor critical bugs in key generation that silently weaken security for years.

For crypto holders, the takeaway is that self custody is powerful but not infallible, and layering safeguards such as passphrases, independent entropy, multisig and diversified storage can materially reduce the impact of similar failures.