The Coldcard Vulnerability: How a Firmware Flaw Exposed Millions in Bitcoin

A five-year-old firmware flaw in Coinkite's Coldcard hardware wallets allowed attackers to drain roughly $130 million in Bitcoin from thousands of addresses in a matter of hours, according to Forbes — and the entry point was a bypass of the device's true random-number generator. Instead of leveraging the dedicated hardware TRNG chip, affected firmware routed seed generation through a weaker software function, collapsing the search space from "a single atom in a billion galaxies" to something modern compute clusters can crack within hours. The vulnerability reportedly slipped past human reviewers but was surfaced by AI coding agents scanning the open-source repository.

How the entropy collapse worked

Coldcard devices are engineered to produce wallet seeds via a hardware TRNG — a chip whose output is so vast that brute-forcing a seed is functionally impossible. When the firmware skipped that path, each generated seed became predictable rather than random. Any wallet created on an affected firmware version is now sitting inside a search space commodity clusters can enumerate in hours. Galaxy Research has tracked at least 15 distinct attackers pulling roughly 2,000 BTC from 4,385 addresses once the exploit went live.

The issue entered the seed-generation pipeline as early as 2021 and persisted across multiple firmware versions. Coinkite has since confirmed in a technical report that affected builds did not use the intended hardware entropy path. The critical constraint: patching the firmware protects only seeds generated after the update — keys already created on compromised firmware remain exposed indefinitely.

Correlated risk inside multi-sig

The incident is not a referendum on hardware wallets or on open-source review. It does, however, expose a structural weakness in how users assemble custody stacks. Multi-signature reduces the blast radius of a single compromised key, yet if signing keys are generated on the same device, firmware, or entropy source, that isolation collapses. In a 2-of-3 setup where two keys share a common origin, a correlated vulnerability can effectively flatten the multi-sig advantage into a single point of failure.

Mixin, a self-custody product team, has highlighted this attack vector in the wake of the disclosure, framing correlated risk as the under-discussed variable in custody architecture. Their technical note on the incident emphasizes that multi-sig security depends not on the count of keys but on who generates them, which devices and entropy sources back them, and whether they share infrastructure or code.

What to verify and what to do

For any operator holding funds on a Coldcard generated before the patched firmware, exposure should be assessed against three variables: device model, firmware version, and seed-generation method. Where risk is confirmed, the prescribed mitigation is generation of a new seed on clean firmware and migration of assets — not simply upgrading in place.

Market signal worth tracking: on-chain analytics platform Santiment has reported 2.27 million new Bitcoin wallets created during the period of custody concerns — a flow that hints at how retail is repositioning around hardware-wallet risk.

Architecturally, the incident argues for diversification across entropy sources, signing devices, and key custodians. Independent audits remain necessary but insufficient when a single buggy commit can undermine an entire hardware class. For exchanges and high-net-worth holders, the takeaway is structural: a custody stack concentrated on one vendor, one firmware branch, or one code lineage carries risk that multi-sig alone does not offset.