Coldcard’s 5-Year Flaw: A $90 Million Security Gap in Hardware Wallets

Hardware wallets, long considered the gold standard for securing crypto assets, have been shaken by a critical security flaw in Coldcard that went undetected for five years, serving as a severe warning that industry independent audit standards require urgent revision.

The $90 Million Entropy Failure

The magnitude of the incident became clear as over 4,500 addresses were impacted and nearly $90 million in Bitcoin was drained, prompting Kraken Chief Security Officer Nick Percoco to label the situation a "wake-up call" for the industry.

■Coinkite confirmed that a migration error during the integration of a new cryptographic library in March 2021 routed the wallet creation process to a weaker MicroPython generator.

■The flaw disabled the device's True Random Number Generator (TRNG), leading to the usage of a weaker source of randomness that attackers could predict and exploit.

■Coinkite officials admitted that while the carefully crafted TRNG code existed in the codebase, it was utilized only by chance for less important operations.

The Audit Illusion: Verification Gaps in Firmware

The core reason this vulnerability remained hidden for five years is that auditors verified the existence of the TRNG code but failed to check if it was actually invoked by the production firmware.

■Current audit processes do not systematically verify that the approved source of randomness is the one actually executing in the production environment.

■Consumers are forced to trust the manufacturer's implementation of the system's most critical function without independent verification of the entropy path.

■Percoco emphasized that hardware wallets currently rely on vendor-sponsored audits and limited certifications, lacking systematic end-to-end verification.

Standardizing the Chaos: The Missing NIST Protocols

While the payments industry and the US government already apply rigorous standards to cryptographic modules, the hardware wallet sector remains an exception to these validation rules.

■NIST SP 800-90B (US government standard) and BSI AIS-31 (German federal office for information security standard) are existing references for designing and testing physical TRNGs.

■The payments industry does not allow PIN entry devices to ship without independent lab testing, yet digital asset self-custody remains exempt from such scrutiny.

■Following the discovery, Coinkite halted all device shipments and destroyed all remaining units containing the affected firmware.

As a blockchain researcher, this event teaches me a crucial lesson: even "trustless" systems are subject to the fallibility of human-written code. Hardware wallets may be the fortress of cold storage, but if the entropy mechanism inside is compromised, the fortress gates are wide open. The sector must transition to dynamic testing standards that verify not just if the code "exists," but if it "actually runs" on the chip. Otherwise, self-custody remains merely a security illusion.