The collapse of FTX in 2022 triggered an unprecedented surge of interest in hardware wallets, allowing manufacturers like Trezor, Ledger, and BitBox to grow their revenue. However, Coldcard always stood apart, positioning itself as a "cypherpunk" tool for maximalists: Bitcoin only, no USB connection, air-gap only. The slogan "Sleep at Night Technology" was meant to instill absolute confidence in security, but reality turned out to be far more mundane and cruel.

The incident that occurred on July 30, 2026, became a dark day for the Bitcoin community. Attackers drained nearly 1,200 addresses without touching a single device. The total damage exceeded $70 million, and the number of affected addresses grew to 5,415 by August 3. This is not an exchange hack and not phishing—it is a blow to the very concept of self-custody, hitting hardest those users who are most disciplined and technically savvy.

Fatal error in the random number generator

The root of the problem, as it turned out, lies in a treacherous integration error. In March 2021, Coinkite, the manufacturer of Coldcard, moved cryptography to the libsecp256k1 library, adding an intermediate layer called libNgU. It was in this commit, which was supposed to improve security, that a fatal mistake was made. The request for entropy generation began going to a software generator that relied on the chip's serial number and timer readings. Such data is not truly random, leading to a catastrophic reduction in strength: instead of the designed 128 bits, the Mk2 and Mk3 models retained only about 40 bits. Even in the newer Mk4, Mk5, and Q models, which used secure elements, entropy was only 72 bits.

The most ironic part is that Coinkite itself warned about the risks of a "retirement attack"—a theoretical backdoor where the manufacturer leaves a "bug" in entropy generation. In 2021, the company assured that manual dice rolls made such an attack impossible. And that is indeed true: 50 dice rolls provide randomness manually, and a broken generator does not affect such a seed. But only a handful of users took advantage of this option.

Who needs to act urgently

Everyone who created a seed phrase on vulnerable firmware (Mk2/Mk3 with version 4.0.1 and newer, Mk4/Mk5 up to 5.6.0, Q up to 1.5.0Q) without using dice rolls is at risk. It is important to understand: exporting such a seed to another device, whether a Trezor or BitBox, does not fix the weak randomness. Multi-signature schemes are also at risk if the quorum consists of vulnerable devices.

Coinkite recommends the following course of action:

  1. Determine the firmware version at the time the seed was generated.
  2. Create a new seed phrase on fixed firmware or on a device from another manufacturer.
  3. Verify access recovery using the new words.
  4. Move funds and completely abandon old addresses.

Toxic marketing and lost trust

This story is not just a technical failure. It exposes the hypocrisy and toxicity that were cultivated around Coldcard. Coinkite co-founder Rodolfo Novak built his marketing on harsh criticism of competitors, calling their devices "scam" and Bluetooth "garbage." At the same time, in 2020, the company changed its license from GPLv3 to MIT with the Commons Clause addendum, effectively closing the code for commercial use. The entropy error entered exactly the commit that removed the last traces of open GPL code. Auditing such code became pointless for third-party developers, and the vulnerability lived in the firmware for five years.

The consequences extend far beyond a single manufacturer. According to CryptoQuant data, on July 31 there was a sharp spike in deposits to exchanges in small transactions—up to 7,300 BTC. This is a mirror image of user behavior after the FTX collapse, when people mass-withdrew coins from platforms. Now, some holders, frightened by the vulnerability of the "most secure" solution, have carried their bitcoins back to exchanges, hoping for compensation. This situation deals a colossal blow to the reputation of non-custodial storage as a whole.

My analysis: This incident is a turning point. We see how the "maximalist" approach, placing a single company on a pedestal, leads to disaster. Trust in hardware wallets as "absolute protection" has been undermined, and now users will have to learn to live in a new reality where even the most reliable tools can have fatal flaws. The only reasonable response becomes diversifying risk through multi-signature on devices from different manufacturers, but this requires a high level of technical expertise. The industry needs a new standard of transparency and independent audits, otherwise we risk witnessing a repeat of such nightmares.