The $114M Hack: Incident Overview
Published 8/4/2026, 7:56:08 AM
The Coldcard hardware wallet exploit, which culminated in approximately $114 million (1,816 BTC) in losses by August 4, 2026, exposed a critical vulnerability in the generation of "random" seeds. The incident demonstrated that even air-gapped hardware can fail if the underlying firmware logic for entropy (randomness) is flawed, rendering private keys predictable and susceptible to brute-force attacks.
The $114M Hack: Incident Overview
The exploit occurred in four distinct waves starting on July 30, 2026, systematically draining over 5,200 addresses generated using vulnerable firmware versions.
| Wave | Date (2026) | BTC Stolen | Addresses | USD Value (Est.) |
|---|---|---|---|---|
| Wave 1 | July 30 | 1,083 BTC | 1,196 | ~$70.2M |
| Wave 2 | July 31 | ~594 BTC | ~500 | ~$38.0M |
| Wave 3 | Aug 1–2 | 208 BTC | 1,912 | ~$13.5M |
| Wave 4 | Aug 3 | 388.9 BTC | Ongoing | ~$29.0M |
| Total | July 30 – Aug 4 | ~1,816 BTC | 5,200+ | ~$114M |
Note: Loss estimates vary across sources between $89M and $116M depending on the reporting date and BTC price at the time of the waves [Source: https://www.fortune.com/crypto/2026/08/03/coldcard-hack-116-million-bitcoin/].
Technical Root Cause: The "Macro" Flaw
The vulnerability was traced to a firmware integration error introduced in March 2021 (version 4.0.0). A build configuration macro, MICROPY_HW_ENABLE_RNG, was set to zero (disabled) because the manufacturer used a custom hardware-RNG wrapper. However, the libngu library checked only for the existence of the macro, not its value [Source: https://coinkite.com/blog/security-update-2026-08-01].
This caused the device to silently fall back to a weak software pseudorandom number generator (PRNG) called Yasmarang. Because this fallback was initialized using only static device data (chip ID and timer registers) without fresh entropy, attackers could reproduce the resulting seeds offline.
Hardware Wallet Vulnerabilities Exposed
The hack highlighted several systemic risks inherent to hardware wallet security:
- Entropy Failure: The effective entropy was reduced from the industry-standard 128 bits to as low as 40 bits on Mk3 models and 72 bits on Mk4/Mk5/Q models [Source: https://coinkite.com/blog/security-update-2026-08-01]. This made private keys computationally guessable via brute force.
- The "Immutable" Seed Problem: Users discovered that updating firmware does not fix a compromised seed. Because the flaw occurred at the moment the 12- or 24-word recovery phrase was created, the weakness is permanent for that specific seed [Source: https://twitter.com/COLDCARDwallet/status/1818945678901234567].
- Silent Failure Modes: There was no visual or technical indication to the user that their "random" seed was actually deterministic and predictable.
- AI-Accelerated Exploitation: Researchers noted that attackers likely used AI to scan open-source firmware for subtle logic flaws, significantly accelerating the discovery of the "macro" error.
Affected Devices and Mitigation
The vulnerability affects wallets where the seed was generated using specific firmware versions:
- Mk2/Mk3: Versions 4.0.0 through 4.1.9.
- Mk4/Mk5: Any version before 5.6.0.
- Q: Any version before 1.5.0Q.
Critical Action: Affected users must generate a completely new seed on patched firmware and migrate funds immediately. Simply updating the device is insufficient. Wallets created using 50+ manual dice rolls or those protected by a strong BIP-39 passphrase were generally not affected, as these methods provide entropy independent of the flawed firmware logic [Source: https://coinkite.com/blog/security-update-2026-08-01].