Coldcard's $116 Million RNG Hack: A 5-Year Entropy Flaw, 5,200 Wallets, and No Easy Fix

The Coldcard drain hit the place investors expected to be safest

Starting July 30, attackers pulled roughly 1,816 BTC from more than 5,200 addresses in four waves. That matters because Coldcard was bought for one reason: to keep keys offline and make theft hard. When that assumption fails, the damage is not just financial. It hits confidence in self-custody itself.

The speed of the attacks made the incident especially damaging. Initial reports described a fast first sweep, with later waves continuing to drain affected wallets. Most victim funds were found pooling at a small number of attacker-controlled addresses with limited onward movement, which means the stolen BTC remained largely visible but also left room for further transfers and more victims to come forward.

The root cause was weak randomness in Coldcard firmware

This was a wallet implementation failure, not a Bitcoin protocol failure.

Bad entropy weakened generated seeds

A March 2021 firmware integration error routed seed generation to a deterministic software pseudorandom number generator instead of the device's hardware RNG. That collapsed effective key strength from 128 bits down to as little as 40 bits on older devices, making some generated keys much easier to brute force remotely.

Block traced the problem to a build configuration issue: the device was initialized from its unique ID and timer registers and collected no fresh entropy after initialization. In practical terms, the wallet produced a smaller and more predictable pool of seed candidates than intended.

Attackers could guess seeds offline and verify them on-chain

Once the randomness is weak, physical access is not required. Block said an attacker who can determine or sufficiently constrain the device UID, timer state, and prior RNG-call history can reproduce candidate output streams offline.

From there, the attack moves on-chain: candidate seeds are turned into addresses, and those addresses are compared with public blockchain data to identify live wallets. The chain looks like this:

• weak RNG
• weak seed
• offline guessing
• on-chain confirmation

Importantly, this does not imply that every Coldcard is affected. Exposure depends on the firmware running when the seed was created, not on the firmware version installed later. Coinkite shipped emergency patched firmware for every affected model and release track on July 31.

The patch does not repair previously generated seeds

The most important limitation is simple: updating firmware does not fix wallets that were already compromised. Coinkite advised users with exposed seeds to generate a new seed on patched firmware and move their funds. Restoring an old, affected seed on updated firmware or another wallet carries the weakness forward.

The clean takeaway is that the failure lived in Coldcard's seed-generation path, not in Bitcoin itself. The root cause was not a flaw in the Bitcoin protocol itself.

The market reaction split between system resilience and trust damage

Even after a major cold-storage breach, Bitcoin price action has not broken into the same panic seen in broader crypto collapses. But sentiment has turned sharply negative, and the incident has reopened debate over how much trust investors should place in self-custody hardware.

Bulls see a vendor-specific failure; bears see a confidence problem

Bulls argue this remains a vendor-specific issue rather than a failure of Bitcoin. The root cause was not a flaw in the Bitcoin protocol itself, and minimal laundering so far means the stolen funds remain largely traceable. If that containment holds, the damage can be treated as a reputation hit to Coldcard and related products rather than a structural shock to Bitcoin.

Bears focus on the longer-term confidence risk. A breach of a device marketed for offline security can damage trust for longer than the direct loss itself, because the point of using a Coldcard is supposed to be one of the safer ways to store cryptocurrency. If investors start viewing self-custody hardware as less dependable, the fallout could spread beyond one vendor's sales.

What would change the read from here?

The most important signposts are straightforward:

• whether stolen funds begin moving more aggressively through mixers or dispersal chains
• whether new affected addresses continue to emerge
• whether the sentiment damage starts showing up more broadly across self-custody products

If those signals stay contained, this looks more like a severe implementation failure. If they worsen, the incident starts to look less like a isolated vendor issue and more like a wider confidence drag on cold storage.