Coldcard Exploit Confirmed at 1,719 BTC Stolen as Galaxy Research Warns Losses May Top $130 Million

Galaxy Research confirms 1,719 BTC stolen, with losses still potentially rising

Coldcard is no longer a niche hardware-wallet headline. Galaxy Research now confirms 1,719 BTC stolen, with total losses likely to exceed $130 million. For portfolio holders, that is large enough to trigger active damage control.

The bigger concern is the spread, not just the headline number. Galaxy Research said it has identified more than 25 separate attack patterns and received reports from more than 250 victims. That makes this less like a single phishing break-in and more like an ongoing cleanup operation.

The core debate is whether this stays a product-specific issue or becomes a wider hit to self-custody confidence. If the bug is contained, many users will move on. If it keeps widening, the trust damage can last longer than the exploit itself.

There is at least some good news in the flow. New units ship with corrected firmware, so the immediate task is firmware action and migration, not an admission that every self-custody setup is compromised. Even so, older devices now carry more stigma, and that could slow replacements or push some users toward simpler, managed alternatives.

The market does not need another 500 BTC to feel pressure from this. Even a brief reference to a possible 2,300 BTC ceiling is enough to keep institutions cautious. In a fragile market, reputational drag and forced migration can matter before the final loss count is settled.

Why the exploit worked: weak randomness made some recovery phrases guessable

This was not a clean phishing attack. The problem lived in the way certain Coldcard devices created recovery phrases, potentially making seeds weaker than a cold wallet should allow even before hackers interacted with the user. Galaxy said the flaw left wallet seeds weak enough for hackers to guess, and Coinkite later confirmed a bug in some firmware versions before shipping a fix.

Who was actually exposed

The exposed group was narrower than a full-device failure, but still large enough to matter. Coinkite said seeds generated on Mk2 or Mk3 firmware from 4.0.1 through 4.1.9 were at risk unless the user had used at least 50 fair, independent, private dice rolls and protected the wallet with a strong, unique BIP-39 passphrase. In plain English, users who relied on the device's default randomness, did not use enough private dice entropy, and had no real passphrase were in the danger zone.

That matters because many users assume a hardware wallet is automatically safe once the recovery phrase is written down. This exploit breaks that shortcut. The weak link was not an internet connection or a stray photo; it was randomness quality at creation time, which can affect even careful users.

The cleanup is still active

The scale helps explain why this has not settled. Attacks have hit more than 5,200 individual addresses, and approximately 1,816 Bitcoin has been moved in waves. That is large enough for investigators and victims to keep sorting out who qualifies, what funds are exposed, and whether stolen coins are still liquid.

This is also not a "patch and forget" incident. Coinkite explicitly said updating the firmware does not change an affected seed, and reiterated that users should replace the seed and migrate rather than assume a firmware update repaired anything. Firmware can stop new weak seeds from being created, but it cannot fix seeds already derived from weaker randomness.

Why the firmware boundary matters

The exact firmware boundary is the whole point. Mk2 and Mk3 devices running version 4.2.0 or later are in the fixed camp, while versions 4.0.1 through 4.1.9 are the exposed range. Users who created seeds outside that window may still be fine, but anyone unsure is left in ambiguity, and ambiguity is where attackers keep pressing.

That is also why the migration message must be taken literally. Coinkite's guidance says a passphrase can reduce immediate exposure, but it does not repair an affected seed. If you are in the affected cohort, the practical fix is a new seed on updated firmware and a test transfer before moving the rest.

Coinkite's disclosed-history record adds trust pressure

This is where the vendor record becomes a pressure point, not a verdict. Coinkite's own public security chronology, not a count of independent vulnerabilities shows years of disclosed issues, which can make users ask whether this was inevitable. Critics will argue cold storage is only as strong as the last undisclosed bug. Supporters will argue the opposite: vulnerabilities were disclosed, reviewed, and hardened over time.

The key point is not whether the product history was perfect. It is that this particular fix stopped future weak seeds from being created, but it did not remove liability for seeds already generated under the older behavior. In crypto custody, that distinction determines how much cleanup is still required.

What investors should watch next: migration drag, final losses, and brand trust

Coldcard is moving from incident response to reputation management. New units ship with corrected firmware, which helps the future product story. But the near-term repricing risk sits in the installed base, where users still need to replace the seed and migrate, not just install a patch.

What can still drive pressure

The clearest pressure point is migration drag. If users are forced to move funds off old seeds, Coldcard starts to look less like a set-and-forget device and more like a product with a long cleanup tail. That matters even after the exploit is contained, because brand damage usually lags the final loss count.

Flow markers to watch

Watch the incident itself, not the narrative around it. A worsening read would come from rising victim counts, new confirmed attacker addresses, or signs that final losses are still drifting toward an earlier ceiling such as losses could exceed 2,300 BTC. Another marker is whether seizure activity shows up in the flow story; About 600 hacked wallet addresses belonged to federal investigators, industry compliance firms and cyber investigators is enough to keep that angle under close observation.

The bear case and the fade case

The bearish argument is that trust broke at the custody layer, not just inside one firmware build. That would be a real threat to Coldcard if it shifts user habits away from self-custody pride and toward simpler wrappers.

The cleaner bullish argument is narrower but concrete: there is currently no evidence the bug affects signing devices or wallets beyond Coldcard Mk3, Mk4, Mk5 and Q. If that containment holds, the issue is more likely to remain product-specific than become a broad self-custody break.

What would weaken the bearish read

The bearish case weakens if three things happen together: new thefts stop, most victims fall outside the affected firmware window, and migration proceeds without fresh public exposure under affected firmware. If that cleanup plays out quietly, the story likely narrows back to a product issue instead of a broader self-custody blowup. If it does not, the brand drag can outlast the exploit by a wide margin.