Jonathan Goodman never shared his seed phrase with anyone. His Coldcard hardware wallet never touched the internet. His recovery backup sat sealed in a fireproof safe, with a second copy in a bank safety deposit box. He did everything the security guides tell you to do — and on the morning of July 31, 2026, he woke up to an empty wallet and a $1.6 million loss.
“I followed every best practice,” Goodman told TechCrunch. “My devices never touched the internet. Everything was kept in multiple safes and safety deposit boxes.” He was not hacked through negligence. He was betrayed by a firmware update from 2021 that no one caught for five years.
Goodman is one of thousands of victims in what is already being called the most consequential hardware wallet breach in Bitcoin’s history. Between July 30 and August 4, 2026, attackers drained 1,816 BTC — worth approximately $116 to $130 million depending on exchange rates during each wave — from more than 5,200 Coldcard wallet addresses. The attack exposed a flaw so fundamental it challenged the core promise of cold storage: that keeping your keys offline keeps them safe.
A Bug Hidden in Plain Sight Since 2021
The vulnerability traces back to firmware version 4.0.1, released by Coinkite — the Canadian company that manufactures Coldcard — in March 2021. Security researchers at Block, who analyzed the exploit, found that this update introduced a build configuration error with catastrophic consequences: instead of pulling randomness from the device’s dedicated hardware entropy source, the seed phrase generation process silently fell back on a software random number generator.
In cryptography, randomness is everything. A seed phrase — the 12 or 24 words that control access to a Bitcoin wallet — is supposed to be drawn from a space so vast that guessing it is computationally impossible. Coldcard’s hardware entropy source was designed to guarantee that. The software fallback did not. According to TRM Labs, which published a detailed on-chain analysis, the effective key strength collapsed from the intended 128 bits of entropy to as little as 40 bits — a reduction that made brute-force attacks not just feasible but fast.
Forty bits of entropy means roughly a trillion possible combinations. That sounds enormous until you realize that a modern GPU cluster can test billions of combinations per second. For an attacker who could mirror Coldcard’s flawed generation process on their own machines, checking every possibility was not a matter of years. It was a matter of hours — or minutes.
The flaw affected every wallet whose seed was generated on firmware 4.0.1 or later versions that inherited the bug, covering five years of Coldcard purchases. Coinkite had not publicly disclosed the vulnerability before the attacks began.
Four Waves, 41 Minutes, $130 Million Gone
The attackers moved with surgical precision. The first wave hit on the evening of July 30, draining 1,083 BTC from 1,196 addresses in just 41 minutes. The second wave followed the next morning — 594 BTC swept from approximately 500 single-signature wallets in 25 minutes. A third wave over the weekend cleared another 208 BTC from 1,912 addresses. By Monday, a fourth wave was detected, bringing the confirmed total to 1,816 BTC and pushing estimated losses past $130 million.
The speed of the attacks suggests significant pre-computation. The attackers almost certainly spent weeks or months generating candidate seed phrases before the first transaction fired — quietly building a database of vulnerable private keys, waiting for the right moment to sweep them all. When that moment came, the executions were automated and nearly simultaneous across thousands of addresses.
TRM Labs noted that differences in transaction construction across the four waves point to multiple attackers operating independently — different scripts, different timing patterns, slightly different fee strategies. This was not a single coordinated group but several bad actors who had apparently discovered or purchased the same exploit methodology and launched separate campaigns in close succession.
On-chain laundering was surprisingly unsophisticated. Analysts observed a single 64.9 BTC deposit into Wasabi Coinjoin and approximately 200 ETH moved into Tornado Cash on August 4. Most of the stolen funds remained consolidated at a small number of attacker addresses with minimal onward movement — an unusual pattern that TRM analysts said was inconsistent with professional groups such as North Korea’s TraderTraitor syndicate, which typically begins layering stolen funds immediately.
Cold Storage’s Trust Crisis
Hardware wallets exist precisely because software solutions — exchange accounts, desktop wallets, browser extensions — are considered too exposed. The pitch is straightforward: your private key is generated inside a dedicated secure element, never touches an internet-connected device, and cannot be extracted without physical possession of the hardware. Coldcard, manufactured in Canada and endorsed by prominent Bitcoin security advocates, was considered one of the most trusted options in the market.
The July 30 attack dismantled that trust in ways that go beyond Coinkite. The exploit required no physical access to the device. It required no social engineering. It required no malware. The attackers simply replicated a flawed process that Coldcard’s own firmware had already performed — and used the predictable output to reconstruct keys that were supposed to be unguessable.
Victims like Goodman are left with a disquieting realization: the attack surface for a hardware wallet is not just the device itself. It is also the software that generates keys on that device, the firmware update pipeline that introduces changes to that software, and the quality assurance process — or lack thereof — that audits those changes before they ship to customers.
The breach now ranks as the third-largest individual crypto hack of 2026. It arrives against an already grim backdrop: total cryptocurrency losses this year exceeded $1.2 billion across 276 documented incidents through early August, according to TRM Labs data — a pace that puts 2026 on track to rival or surpass 2022’s record-setting theft totals.
Coinkite’s Response and Its Limitations
Coinkite issued a public advisory on August 4, describing the preceding days as “some of the hardest in this company’s history.” The company confirmed the vulnerability, urged all users who had generated wallet seeds on affected firmware to migrate funds immediately, and released an updated firmware version that patches the random number generation flaw.
The advisory carries an important and painful caveat, noted explicitly in TRM Labs’ analysis: “A firmware update prevents new wallets from being generated with weak randomness, but it does not retroactively fix a seed that was already generated on vulnerable firmware.” For users whose wallets were created between March 2021 and the patch release, the only safe course of action is to generate an entirely new seed on updated hardware, transfer funds to the new wallet, and verify the new wallet’s fingerprint before treating the old one as abandoned.
For some users, that window has already closed. Wallets that had not yet been drained at the time of the advisory were likely swept in subsequent waves before their owners could act. The gap between when Coinkite became aware of the issue and when the public advisory was issued remains unclear, and the company has not addressed that question publicly.
What This Means for You
If you own a Coldcard hardware wallet, the first step is straightforward: do not wait. Open the Coinkite website directly, verify you are on the official domain, download the latest firmware, and check which version was used to generate your current seed. Coinkite’s advisory outlines the affected firmware range.
If your seed was generated during the vulnerable period, treat that wallet as compromised regardless of whether funds have been moved. Generate a new seed phrase on updated firmware. Do this on the device itself — not on a computer or phone. Verify the new wallet fingerprint matches what appears on the device screen. Then transfer funds to the new address using a small test transaction first before moving everything.
If you use a different hardware wallet manufacturer, this incident is still a prompt for action. Verify that your device’s firmware is current. Review whether your manufacturer has ever issued security-related firmware advisories and whether you applied them. Consider whether the seed phrase generation process for your device has been independently audited. The question “could my device have a similar flaw?” is worth asking of every manufacturer, not just Coinkite.
More broadly, this attack exposes the limits of a security model that places blind trust in hardware manufacturers. Independent firmware audits, open-source firmware with public code review, and reproducible builds — where anyone can verify that the compiled firmware matches the published source code — are not optional extras in a world where a silent configuration error can sit dormant for five years before costing users $130 million.
For users who keep significant amounts of Bitcoin in cold storage, this is the moment to revisit the supply chain of trust underneath your setup. The device is not the whole story. The firmware is. The update process is. The audit trail is. And if any of those elements is opaque, it is an exposure you may not know about until it is too late.
The Broader Lesson for the Industry
The Coldcard exploit lands at a moment when the hardware wallet market is growing faster than its security infrastructure can keep pace. Rising Bitcoin prices throughout 2025 drew millions of new users into self-custody, many of them purchasing hardware wallets for the first time. Those users were told they were safer than leaving funds on an exchange. For the most part, that remains true — but “safer than an exchange” is a low bar, and “safe enough to hold life-changing amounts of money” requires something much more rigorous.
The incident also reinforces an underappreciated point about crypto security: the most devastating attacks in recent years have not exploited complex smart contract logic or broken cryptographic primitives. They have exploited the human and engineering systems layered on top. A build configuration error. A weak default. A dependency that slipped through review. In July 2026, as in so many breaches before it, the lock was fine. The key that opened it was made wrong.
Jonathan Goodman did everything right. The device he trusted did not. Until the hardware wallet industry standardizes independent firmware audits and mandatory reproducible builds, that gap between user diligence and manufacturer accountability will keep producing victims — regardless of how many safes their seed phrases are stored in.
Sources
- Fortune — Bitcoin owners rocked by $116 million hack: What we know about the Coldcard exploit
- TRM Labs — The Largest Hardware Wallet Exploit of 2026: Inside the $116 Million Coldcard Hack
- TechCrunch — Hackers steal over $130M by exploiting bug in offline hardware wallets
- Bloomberg — Coldcard Bitcoin Wallets Compromised as Hackers Exploit Software Flaw
- Fieldfisher — Coldcard hack: what happened and what victims can do to recover