80,689.
That is the number of Trezor customers whose contact details and delivery addresses leaked through the ShipMonk logistics breach. It is also, weeks later, the number of inboxes that received a polished email warning that "25% of Trezor devices" carried a catastrophic entropy flaw inside their STM32 microcontrollers.
The email was elegant in the way only malicious things are. It named real silicon. It cited a chip family Trezor genuinely ships. It used the word "entropy" correctly, in the right context, with the right tone of technical concern. And its central claim was manufactured from nothing.
I pulled the original disclosure, cross-read the security threads, and did what I did back in 2017 when I spent 150 hours tracing The DAO's reentrancy logic line by line: I stopped asking whether the claim was true and started asking why it felt true. In a bear market — where the only commodity scarcer than liquidity is attention — that question is worth more than the answer.
Trezor is the oldest consumer hardware wallet still shipping. Founded under SatoshiLabs in Prague in 2013, it built its reputation on a single architectural bet: open-source firmware running on general-purpose microcontrollers, verifiable by anyone, rather than a closed certified secure element. That bet is the entire identity of the brand. It is also the reason this phishing campaign worked as well as it did.
To understand the sequence, you have to see three events as one arc rather than three headlines.
First, ShipMonk, Trezor's fulfillment partner, was breached — names, emails, phone numbers and physical addresses for 80,689 customers, including roughly 67,000 in the United States. Second, attackers used a hijacked Trezor-adjacent domain to push a phishing email to that exact list. Third, Ledger's research arm, Donjon, published work on 1064nm laser fault injection against chips in the Trezor lineage, while Tropic Square's TROPIC01 — the security chip inside the newer Trezor Safe 7 — drew fresh scrutiny for lacking independent third-party certification.
Layer onto that Casa's CEO publicly noting that BitBox users received similar messages, and ZachXBT's blunt verdict that "all hardware wallets are complete garbage," and you have a textbook trust cascade.
Note what is missing from the arc: any evidence that a device was actually compromised. No seed phrase was extracted by mathematics. No firmware was silently replaced. The silicon held. The org chart did not.
Start with the silicon, because that is where the fear is aimed.
STM32 is a real family of ARM Cortex-M microcontrollers from STMicroelectronics, and Trezor has used them across multiple device generations. Entropy weakness in general-purpose MCUs is a legitimate topic in embedded security circles — the class of problem is real, and it has been discussed publicly before. That is precisely why the "25% of devices" claim was calibrated so carefully. A number like 90% would have triggered disbelief. A number like 2% would have been ignored. Twenty-five percent sits in the narrow band where panic and plausibility coexist.
That is not amateur work. That is someone who has studied how humans process risk.
Now move to where the attack actually landed. The phishing email did not exploit a chip. It exploited a domain, a mailing list, and a customer's trust in a brand. The breach at ShipMonk delivered the addresses. The hijacked domain delivered the authority. The email delivered the pressure. Every link in that chain is an operations problem — vendor management, domain hygiene, list security. Not one of them is a cryptography problem.
Compare the cost of the two attack vectors that got conflated in the coverage. Ledger Donjon's 1064nm laser fault injection requires physical possession of the device, a laboratory, precision optical equipment, and an attacker who knows exactly which chip revision to target. The phishing email required a breached marketing vendor and a copywriter. One is a nation-state-grade curiosity. The other scales to eighty thousand people in an afternoon.
The industry is spending its security budget on the attack that is hard to execute and ignoring the one that is trivial.
Trezor has no token. There is no supply schedule, no unlock cliff, no emissions curve to model. What it has instead is trust capital, and trust capital behaves like a reserve that only depletes. The logistics breach, the phishing wave, the chip certification questions — none of these are individually fatal. Stacked together, they form a withdrawal pattern. Hardware wallets are the rare sector where the balance sheet is made almost entirely of something you cannot see and cannot easily refill.
The regulatory risk here is not securities law. It is data protection. Eighty thousand-plus records containing contact and delivery information is a serious GDPR event, and if European regulators decide to make an example, the ceiling is 4% of global annual revenue. That is a larger number than most people realize for a privately held Czech company. Watch the Czech data protection authority. If a formal investigation opens, the financial pain will outlast the phishing narrative by years.
Then there is the upstream question nobody on the marketing side wants asked. TROPIC01 comes from Tropic Square, and Tropic Square shares lineage with SatoshiLabs. The chip may be excellent. But "independently evaluated" and "evaluated by a related party" are not the same sentence, and the distinction matters enormously when your entire pitch is verifiability.
A trust anchor that certifies itself is not a trust anchor. It is a mirror.
The attack did not stop at Trezor. BitBox users received similar messages. That detail is easy to skim past and it should not be. If two separate hardware wallet brands were targeted in the same campaign, the attacker is not working from one stolen list. They are working from a dataset that spans the self-custody industry — a broker, a scraped aggregation, or a leak whose true scope has not surfaced yet.
Treat cross-brand phishing as an indicator of industry-level data compromise, not a single-vendor incident.
Here is the part that keeps hardware wallets alive through all of this, and it is not cryptography. It is friction. Moving from one wallet to another means regenerating a seed, re-deriving accounts, re-registering multisig cosigners, re-testing every backup path. Users do not stay because they have re-verified the security model. They stay because leaving is exhausting. That switching cost is the real moat, and it buys the industry time it has not earned.
And a harder thought, one I have not seen articulated cleanly. Trezor's open-source promise makes the attack more credible. Because the stack is public, an attacker can name real chips, real firmware families, real entropy concepts, and every detail checks out against the documentation. Transparency is a security virtue that doubles as an attack surface for social engineering.
We don't usually price that trade-off. This campaign is the invoice.
In 2025 I built TruthLayer, a prototype registry for AI-generated media, with three other Nairobi developers. The lesson that stuck with me was not technical. It was that users do not evaluate authenticity by verifying provenance — they evaluate it by tone. This phishing email had the tone. It had the vocabulary. It had the plausible specificity that machine-generated text now produces at near-zero cost.
The signature of a 2026 phishing campaign is not bad grammar. It is technically accurate language deployed by someone with no technical intent.
Here is where I will push back on the consensus forming around this story.
The loudest take is that hardware wallets are broken and self-custody is a fantasy for normal people. That reading gets the causality backwards. Nothing about this event suggests hardware wallets are unsafe as devices. It suggests that the companies selling them run their marketing, logistics and domain infrastructure with less rigor than they apply to silicon. Those are problems of prioritization, not fatal flaws of design.
The second loudest take — ZachXBT's blanket dismissal — is more damaging than the phish itself. When a respected investigator says every hardware wallet is garbage, users do not migrate to safer hardware. They migrate to exchanges, where the trust model is strictly worse and the failure modes are custodial rather than personal.
Panic does not push people toward better security. It pushes them toward convenience.
And notice who benefits from the confusion. Ledger's research team published genuinely excellent work — and that work happens to argue that the certified-secure-element architecture is superior. Both things can be true at once. The research can be rigorous and the timing can be convenient.
The replacement candidates are not automatically better either. MPC wallets, social recovery, delegated custody — each one introduces a new trust chain, a new vendor, a new operational surface to attack. The bear market didn't kill the hardware wallet. It just made the industry's sloppiest layer visible to everyone at the same time.
The real signal to watch is not the next email. It is the blockchain.
If the leaked list gets weaponized into actual theft — three or more attributable incidents — this stops being a phishing story and becomes a self-custody crisis. Until then, the correct posture is boring and unglamorous: verify firmware only through the official app, treat every unsolicited security warning as hostile, and assume your address and phone number are permanently public. The chain keeps no secrets about who moves funds.
Neither, it turns out, do the vendors who hold your data.
About Me — I'm Chris Thompson, a decentralized protocol PM based in Nairobi. I have been auditing, breaking and rebuilding crypto infrastructure since I traced my first smart contract vulnerability in 2017, and I write about the places where cryptography collides with the messy reality of human operations. In a bear market, I care less about which protocol wins and more about which ones survive contact with their own vendors.
