The Ledger "Attack" Nobody Lost Money To — and the Version Problem Behind a $4 Billion Story

A rival wallet maker's security team announced this week that it had reproduced an attack on Ledger, the best-known name in hardware wallets. Ledger answered that no user had been hacked. Both statements are true, and the space between them is where the useful reading sits — because Ledger has reportedly been exploring a New York listing that could value it near $4 billion, which means the whole company is a bet on a single asset: trust.

Here is what the "attack" actually was. A hardware wallet works on a simple promise: your crypto keys never leave the device, and the device only ever signs what it displays on its own small screen. You send a transaction, the screen shows you the details, you press confirm, the key signs exactly what you saw. "What you see is what you sign" is the entire pitch.

In the Ethereum app as it existed through version 1.22.1, that separation could break. If a malicious program on the connected computer slipped a second signing request in while a first review was still on screen, the device could end up signing a different transaction than the one displayed — a race condition between the display logic and the signing buffer, in security parlance. The screen showed Transaction A; the chip signed Transaction B.

The catch is the size of the threat, and it is genuinely large-to-small. To exploit the flaw, an attacker had to already control the machine talking to the device — malware on your laptop, a compromised wallet app, a hostile webpage. The device was never asked to hand over its keys. And Ledger had fixed the bug before it ever became a headline: Ethereum app 1.22.2 shipped on August 13, an underlying fix to the company's Secure SDK followed on August 21, and its Ethereum apps were rebuilt on top of that. A security firm, TestMachine, aired the finding publicly in late August; a few days later, on August 27, OneKey announced its lab reproduction against the older, unpatched version. Ledger's chief technology officer called it "a lab exercise, not a finding," and said no user was hacked and nothing had been exploited in the wild. No funds moved.

Now put that scare next to the hardware wallet's actual catastrophe of this summer, and the scale snaps into focus. In late July, the Coldcard — the austere Bitcoin-only wallet prized by the most security-obsessed users — was picked clean. A 2021 bug in how some Coldcards generated their recovery seeds let the device fall back to weak randomness, cutting key strength from 128 bits down to as little as 40, which is brute-forceable. Over four days starting July 30, attackers drained roughly 1,816 BTC — about $116 million in TRM Labs' count — from more than 5,200 addresses. The unforgiving detail: updating the firmware only protects seeds created after the update, so keys made under the flawed version are weak forever. That was a foundation crack and real money lost. The Ledger finding was a lab demonstration against a version the owner had already shipped a fix for. One is a catastrophe; the other is a scare that happened to be true at one moment in time.

But the two incidents share a shape, and that shape is the investment point. Both were version problems. Coldcard's flaw was baked into seeds generated on old firmware; Ledger's lived in an Ethereum app from before August 13. A hardware wallet's protection is only what the version you are running delivers — which is a fact worth staring at, because it contradicts how the product is sold. Hardware wallets are marketed as a durable purchase: a vault you buy once, keys that never leave it, long-term storage you can set and forget. In practice the security is leased, not owned. The vendor must keep patching code you cannot see, and you must keep installing the patches — through vendor software like Ledger Live, and for each coin app separately, updating the firmware alone is not enough. Ledger's own security team reached the conclusion out loud, noting the episode underscored the need for hardware wallets to support software updates. The guarantee holds only at the latest version, and the latest version only exists if both sides keep paying a recurring cost the customer never sees on the invoice.

That structure is where the economics get interesting, because the money flows the wrong way for a simple gadget business. Ledger posted nine-figure revenue in 2025 — over $100 million — and says it secures more than $100 billion of Bitcoin. Yet the entire hardware wallet market is small: one industry estimate put it near $0.7 billion in 2026. Selling plastic boxes caps the story, which is exactly why the reported IPO math looks the way it does. Reports this year said Ledger was seeking a valuation near $4 billion, nearly triple what investors assigned it in 2023, and pre-IPO shares on secondary platforms were trading early this year at a level implying roughly $1.4 billion. That gap is the bet: the company is worth several times its device business only if it sells its installed base something recurring — backup and subscription services, enterprise security tools — and converts a one-time purchase into an annuity.

Here is the tension that makes the bet fragile. In a device business, the buyer pays once and the seller carries the maintenance and the tail risk forever. Coldcard is the tail risk made visible: a hardware maker whose annual revenue pool is a rounding error beside the hundreds of millions its flaw cost users. When a vulnerability is found in this category, the bill lands on the vendor's trust premium, not on the researcher who found it. The researcher gets a headline; the vendor gets the repair costs, the support load, and the disclaimer in every future thread. And the cumulative effect of repeated disclosures is exactly what a trust-premium valuation is most sensitive to.

Which is the lens for reading OneKey's role. OneKey is not a neutral laboratory. It is the challenger: open-source hardware, cheaper devices, a mid-2025 Series B at a $150 million valuation led by YZi Labs — Binance's venture arm — with Coinbase Ventures, Dragonfly, and Ribbit in the cap table, and co-branding with the largest exchanges. Its security unit had published its own teardown of the Coldcard failure in late July and turned the same lens on the industry leader before August was out. Publishing a verified reproduction of an already-patched bug costs OneKey almost nothing and buys what ads cannot: the attention that comes from attaching its name to the market leader's problem. Free, verifiable security research is the cheapest distribution available in a trust market. The original finder and the vendor's patch — the actual security work — end up in the footnotes, while the messenger owns the moment.

That is the moat question, restated. A hardware wallet leader's defense is not an unbreakable chip — the industry periodically disproves that claim, and automated tools now scan device firmware for exactly this class of bug. The defense is operational: how fast the vendor finds and patches flaws, how clearly it talks about them, and how little friction stands between a user and the update that closes the hole. Every security episode is a test the leader passes or fails in public. This week was the leader passing: patch first, communicate fast, tell users precisely what to do.

For anyone holding crypto in self-custody, the instruction is small and concrete: update the apps and firmware, verify the version on the device. That is the whole remedy, and a scare like this is not a reason to abandon the device. As a reading of the company, the episode is near-noise on its own — an already-patched flaw with nothing stolen — and a small, recurring tax on the exact assumption the $4 billion valuation leans on. Each future disclosure, from a researcher, a rival, or an automated scanner, is a debit against the durability of brand trust and against a services business that Ledger has not yet proven it can sell at scale. The observable evidence is public: patch cadence, disclosure discipline, and whether users actually run the latest version. A leader that wins those three things converts security scares into proof of reliability. A leader that does not converts them into erosion.

The OneKey demo did not hack Ledger, and that is the point in miniature. In this category, risk does not announce itself as a hack. It shows up as a race condition in a version you were supposed to update — which is why the business is only as strong as the trust it re-earns, one disclosure at a time.