Uneasy Money: Why Ledger's Supply Chain Became a Nearly $100M Target

The Ledger-related theft was estimated by the hosts at roughly $70 million to $100 million over about three to five months, with reports placing purchases of affected devices between June and early October. Unlike earlier scams involving fake USB devices, preloaded seed phrases, or deceptive scratch-off cards, this incident reportedly involved genuine Ledger devices entering an authorized distribution channel. The resellers were described as known and authorized; Crypto Billus had reportedly worked with Ledger and other industry participants since about 2021.

The devices were reportedly modified with an LTE chip, CPU, and eSIM added to the Ledger board. The hosts said the suspected purpose was to capture information displayed from the secure element and transmit seed phrases to an attacker’s command-and-control server over LTE, but they did not establish that this transmission occurred. Because the modifications reportedly left the CPU, secure element, and firmware intact, the devices could still pass Ledger’s genuine-device check. Confirming whether they transmitted data would require examining or testing a device, potentially somewhere without a signal, such as inside a Faraday cage.

The hosts urged Ledger users not to panic: users must display their seed phrases in plain text to write down a backup. They said the recent losses they were discussing came not from rogue Ledger devices, but from people panicking and responding to malicious Google ads, fake Ledger Live apps, phishing sites, alarming emails, and phone calls. Hardware wallets can reduce exposure to computer malware by narrowing the points of compromise, but a device costing about $100 may protect assets worth $20 million. Their advice was to buy carefully, avoid concentrating assets in one place, and consider an air-gapped machine for substantial holdings.

For custody, one speaker recommended spreading assets across multiple baskets and considering simple multisig or Shamir secret sharing; a two-of-three multisig can limit the damage from one compromised device. But more elaborate arrangements bring their own risks: users can lose access by recording seed words in the wrong order or misreading them later. The discussion emphasized that security choices involve balancing the protection gained from distributing custody against the chance of making a costly mistake.

The episode also discussed an approximately decade-old XRP code bug that could have enabled infinite minting. A speaker said AI may have helped researchers search and parse old code, but it was unclear whether AI found the bug or whether anyone attempted to exploit it. XRP paid a $250,000 bounty, which one speaker considered substantial and higher than many crypto programs offer. Zach XBT and others reportedly argued the reward should have been $2.5 billion; the speaker countered that companies should generally honor published caps, while noting that caps can become outdated as firms grow. The comparison with Google’s top-end Chrome reward also raised the costs and uncertainty of security research, including researcher time, compute expenses, and the possibility of finding nothing.

The speakers anticipated renewed interest in layer-one networks, arguing that credible L1s may capture more value than L2s. Starknet was described as considering an L1 focused on quantum resistance, AI, STARKs, and SNARKs, and as expecting to be quantum-resistant by 2027 and to outperform Ethereum. The speakers disagreed about whether that move would benefit Starknet at the broader ecosystem’s expense. They said L2s can lose integration advantages as they add custom features, while EVM compatibility alone does not ensure usability or durable network effects. More Ethereum block space could weaken the original case for L2s, though they may still suit consumer apps needing cheaper, faster transactions that settle to Ethereum.

Wallet design and AI were framed as another unresolved trade-off. Argent’s smart-contract wallet faced high Ethereum gas costs, and its signatures reportedly were not consistently respected across apps; MetaMask’s broad interoperability made it easier for users and apps to connect, despite limitations such as not including Tether as a default token for years. One prediction was that AI could let users state an intent and authorize a smart-contract wallet with a passkey, while another speaker argued this shift need not require smart-contract wallets and users may distrust a single action that makes funds “magically float away.” AI could assess transaction flows before execution more effectively than current wallet indicators, but AI agents could also become an attack vector.