CryptoReal
CASE FILE — Oct 16, 2026

Ledger Hardware Wallets Compromised by Sophisticated Supply Chain Implants

A string of seed words, discreet copper wiring, a compact screen, and an unrequested cellular module quietly transmitting information.

On October 9, wallets linked to devices distributed through CryptoBilis—a Ledger-approved seller in Southeast Asia—began seeing assets siphoned across six blockchains.

According to Bitquery’s blockchain research, as much as $102 million was drained from 324 wallets, though Ledger itself has not confirmed the amounts or number of affected users.

The wallets’ core cryptography was not breached: Ledger’s Secure Element performed as designed. However, an illicitly added circuit board captured the seed phrases during device initialization and relayed them out via an LTE connection.

Ledger’s official threat analysis admits its Genuine Check cannot identify additional hardware unless the original chip is altered.

This vulnerability was previously flagged. In August 2025, a Reddit user from Thailand discovered unfamiliar electronics inside a Ledger Nano X; hardware security expert Joe Grand later examined the device, confirmed it was an implant, and briefed Ledger in March 2026.

By October, one device analyzed by Mark Karpelès bore a visible part number with a revision label, and similar tampered devices were reported in Flex models, at a German MediaMarkt store, and from a Singapore airport purchase.

An elaborate supply chain compromise appears to have been ongoing for over a year, with much of the early investigation led by independent researchers purchasing suspect wallets from third-party sellers.

When every visible sign—the packaging, the reseller, the authenticity verification—indicated legitimacy, how could consumers detect the deceit?

Credit: The Block, Bitquery, Ledger, Zero Day, Mark Karpelès, Tropicon 2026, hardwear.io, Grand Idea Studio, Tibane Labs, Wentrre, tanuki42, Specter, YFarmX, Arravind Prabu, cryptonews, Changpeng Zhao, Ben Yorke, Johannes, CoinTelegraph, Danny Sanders, Karpelès Lab

August 2025. A Thai Redditor purchased a Ledger Nano X from an online marketplace and, upon inspection, discovered non-standard circuitry.

He forwarded the unit to Joe Grand, a hardware security specialist familiar with rumors of wallet implants but never a confirmed case in the field.

Grand’s examination revealed additional components: a secondary circuit board, antenna, LTE module, and eSIM.

The implant was programmed to activate during wallet setup, capturing each character of the recovery phrase as it appeared on screen and then using cellular hardware to transmit the data.

Grand later reported that of eight more Nano X units ordered from Thai sellers on Lazada, four were tampered.

On March 4, 2026, Grand states he informed Ledger of his findings during a Zoom meeting. Ledger’s response was that they would investigate, but Grand says he received no update until he inquired months later.

According to Grand, Ledger told him: “Ledger had found no evidence anyone was using the implants to steal crypto.”

Grand continued his research, presenting his findings at Tropicon in April and hardwear.io USA in May, and later shared slides and the extracted firmware online.

Discussions expanded to a Signal chat with Ledger engineers and a researcher from Japan who had independently discovered similar implants.

In September, Mark Karpelès reviewed Grand’s research, acquired two Ledgers, and found implants in both. These devices featured more refined iterations, as documented on Tibane Labs and discussed on Twitter.

After the teardown and firmware became public by May, questions emerged: Who else took note, and how did Ledger respond to the warnings?

Materials and Construction

Grand’s initial sample resembled a prototype, with crude, hand-soldered wires connecting the implant to the display.

Subsequent devices showed a leap in manufacturing sophistication.

A Nano X examined by Karpelès had a custom polyimide circuit laminated to the main board, with neatly soldered points, hiding all wiring. Beneath the battery and antenna were an ST STM32G431KBU6 microcontroller, a compact eSIM, and an LTE module, with a modified battery for space. Markings indicated manufacture no earlier than December 2025.

Another device, sent from Malaysia by a Chinese Amazon Japan seller, was sold at half standard price. Its flex circuit was coated black, and part numbers had been removed. Instead of foam between screen and case, the implant occupied the space, and communications were handled by a Nordic nRF9151 modem.

A slender black antenna along the device’s edge bore a silk-screened label: AC-A2016-NANO-V1.1. Karpelès interpreted this as an official product code, implying multiple iterations. “That is a production run, not a one-off,” he wrote.

Tibane Labs noted that the flex circuit was panelized, suggesting commercial-scale fabrication.

Ten Nano X units purchased by a Japan-based researcher from unauthorized sources, mainly via Amazon Japan, all contained implants, with five distinct designs.

Around July or August, the listings switched from Nano X to Flex, and both Flex units he acquired were also compromised.

Photos of a Ledger Stax with hidden LTE components appeared on Twitter in October, but independent verification is lacking.

Ledger’s Genuine Check cannot detect such additions. Their public threat model confirms the test authenticates only the Secure Element, not external hardware.

Karpelès’s black-coated implant arrived in pristine, factory-sealed packaging from Malaysia and passed Ledger’s authenticity test.

While a hobbyist might create one device, the presence of a second version points to organized, ongoing production.

The Timeline of the Heist

The theft began with a funding transaction.

September 24: 14,512 TRX withdrawn from Bybit was sent to a wallet funding the attacker’s TRON address the following day.

Bybit withdrawal transaction:
b284a2542afed3f5e00a4f9370cae1a9675b94c29967d49d8c45819cb1d7686e

Wallet receiving funds for the TRON address:
TWGHw7ZsDdXaP6Jw3vJgJ35FHKUvetBf3E

Three minutes before that TRON address was funded, the Ethereum control address received its own deposit.

Ethereum control address:
0x381825deb972b3d0a031a9ce86902ac8b22df8f2

A series of 41 test transactions were run over two weeks by these control addresses—21 on TRON, 20 on Ethereum—testing fund transfers and approval signatures.

The final two tests occurred just 23 seconds apart on October 7.

October 9, 04:59:57 UTC: Thirty TRON wallets simultaneously added the same new key, granting it sole fund movement authority.

Key added to 30 TRON wallets:
TEqZsE39aNXRnuTwrXJ9ooWMc87puuZxpK

Only the legitimate wallet key could authorize this, meaning the attacker already controlled all affected wallets. Owners remained unaware their access was revoked.

Eight minutes following this, the withdrawals began. Thirteen TRON wallets approved outgoing transactions in one block, twelve more in the next, and $29 million in USDT was extracted within seconds.

At 05:54 UTC, 111 Bitcoin wallets were emptied in a single block.

In under an hour, the attacker targeted TRON, Ethereum, BNB Chain, Polygon, Solana, and Bitcoin.

Bitquery found a phishing attack improbable: The precision suggested software with full private key access.

Researchers tanuki42 and Specter publicly listed the draining addresses by noon UTC, and Tether responded within ten minutes, blocking 37 addresses and freezing $10 million across 20 of them.

Sums referenced in this case file

The perpetrator had already converted 23.7 million USDT to USDD, which Tether cannot freeze, and subsequently routed 26.9 million USDT via THORChain for 10,636 ETH.

Roughly 6,840 ETH was laundered through Tornado Cash over five days, much of it later passing through Zcash. Bitquery and YFarmX correlated 39 of 41 Zcash outputs to the stolen funds.

Draining continued despite publicity. At 20:49 UTC on October 9, nine more TRON wallets lost 8.8 million USDT.

The next day, a routine payment of 70,000 USDT was sent to an already compromised wallet, and the attacker claimed it shortly after, using a prior approval.

Bitquery’s October 13 analysis noted that 203.8 BTC had not yet moved.

Karpelès began sharing implant findings on Twitter the day before the main theft, and Zero Day’s Kim Zetter speculates this exposure may have accelerated the attacker’s cashout.

All affected wallets were under the control of a single entity.

Who accessed the hardware before it reached end users?

Changes in Company Ownership

CryptoBilis’s separate ownership change surfaced only after the thefts became public.

On October 9, co-founder Arravind Prabu announced CryptoBilis had changed hands earlier in the year, with founders having “fully stepped down from all operations, system access, and management.”

In a detailed statement shared by Zero Day, Prabu and co-founder Vimalatheethan specified that the transition completed in March 2026, after which they lost all access to company systems and records.

Per Prabu, the acquisition agreement prevented disclosure until mid-October 2026. All suppliers were informed of the new management, but the new owner was not named. He directed further inquiries to Nicholas Chang, previously listed as marketing director. Chang did not comment when contacted by Zero Day.

Philippine publication BitPinas reported corporate records showing 100% of CryptoBilis shares transferred on August 3, 2026, to an individual listed as “Jiaming,” with a registered address in Heilongjiang, China.

Binance’s Changpeng Zhao suggested hackers had acquired the reseller, but no evidence supporting this claim has been made public.

There is no evidence directly linking either the former or current owners of CryptoBilis, or their staff, to the implants or the theft.

However, the timeline suggests the ownership change alone cannot explain the full incident, as Grand’s implants in Thailand were found in August 2025, seven months before the transition.

Implicated devices originated from Lazada, Amazon, airport retailers, and MediaMarkt, not exclusively CryptoBilis.

Geographic patterns stand out. Karpelès’s black-coated device shipped from Malaysia, CryptoBilis’s home base.

The Japan-based researcher’s devices arrived from Malaysia, Thailand, and an Osaka location, with one address linked to multiple companies.

CryptoBilis has now halted sales of all hardware wallets across Malaysia, the Philippines, and Indonesia, closed physical outlets, and brought in external auditors.

One online shop has shut down.

Who continues to receive compromised devices?

Tracking the Implants’ Communications

Every hardware implant needs a remote recipient, but tracing this lead has proven difficult.

The most comprehensive insight comes from a Japan-based researcher, who asked not to be identified. He analyzed compromised units in his possession, finding two types of data being transmitted.

First, location: The embedded eSIMs function globally, and even unused ones retain the last-accessed cell tower. He claims these records pointed to suspected manufacturing centers, though he has not made them public.

Second, destination: Firmware analysis revealed communication with command-and-control servers hosted at an Alibaba-operated Hong Kong data center.

He stated that at least 90 compromised Ledgers were in contact with these servers in June, likely underestimating the scale, and that this was four months before any wallet was drained.

This evidence is not independently corroborated, and he could not confirm whether the same group created all the implants or if multiple actors copied the design.

Karpelès speculates the implant is a standalone product, manufactured and sold as a kit for installation in Ledger wallets before they reach customers.

Grand describes the operation as hardware espionage, a term reflected in the covert methods, concealment, and exfiltration infrastructure, with the objective being financial theft.

The methods employed resembled intelligence operations, but the motivation was criminal gain.

Who was responsible for maintaining the integrity of the supply chain?

The Authorized Seller Dilemma

For years, security advice for hardware wallets was consistent: Only purchase directly from manufacturers or official resellers.

CryptoBilis was officially recognized by Ledger, as was MediaMarkt.

Johannes pointed out that Ledger’s supply chain issue extended beyond Asia: He found a spy implant in a device bought from a European authorized reseller, later identified as MediaMarkt.

Zero Day’s reporting indicates that the German MediaMarkt purchase occurred in August, and that Johannes only inspected the device after the thefts became public. A second purchase at a different branch also contained an implant.

A MediaMarkt staff member told Johannes his device was not a returned item, noting that any returns are sent back to the manufacturer and not resold.

On October 11, Ledger reminded sellers to use only authorized supply channels and never restock returned units.

By October 13, MediaMarkt Saturn, with Ledger’s consent, removed all Ledger products from stores and websites, citing one customer report and no confirmed losses.

That same day, Ledger instructed resellers in Southeast Asia to suspend Ledger sales and pull items from shelves, while noting it had not confirmed any thefts linked to devices acquired outside CryptoBilis.

Trezor also acted, notifying some resellers that their authorization would expire on January 1, 2027.

Danny Sanders, Trezor’s Chief Commercial Officer, asserted there was "not a single piece of evidence" of Trezor wallets being compromised.

Grand reported to Trezor on August 10 about a Model T purchased via Facebook Marketplace in Canada in 2025. The microcontroller had been swapped, causing the wallet to generate one of a dozen pre-known seed phrases.

He found a similar replacement chip in a Trezor One bought on TikTok Shop in August.

Trezor responded promptly and subsequently asked Grand to notify them before publication. Both instances involved unauthorized sellers, so Trezor’s denial applies only if “compromised” excludes post-distribution tampering.

Mark Karpelès maintains a public tracker summarizing the issue: It lists 15 devices from nine sellers, spanning CryptoBilis, MediaMarkt, Lazada, Amazon Japan, Yahoo! Flea Market, JD.com, a Russian marketplace, and Singapore outlets, including an airport store.

Nearly all are Ledger devices retrofitted with a circuit and antenna that reads seed phrases from the display; others are counterfeits lacking secure elements.

The database covers only confirmed public cases, with some pending further inspection, but gives buyers a chance to verify their device or seller.

Even Ledger’s warnings lagged behind. The initial alert targeted only recent CryptoBilis customers, but Bitquery established that the vast majority of victim wallets were funded prior to that period.

By this point, “authorized” had failed repeatedly.

If being on an official reseller list was insufficient, what protection did buyers actually receive?

Has the concept of self-custody been undermined?

Not entirely. But the mantra “not your keys, not your crypto” always assumed the threat came from custodians.

This event demonstrates a new risk: If you don’t control your hardware, you don’t control your keys.

The underlying cryptography was uncompromised. Private keys never left the Secure Element.

But the recovery phrase—enough to regenerate the keys—was intercepted as it appeared on the device’s screen, the very interface meant for the user’s eyes.

Self-custody faltered not due to mathematical weakness, but because trust shifted to packaging, branding, reseller status, and the logistics chain.

Each of these presented an opportunity for interception.

The key takeaway is not that self-custody is futile, but that its security depends on every link between manufacturing and the user’s hands.

If you didn’t personally seal, ship, or unbox your hardware, whose wallet is it, really?

LedgerSupply Chain Attack
Investigation alerts

Get new scam files the moment we publish them — usually 2–3 emails a week.

Enter a valid email address.

No spam, unsubscribe anytime. We never sell your data. Crypto assets are volatile and high-risk; nothing here is financial advice.

You're on the list. Watch your inbox for the next scam file.