Hardware Wallet Reviews 2026: Cold Storage After the Coldcard Hack
A five-year-old firmware bug drained $116 million from Coldcard wallets in 2026. We review every major hardware wallet, Ledger to Tangem, and score what actually keeps your crypto safe.
The five-year-old bug that rewrote hardware wallet reviews
On 30 July 2026, Bitcoin began leaving Coldcard wallets that nobody had touched. There was no phishing link, no malicious token approval, no seed phrase left on a kitchen notepad. The keys themselves were weak, and they had been weak since a firmware build shipped in March 2021. By the time investigators reconstructed the pattern, TRM Labs put confirmed losses at about 1,816 BTC, roughly $116 million, while Galaxy Research counted 1,596 BTC across three attack waves, with a suspected fourth that could push the total near 2,055 BTC.
That one incident reshaped how we review hardware wallets more than any product launch this year. For most of the past decade, a hardware wallet review read like a gadget review: screen size, supported coins, whether the companion app looked nice. The Coldcard entropy failure was a blunt reminder that these devices have exactly two jobs, generate a secret nobody can guess and sign only the transactions you intend, and that a review which does not interrogate how a device does both is not really a review at all.
So this is a reset. We look at every major hardware wallet on sale in late 2026, Ledger, Trezor, Coldcard, Keystone, BitBox02, Foundation and Tangem, and score them on the criteria that decide whether coins survive contact with a hostile world. Along the way we fold in the year’s hard lessons: a firmware flaw that drained nine figures, a laser lab that reset a rival’s card without the password, a data breach at a shipping partner, and a broad surge in violent robberies aimed at holders. The hardware is genuinely good. The threat model got harder.
A hardware wallet is not the same category of product as the browser extensions and phone apps most people meet first. If you want the trade-offs on that side of the fence, our MetaMask vs Phantom vs Rabby showdown covers the hot-wallet field in detail. The one-line version: a hot wallet keeps your keys on an internet-connected device, a hardware wallet keeps them on a dedicated chip that never exposes them, and the distance between those two designs is the whole reason cold storage exists.
The backdrop matters too. With Bitcoin softening into the high $70,000s in late August after a hawkish Federal Reserve turn at Jackson Hole, and with self-custody now the default advice after every exchange scare, more coins are moving into personal custody than ever. The hardware wallet market was worth about $0.54 billion in 2025 and is forecast to reach roughly $2.25 billion by 2031, with unit sales up around 31% last year, according to CoinLaw. Buyers have never had more choice, or more ways to get it wrong.
What a hardware wallet actually protects (and what it does not)
Before scoring anything, it helps to be honest about the threat model. A hardware wallet keeps your private key, or the seed that derives it, inside a chip and signs transactions internally, so the secret never touches your laptop or phone. That design defeats a specific and very common set of attacks: malware that scrapes keys off a computer, remote intrusions, keyloggers, and the platform risk of leaving coins on an exchange that freezes, fails or decides you are no longer welcome. That last point is not hypothetical; as our piece on crypto debanking lays out, losing account access is a live risk, and a device you control cannot be closed by someone else’s compliance department.
What a hardware wallet does not do is equally important, because most real-world losses in 2026 exploited the gaps rather than the chip:
- It cannot save you from approving a malicious transaction yourself. If you blind-sign a drainer, the device faithfully signs what you told it to.
- It cannot fix bad entropy. The Coldcard victims held their keys offline the entire time; the keys were simply guessable from the start.
- It cannot restore a lost or destroyed backup. Your recovery phrase, or a seedless card’s backup cards, is the real master key.
- It cannot stop someone pointing a weapon at you. So-called wrench attacks bypass every chip on the market.
- It cannot undo phishing that talks you into typing your seed into a website or a fake support agent.
There is also a regulatory wrinkle worth stating plainly. Self-custody sits outside the perimeter that supervisors actually regulate. In the United States the SEC and its peers police exchanges, brokers and custodians, not the code on a device you own, and in the EU the MiCA regime governs crypto-asset service providers rather than the act of holding your own keys, a distinction our guide to DeFi compliance unpacks. Practically, that means nobody is coming to reverse your mistakes. You are the custodian, the compliance desk and the disaster-recovery plan.
How we score a hardware wallet in 2026
With that in mind, here are the nine criteria this guide weighs, roughly in order of how much they should move a buying decision after the year we have had.
- Key generation and entropy. Where does randomness come from, is the process auditable, and can you verify it? The single most important line in any 2026 review.
- Secure element and certification. Is the key stored in a dedicated, tamper-resistant chip, and to what Common Criteria level is it evaluated?
- Source model and reproducibility. Is the firmware open, can builds be reproduced bit for bit, and has anyone independent actually looked?
- Transaction verification. Does the device show human-readable details on its own trusted screen (clear signing), or ask you to approve an opaque payload (blind signing)?
- Connectivity and air-gap. USB only, Bluetooth, NFC, or fully air-gapped over QR and microSD? More radios mean more convenience and more surface.
- Backup, recovery and inheritance. Standard BIP-39 seed, Shamir backup, seedless cards, or a proprietary scheme, and how does your family recover funds if you are gone?
- Supply-chain integrity. Tamper evidence, genuine-device checks, and how the unit reaches your door.
- Physical-attack resistance. PIN and passphrase design, duress options, and how the hardware holds up to fault injection.
- Price and asset support. What you pay once, and what you can actually hold.
Notice that price sits last. It matters, but a $79 device that generates strong keys and shows you what you are signing protects your coins better than a $399 showpiece you blind-sign on. The Coldcard losses were not about money spent; several drained wallets ran a device widely regarded as one of the most security-focused on the market.
Secure elements and the EAL numbers, explained
Half the marketing in this category revolves around the secure element, so it is worth understanding what the term buys you. A secure element is a dedicated chip designed to store secrets and resist physical extraction, the same family of hardware used in passports and bank cards. It is different from a general-purpose microcontroller, which is easier to probe. Devices without one, notably Trezor’s older Model One and Model T, keep the seed in ordinary flash, which is why Kraken’s security team showed that a stolen older Trezor could be opened with about 15 minutes of physical access and a kit costing under $100. The newer Safe line closed that gap by adding a secure element.
Certification is measured on the Common Criteria scale, where EAL5+, EAL6+ and the rare EAL7 describe how rigorously a chip was evaluated against physical and logical attack. Ledger builds around STMicroelectronics ST33 secure elements rated EAL5+ to EAL6+; Trezor’s Safe 3 and Safe 5 use an Infineon Optiga rated EAL6+; Tangem cards use a Samsung S3D232A at EAL6+; Coldcard pairs two secure elements from different vendors; and Keystone stacks three Infineon chips. NGRAVE markets an EAL7 operating system, the highest consumer claim on the market.
Here is the nuance most reviews skip: a high EAL rating describes the chip’s resistance, not the whole product’s safety. Ledger’s own Donjon security lab spent 2026 proving the point, defeating logic that ran on top of EAL6+ silicon in both a Tangem card and a Trezor Safe 7 without breaking the chip itself. Certification raises the cost of an attack; it does not make a device unbreakable, and it says nothing about firmware bugs like the one that sank Coldcard, whose secure elements worked exactly as designed while the software fed them a weak seed.
The open-source question: sealed versus auditable
The deepest philosophical split in this market is whether you should trust a sealed, certified black box or an open design you can inspect. Ledger sits firmly on the first side: its BOLOS operating system is closed, and the company argues that secrecy plus certification plus its Donjon red team is the stronger model. Trezor, Coldcard, Keystone and BitBox sit on the other: fully open firmware that anyone can read, on the theory that sunlight finds bugs faster than secrecy hides them.
Trezor’s chief executive Matej Žák puts it bluntly. “We are the most secure because we are the most open-source,” he told DL News around the Safe 7 launch, adding that the old approach of proprietary chips had a fatal flaw: “In the past, we used chips under NDA, and if we found a problem, we couldn’t warn anyone.” The Safe 7’s headline feature answers exactly that complaint. It pairs a conventional Infineon secure element with TROPIC01, from Czech firm Tropic Square, billed as the first openly auditable secure element with no NDA required to inspect it.
Ledger’s leadership frames the trade-off differently. Chairman and chief executive Pascal Gauthier argued in a July 2026 essay that the harder problem is software surface area, not secrecy: “Everything that is in software is close to impossible to protect,” he wrote, and “a device built to do everything cannot fully protect anything.” Both men are describing real risks. And the Coldcard hack is the case that refuses to settle the argument, because Coldcard is open source and still shipped a catastrophic entropy bug that sat in public code for five years. Open source is necessary, but it is plainly not sufficient; what closes the loop is reproducible builds and people who actually read the code and reproduce the binaries, not just the promise that they could.
Ledger, reviewed
Ledger remains the market’s default, with more than eight million devices sold, per CoinLaw, plus the widest asset support and the most polished app in the category. The 2026 lineup runs from the $79 Nano S Plus through the Bluetooth Nano X at $149 and the touchscreen Nano Gen5 at $179, up to the E-Ink touchscreen Flex at $249 and the curved-screen Stax at $399, according to retail pricing trackers. All share the closed BOLOS operating system on an ST33 secure element.
The strengths are real: the broadest coin and app support, a genuinely good trusted-display experience on the Flex and Stax, NFC for tap workflows, and a leadership role in clear signing (more on that below). If you hold a diverse portfolio across many chains and want the smoothest software, Ledger is hard to beat.
The weaknesses cluster around trust and history. The 2020 e-commerce breach exposed close to a million customer emails and hundreds of thousands of physical addresses, feeding years of phishing and extortion. In December 2023, a compromised ex-employee’s account poisoned Ledger’s Connect Kit library and drained several hundred thousand dollars from web3 users before a fix landed. And the 2023 launch of Ledger Recover, an optional paid service that shards your seed across custodians with identity verification, triggered a backlash over the idea that a seed could ever leave the chip at all. None of these broke the secure element, but they are the reason closed-source skeptics stay skeptical.
Verdict: the best all-rounder for multi-asset holders who value software polish and are comfortable trusting a certified black box. Buy direct, never secondhand, and treat every email claiming to be from Ledger as hostile until proven otherwise.
Trezor, reviewed
Trezor is the open-source standard-bearer, and the 2026 range finally pairs that transparency with modern secure elements. The Safe 3 sells for about $79, the Safe 5 lands around $169, and the flagship Safe 7 launched in Prague in October 2025 at $250 and began shipping that November. The Safe 7 is the most interesting device of the year on paper: dual secure elements (the auditable TROPIC01 plus an Infineon Optiga rated EAL6+), an STM32U5 processor, a 2.5-inch color touchscreen, USB-C, Bluetooth and Qi2 wireless charging, and the industry’s first post-quantum firmware-signing scheme.
The strengths: full open-source firmware, an auditable secure element that lets researchers inspect what was previously hidden, and a company culture that treats disclosure as a feature. When Ledger’s Donjon lab turned its laser fault-injection rig on the Safe 7 in 2026, the open model meant the finding could be discussed publicly, with no user funds lost. That is the transparency dividend in practice.
The weaknesses: the older Model One and Model T lacked a secure element and should be retired for serious sums. On the operational side, Trezor has had two customer-data incidents worth knowing about, a January 2024 support-portal breach affecting tens of thousands of users, and a fresh one in August 2026 when a third-party fulfillment partner, ShipMonk, was accessed without authorization, exposing contact and shipping details for roughly 13,700 recent customers. Trezor stressed that “our systems and devices remain secure,” and no keys were involved, per Bitcoin Magazine, but the episode is a reminder that your name and address can leak through the supply chain even when your seed does not.
Verdict: the top pick for buyers who want to verify rather than trust, and the Safe 7 is the most future-proofed flagship on sale. The Safe 3 is also the best-value secure-element device for beginners on a budget.
Coldcard, reviewed after the hack
Coldcard, made by Canada’s Coinkite, is the Bitcoin-only purist’s tool: the Mk4 at about $178 and the keyboard-equipped Q at $249, both with dual secure elements, true air-gapped signing over microSD and QR, a duress PIN that opens a decoy wallet, and the deepest multisig support in the category. For years it was the device serious Bitcoiners recommended without hesitation. Then came the hack.
The mechanics are worth understanding because they are the lesson of the year. A March 2021 firmware build (version 4.0.1) shipped with a build-configuration error that caused seed generation to fall back on a weak software random number generator instead of the hardware source. Effective key strength collapsed from the intended 128 bits to roughly 72 bits on Mk4-class devices, and as little as 40 bits on older Mk2 and Mk3 units, low enough to brute-force with ordinary computing power and no physical access, as crypto.news detailed. Crucially, a later firmware fix does not regenerate an already-created seed, and even a passphrase did not save affected users. Anyone who generated a seed on a Coldcard between that build and the patch has to assume it is compromised and migrate to a freshly generated one.
The stolen coins did not vanish into thin air; TRM Labs traced the drains across multiple waves, and the laundering that followed looked much like the flows we mapped in the bridge-hack money trail, funds hopping through mixers and swaps in an attempt to break the chain. Coinkite patched the flaw and urged migration, but a fix cannot un-guess a key.
Verdict: still a formidable Bitcoin-only, air-gapped, multisig-ready device, and the Q’s full keyboard and large screen make careful verification genuinely pleasant. But the incident is a permanent asterisk, and it is the clearest argument in this whole guide for not trusting any single device or vendor with everything. For advanced Bitcoiners running multisig, Coldcard still earns a place; just not as your only signer.
Keystone, BitBox02, Foundation and the air-gapped field
Beyond the big three, a strong field of specialists deserves your attention, especially if you prefer air-gapped designs that never plug into anything.
- Keystone 3 Pro (around $149). Fully air-gapped, signing only over QR codes, with no USB data path, Bluetooth or NFC, three Infineon secure elements, a 4-inch touchscreen, a fingerprint reader and open-source firmware. The strongest QR-native all-rounder, and a natural match for mobile-first users.
- BitBox02 (around $150). A Swiss, open-source, dual-chip device with a minimalist capacitive-touch design and USB-C. Understated and well-audited, with a Bitcoin-only edition for purists.
- Foundation Passport (Core $199, Prime $349). A Bitcoin-focused, air-gapped design with a phone-like build; the Core is fully air-gapped over QR and microSD, while the Prime adds NFC and Bluetooth and therefore is not air-gapped, a distinction worth reading carefully before you buy.
- NGRAVE Zero (around $398). A premium, air-gapped device marketing an EAL7-rated operating system, closed-source, aimed at buyers who prioritise certification above openness.
- SafePal S1 (around $50). A budget air-gapped option; not the equal of the above on secure-element pedigree, but a real cold-storage step up from a phone app.
The pattern across this group is that air-gapping, no wireless radios and all communication over QR or microSD, has become the premium security signal of 2026, precisely because it removes the connectivity surface that supply-chain and remote attacks rely on. The trade-off is convenience, and for many buyers that is a fair price.
Tangem, reviewed: the seedless bet
Tangem is the outlier, and the most interesting onboarding story in the category. Instead of a screen-and-buttons gadget, it is a set of NFC smart cards. You tap a card to your phone to set up and to sign, the key is generated on the card’s Samsung S3D232A secure element and never leaves it, and there is no seed phrase to write down. Backup works by cloning the key to one or two backup cards during setup only. A two-card set runs $54.90, a three-card set $69.90, and a wearable Ring with two backup cards about $160. The cards have been audited by Kudelski, Riscure and, in 2026, Cure53.
The seedless design is a genuine philosophical choice, not a shortcut. There is no 24-word phrase for a phisher to steal or a burglar to photograph, and the tap-to-sign flow is the closest thing in crypto to using a contactless bank card. The costs are real too: the cards cannot receive firmware updates, and recovery depends on holding one of the physical backup cards rather than a portable phrase you can memorise or split.
That non-updatable design collided with Ledger’s Donjon lab this year. In a report published in July 2026, Donjon showed that a single nanosecond laser pulse aimed at one conditional check in the card’s SetPin handler could reset the access password without the existing password or a backup card, then drain the wallet. The chip was not broken; the fault flipped one branch in the firmware, and because the cards cannot be patched, the weakness cannot be closed by an update. The attack demands physical possession of the card, a lab kit worth roughly $250,000, about two hours per card, and specialist expertise, and Donjon reported no real-world losses.
Tangem’s co-founder Andrey Kurennykh pushed back hard in a public response, arguing that “LFI attacks are not scalable, and for everyday users, the practical risk is virtually non-existent,” and that “there have been no known real-world losses from laser fault-injection attacks on any hardware wallet to date.” He framed the non-updatable firmware as a deliberate trade-off of the seedless model rather than a defect. Both things are true at once: the finding is real, and the everyday risk to a normal holder is low.
Verdict: the best beginner and low-friction option on the market, and a smart choice for small-to-mid holdings where usability drives adoption. Buyers with large balances or a strong preference for standard, portable seed backups should weigh the seedless and non-updatable trade-offs before committing everything to cards.
Prices, specs and security models at a glance
The first table summarises the current field. Prices are typical US retail as of late August 2026 and exclude shipping and any local tax.
| Device | Price (USD) | Secure element | Open source | Connectivity |
|---|---|---|---|---|
| Ledger Nano S Plus | $79 | ST33 (EAL5+/6+) | No | USB-C |
| Ledger Flex | $249 | ST33 (EAL6+) | No | USB-C, NFC |
| Ledger Stax | $399 | ST33 (EAL6+) | No | USB-C, NFC, wireless |
| Trezor Safe 3 | ~$79 | Infineon Optiga (EAL6+) | Yes | USB-C |
| Trezor Safe 5 | ~$169 | Infineon Optiga (EAL6+) | Yes | USB-C |
| Trezor Safe 7 | $250 | TROPIC01 + Optiga (EAL6+) | Yes | USB-C, Bluetooth, Qi2 |
| Coldcard Q | $249 | Dual secure element | Yes | Air-gapped (QR, microSD, NFC) |
| Keystone 3 Pro | ~$149 | Triple Infineon | Yes | Air-gapped (QR only) |
| BitBox02 | ~$150 | Dual-chip | Yes | USB-C |
| Foundation Passport Core | $199 | Secure element | Yes | Air-gapped (QR, microSD) |
| Tangem (3-card) | $69.90 | Samsung S3D232A (EAL6+) | Partly | NFC |
Specs only tell half the story. The second table compares the security posture and the 2026 event that most shaped each brand’s reputation.
| Brand | Trust model | Air-gap | Signing verification | Defining 2026 event |
|---|---|---|---|---|
| Ledger | Closed, certified | No | Trusted display, clear-signing leader | Handed ERC-7730 clear signing to the Ethereum Foundation |
| Trezor | Open source | No (wired) | On-device display | Safe 7 launch; ShipMonk customer-data breach |
| Coldcard | Open source | Yes | On-device, air-gapped | ~$116M entropy exploit from a 2021 build flag |
| Keystone | Open source | Yes (QR) | On-device, QR | Triple-secure-element air-gapped design gains share |
| Tangem | Seedless | No (NFC) | On phone, keys on card | Donjon laser attack and public rebuttal |
| BitBox02 | Open source | No | On-device display | Steady Swiss dual-chip alternative |
Physical attacks: laser labs, wrench attacks and supply chains
The most sobering shift of 2026 is that the frontier of hardware wallet risk moved off the internet and into the physical world. It showed up in three distinct ways.
First, laboratory fault injection. Ledger’s Donjon lab spent the year demonstrating laser attacks against rival secure elements, resetting a Tangem card’s password and peeling back a layer of the Trezor Safe 7, both on EAL6+ chips. The reassuring detail is the cost: physical possession, a roughly $250,000 rig, hours of work per device, and no reported losses. These are nation-state-grade or targeted-theft techniques, not something aimed at a random buyer, and open vendors turned the findings into public fixes and disclosures rather than silent risk.
Second, and far more relevant to ordinary holders, wrench attacks. Chainalysis counted more than $30 million stolen through physical violence and coercion across at least 46 incidents in the first half of 2026, on pace to beat the $58 million full-year record set in 2025. France has become the epicentre, with reporting tying part of the surge to a 2024 tax-authority data leak that exposed high-net-worth holders, and home invasions linked to crypto jumped from a single case in the first half of 2025 to 20 a year later. No secure element defends against a threat to your safety, which is why duress PINs, passphrase-protected hidden wallets, and simple discretion about your holdings now belong in any serious security review.
Third, the supply chain. Your device can be perfect and your data can still leak around it. Ledger’s 2020 breach and 2023 Connect Kit compromise, and Trezor’s 2024 support-portal and 2026 ShipMonk incidents, all exposed customers to phishing without ever touching a private key. The defences are unglamorous but effective: buy only from the manufacturer or an authorised reseller, never a marketplace listing or a used unit; check tamper evidence and run the vendor’s genuine-device verification on first boot; and assume any message referencing a recent order could be a targeted phish built on leaked shipping data.
Blind signing: the bug reviews used to ignore
If entropy was the lesson of July, blind signing is the slow-burning issue underneath most large 2026 thefts. Blind signing is when your wallet asks you to approve a transaction it cannot decode into plain language, so you confirm a raw payload and hope it does what you think. Clear signing is the opposite: the device shows you the actual destination, amount and contract action on its own trusted screen before you approve.
This is not academic. The industry now openly blames blind signing for a large share of catastrophic losses, including the Bybit hack, where the final step was a human approving something they could not meaningfully read. The fix is an open standard, ERC-7730, that formats transactions into human-readable descriptions. Ledger initiated the standard and early tooling, and in May 2026 the Ethereum Foundation took stewardship of clear signing under its Trillion Dollar Security initiative, positioning a neutral party to maintain the registry that wallets draw on.
For a buyer, the practical takeaway is simple: the value of a hardware wallet’s screen is not the pixels, it is the ability to verify the address and amount on the device rather than on a computer that malware may control. Larger trusted displays (the Ledger Flex and Stax, Trezor Safe 7, Keystone 3 Pro and Coldcard Q) make that verification easier and less error-prone, and the habit of reading every field before you approve is worth more than any certification.
Which hardware wallet should you buy?
There is no single best hardware wallet, only the best fit for how you hold and what you fear. The matrix below maps common profiles to a sensible starting point.
| If you are… | Start with | Why |
|---|---|---|
| New to self-custody, small balance | Tangem or Ledger Nano S Plus | Lowest friction, strong chips, hard to misuse |
| A multi-asset DeFi user | Ledger Flex or Trezor Safe 5 | Broad support, good trusted display, clear signing |
| A Bitcoin-only holder | Coldcard Q or Foundation Passport | Air-gapped, multisig-ready, Bitcoin-focused |
| An open-source purist | Trezor Safe 7 or BitBox02 | Auditable firmware, auditable secure element on the Safe 7 |
| Committed to air-gapping | Keystone 3 Pro or Coldcard Q | No wireless radios, QR-only signing |
| Holding a large, long-term stack | Multisig across two or three vendors | No single chip, brand or bug can drain you |
That last row is the real conclusion of the year. The Coldcard hack, the Donjon lasers and every supply-chain leak point the same way: concentration is the risk. For meaningful sums, a 2-of-3 or 3-of-5 multisig that spreads keys across different vendors and different secure elements means a single firmware flaw, a single compromised device or a single stolen backup cannot move your coins. It is more work to set up, and for inheritance you will want clear instructions so your family can actually recover funds, but it turns the weaknesses in this guide into a strength: the failure of any one wallet becomes survivable.
Whatever you choose, generate the seed yourself on the device, verify it, keep a durable offline backup that no camera has seen, buy from the source, and read what you sign. The hardware in 2026 is excellent. Most losses still come from the human standing next to it.
Frequently Asked Questions
Are hardware wallets still safe after the Coldcard hack?
Yes, for the vast majority of users, but the incident changed what safe means. The Coldcard losses came from a firmware build flag that weakened random key generation, not from an attacker breaking a secure element, so the lesson is to keep firmware current, prefer devices with reproducible builds and independent review, and generate and verify your seed on an up-to-date device. Anyone who created a Coldcard seed between the March 2021 build and the patch should move funds to a newly generated seed. A properly generated seed on a maintained device remains far safer than leaving coins on an exchange or in a hot wallet.
What is the best hardware wallet for beginners in 2026?
For most newcomers, Tangem and the Ledger Nano S Plus are the easiest safe starting points. Tangem’s tap-to-sign cards remove the seed phrase entirely and are almost impossible to misconfigure, while the Nano S Plus offers a certified secure element and broad asset support at $79. The right pick depends on whether you prefer the seedless card model or a traditional device with a recovery phrase you control.
Do I need a hardware wallet if I already use a software wallet like MetaMask?
If you hold more than pocket change, yes. A software wallet keeps keys on an internet-connected device, which is convenient for daily use but exposed to malware and malicious approvals. A hardware wallet keeps the keys offline and makes you confirm transactions on a trusted screen. Many people use both: a software wallet for small, active balances and a hardware wallet for savings.
Can a hardware wallet be hacked remotely?
A well-designed hardware wallet is built so the private key never leaves the secure chip, which makes purely remote key theft very hard. The real remote risks are indirect: tricking you into approving a malicious transaction, phishing you into revealing your seed, or software supply-chain attacks like the 2023 Ledger Connect Kit incident. The Coldcard losses were remote in effect but caused by weak local key generation, not a live network intrusion. Physical attacks such as laser fault injection require the device in hand and a six-figure lab.
Ledger vs Trezor: which is more secure?
Neither is simply more secure; they make different bets. Ledger uses certified, closed-source secure elements and a large in-house red team, and argues that secrecy plus certification is stronger. Trezor’s Safe line uses open-source firmware and, on the Safe 7, an auditable secure element, arguing that transparency finds bugs faster. For most users both are safe when bought direct and kept updated. Open-source advocates prefer Trezor; buyers who value the broadest asset support and app polish often prefer Ledger.
Written by Yuki Tanaka, senior wallets and security editor at HOGE Wire.