Hardware wallet

The screen matters as much as the chip. A hardware wallet assumes the computer it plugs into is compromised, so the address and amount appear on a display malware cannot rewrite. Kraken Security Labs showed in January 2020 that a Trezor could still be opened with physical access and about 75 dollars of gear, which makes these devices a defense against remote theft rather than against a burglar.

How it works

The device creates the seed itself, using its own random number generator, and there is no interface for exporting it. Everything else follows from that: the recovery words are shown once on the device screen for you to copy by hand, and the keys derived from them stay behind a chip boundary for the rest of the device's life.

Signing is a conversation with strict rules. Your computer or phone builds an unsigned transaction, commonly as a PSBT under BIP-174, and sends it over. The device parses it, displays each destination address, each amount and the fee, and waits for a physical button press. Only then does it return signatures. The host never sees a key, and a host that lies about where the money is going gets contradicted by the display in your hand.

Hardware differs between vendors in ways worth knowing. Ledger devices are built around a certified secure element, a chip designed to resist physical extraction, with firmware that is not fully open source. Trezor's earlier models, the One and the Model T, use a general purpose microcontroller with completely open firmware, which is the trade-off Kraken's researchers exploited by glitching the chip's voltage. Neither approach is simply better: one resists a thief with a lab, the other resists a vendor with a secret.

Two features do the day-to-day protective work. A PIN limits guesses, wiping the device after a set number of wrong attempts, so a lost wallet is not an open wallet. A BIP-39 passphrase acts as an extra word that is never stored anywhere, producing a completely separate wallet from the same seed, which is what makes a coerced handover survivable.

Why this matters when you buy bitcoin

Buy the device new, from the manufacturer's own store or an authorized reseller, and never second hand. The classic attack needs no technical skill: a resold device arrives with a recovery card already filled in, the buyer follows the enclosed card instead of generating a new seed, and the seller sweeps the funds weeks later. A genuine device always makes you write down words it shows you for the first time.

Guard the delivery details as well. In July 2020 Ledger disclosed that its e-commerce database had been breached, exposing around 1 million email addresses and roughly 272,000 detailed records including names, postal addresses and phone numbers. The coins were never at risk, but customers received convincing phishing mail and, in some cases, physical threats. Shipping to a locker or a business address is a reasonable response.

Ledger Recover, announced in May 2023, changed how the category is judged. The service splits an encrypted copy of the seed between custodians for subscribers, and the backlash focused on the demonstration that firmware could be built to move seed material out at all. It is opt-in, and the useful takeaway is a general one: your threat model has to include the vendor's future firmware, which is why open builds and reproducible releases are worth paying attention to.

Use the device from the first purchase, not from the first big purchase. When you withdraw from an exchange, generate the receiving address on the device, verify it on the device screen, and paste it into the withdrawal form. Our reviews cover the models we consider worth buying, and the honest summary is that any current model from a serious vendor beats leaving coins on an exchange.

What the screen is actually for

Consider a routine withdrawal of 0.05 bitcoin. You copy an address from your wallet software, paste it into the exchange, and send. If the laptop is running a clipboard hijacker, the string you pasted is not the string you copied, and nothing on screen will tell you, because the malware controls the screen.

The hardware wallet breaks that loop in one specific way: it shows the receiving address on a display the laptop cannot draw on. Comparing the first four and last four characters takes three seconds and catches the substitution completely.

The same logic explains the largest theft the industry has recorded. The FBI attributed the theft of roughly 1.5 billion dollars from Bybit in February 2025 to North Korean actors who manipulated what the signing interface displayed, so the people approving the transfer authorized something other than what they believed they were seeing. Signing what a device shows you is only protective when the device is actually showing you the transaction, which is why blind approval of anything you cannot read on the hardware is the habit to avoid.

Hardware wallet vs software wallet

A hardware wallet is a computer with one job and one interface; a software wallet shares a machine with everything else you run. That difference is the entire security argument, and it is narrower than the marketing suggests: both derive keys from the same standards, produce the same addresses, and can restore each other's recovery phrases. What you buy is isolation and a trustworthy display, not stronger cryptography.

Hardware wallet vs cold storage

Owning a hardware wallet is not the same as practicing cold storage. The device is a tool and cold storage is a discipline: the key must be born on the device, never leave it, and every transaction must be verified on its screen. A device initialized from a phrase you generated on a laptop was compromised before you started, and a device you connect daily to click through prompts without reading them provides much less than its price suggests.