Google says a breakthrough involving its Willow quantum chip has reduced errors in preparing specialised quantum states needed for advanced calculations, addressing one of the hardware challenges that would have to be solved before a future machine could threaten bitcoin wallets.
According to CoinDesk, the experiment, published this week in Science, reduced errors in preparing so-called “magic states” by about 40 times. The researchers reported roughly 99.99% fidelity among the states that passed inspection. However, only about 8% of preparation attempts succeeded, and the system remains far from the scale needed for a practical attack on Bitcoin.
The result does not mean that attacking bitcoin is now 40 times cheaper. Instead, it demonstrates progress in an engineering technique that could help future quantum computers produce the reliable states required to run a large calculation for long enough to matter.
Why exposed bitcoin keys matter
Bitcoin transactions are authorised with a secret key and verified using a corresponding public key. Recovering the secret key from the public key is considered impractical with conventional computers. A sufficiently capable quantum computer running an algorithm developed by mathematician Peter Shor could, in theory, reverse that relationship and allow an attacker to sign transactions as the owner.
The risk differs depending on how a bitcoin address is used. Some payment formats keep the public key hidden behind a cryptographic hash until the owner spends the coins. In that situation, an attacker would need to wait for the key to become visible and then work quickly enough to interfere with the transaction.
Other holdings have had their public keys exposed since they were mined. This includes some of bitcoin’s oldest coins. A recovery phrase stored offline would not solve that particular problem, because a future attacker would be targeting information already available on the blockchain rather than the phrase in the owner’s possession.
Lost coins create an additional complication. Bitcoin could adopt stronger protections, but moving existing holdings into those protections would still require the original keys. If those keys are unavailable, the coins cannot be transferred, while an exposed public key would remain visible to a future quantum attacker.
Some dormant bitcoin has moved as discussion of the issue has intensified. On October 7, 100.02 BTC that had not moved since July 2010 were transferred after more than 16 years, with a value of about $8.5 million. In August, six wallets inactive since 2011 to 2014 moved 553.59 BTC, worth approximately $40 million at the time. Those transactions alone do not show that owners were responding to quantum risks; they could also reflect a wallet change, portfolio reorganisation or preparations to sell.
The hardware challenge behind a quantum attack
Quantum computers use qubits, which are highly sensitive to heat, vibration and other disturbances. Errors become harder to manage as systems grow, so engineers combine physical qubits into error-detecting groups. These groups can function as more reliable logical qubits.
Logical qubits also need specialised resources to perform certain operations. Magic states are prepared quantum states that are consumed during calculations, meaning a machine must continually produce new ones while its other components perform the main computation.
Google’s approach, known as cultivation, begins with a less reliable state, checks it for errors and discards attempts that fail. The reported 99.99% fidelity applies to the states that survived those checks. The process retained about 8% of attempts during preparation and measurement, meaning roughly eight out of every 100 tries made it through.
That distinction is important. Producing a highly accurate state is useful only if the process can be scaled, operated in real time and integrated with stronger error-correction systems. Google’s findings address part of that challenge, but the researchers said additional work is needed to make the states useful in longer calculations and to move them into more robust error-correcting arrangements.
What the finding means for bitcoin
Google estimated in March that a future superconducting quantum computer with fewer than 500,000 physical qubits could break the wallet cryptography used by Bitcoin and Ethereum under a specific set of hardware assumptions. That estimate already assumed efficient production of magic states.
The same work modelled an attack lasting about nine minutes after a public key became available. It estimated that the machine would need approximately 500,000 sufficiently reliable magic states every second. The new Willow result is relevant because it tests an engineering method for supplying that ongoing demand.
Other recent research has focused on reducing the mathematical workload of a key calculation in a potential attack. That work addressed one part of the problem rather than the complete attack or the cost of building and operating the required hardware. Google’s cultivation method targets the supply of quantum “fuel” needed while the calculation runs.
The advances should not be combined as if each produced a separate reduction in attack costs. The March estimate already included assumptions about efficient magic-state production, while the latest experiment demonstrates only part of a possible implementation. It therefore represents progress toward a future capability, not evidence that bitcoin wallets can currently be cracked.
Market picture
Moneyiar market data at 21:29 on October 17, 2026, showed Bitcoin at $81,188.41, down 0.54% over 24 hours and 2.15% over seven days. Quant was $194.34, down 4.61% in 24 hours and 19.17% over seven days. Akash Network stood at $0.6913, down 0.72% and 6.57% over the same periods. USD.AI was $0.04323, down 6.11% over 24 hours and 35.98% over seven days.
