Google has succeeded in cutting errors in the “magic state,” a core computational resource for quantum computers, to about one-fortieth of the previous level. The result improves the reliability of quantum operations, but a large gap remains to the level needed to crack bitcoin or RSA-2048 cryptography.

On Oct. 17 local time, blockchain outlet Cryptopolitan reported that Google’s Quantum AI research team said in a paper published in the journal Science that it achieved 99.99% logical accuracy for magic states using its own quantum processor, Willow.

At the center of the work is a technique Google calls “magic state cultivation.” Magic states are a key resource quantum computers need to perform complex operations, and the process of distilling error-mixed quantum states to obtain higher-accuracy states is important.

Quantum computers are prone to errors during computation because qubits, the unit of information, are sensitive to external noise. Even with error-correction technology, some high-difficulty operations require separate quantum states. Creating and refining magic states is cited as a major bottleneck that determines overall computational efficiency.

Google explained that generating magic states can account for up to 95% of a quantum computer’s total resources. The research showed the potential to improve the efficiency of quantum operations by reducing errors in that process. It said the same performance must be demonstrated in larger-scale systems to reach practical use.

Google said Craig Gidney, Noah Shutty and Cody Jones designed the method in 2024 by advancing existing theory. In the cultivation process, quantum states that include noise are placed in a limited region and then measured repeatedly. States found to have errors are discarded, and only those that pass a threshold are kept to secure more accurate magic states.

The researchers also reviewed the criteria used to interpret measurement results during experiments. Researcher Emma Rosenfeld (엠마 로젠펠드) repeatedly obtained results that were hard to predict, and the team traced the cause for months. It found that lab noise from temperature changes and drift, a long-term fluctuation in equipment performance, was distorting the reference values.

The team then widened the system’s control range and improved its measurement method. It also suggested the cultivation technique could be used not only to refine magic states but also as a new benchmarking tool to assess quantum system performance.

Google also demonstrated an integration process that moves refined quantum states from a “distance-3 Color Code” to a “distance-5 Surface Code.” Both are error-correction codes designed to protect quantum information from errors. It explained that linking different error-correction structures could help improve system reliability without simply increasing hardware scale.

Still, the result does not mean quantum computers’ code-breaking capability has reached a practical stage. The 99.99% logical accuracy cited in the paper implies an error rate of about 1 in 10,000 operations. By contrast, the magic states needed to crack RSA-2048 are estimated to require an error probability of about 1 in 10,000,000,000.

Comparing the two figures leaves about a 1,000,000-fold gap between required accuracy and current performance. The sharp reduction in magic-state errors is meaningful, but further advances are needed to reach levels required for actual decryption.

Research continues into potential threats from quantum computers to bitcoin’s cryptographic system. In a 2025 estimate, Craig Gidney suggested that fewer than 1,000,000 noisy qubits may be needed to crack RSA-2048. The magic state cultivation technique was cited as one of the technical elements needed to implement such quantum operations.

A separate Google Quantum AI study also produced an estimate that the number of physical qubits required to break the 256-bit elliptic-curve cryptography used by bitcoin could be fewer than 500,000. That is about 80% lower than previous estimates.

That study estimated that, if a sufficiently powerful quantum computer is secured, it could target the encryption protecting bitcoin transactions in about 9 minutes, with a success probability of 41%. It said the estimate is a theoretical calculation based on specific attack conditions and assumptions. Google’s current Willow processor is at about 105 physical qubits, far from the 500,000 estimated as needed.

As a result, it is hard to conclude that bitcoin or existing public-key cryptosystems will be neutralised immediately based on this research. For quantum computers to be used in actual decryption, they must solve multiple technical challenges at once, including increasing qubit counts as well as error correction, magic-state generation and operational stability.

Even so, calls are growing to prepare for a transition to post-quantum cryptography. Google has already set out a timeline aiming for 2029 to transition to post-quantum cryptography. Post-quantum cryptography is designed to withstand attacks not only from conventional computers but also from sufficiently advanced quantum computers.

In the bitcoin ecosystem, the BIP 360 proposal related to quantum-safe addresses is under discussion, but it was reported that a comprehensive plan to transition all users’ addresses and assets to a new cryptographic system has not yet been 마련되지 않은 것으로 전해졌다. An actual transition would require not only technical validation but also compatibility across wallets and networks, and user participation.

The research shows progress in technologies that reduce quantum-computing errors while also underscoring that a significant gap remains between lab achievements and real-world decryption capability. How quickly work proceeds on improving quantum processor performance as well as developing and preparing to transition to post-quantum cryptography is expected to be a key point to watch.