Technology

Quantum computers can't break Bitcoin or RSA today, but the estimates keep shrinking

Resource estimates for an attack have fallen sharply, and NIST and Bitcoin developers are building defenses before any machine is large enough.

12 min read
Blue-lit 3D rendering of a square, chip-like quantum processor surrounded by intricate metal framework and striped circuit structures.
Image by TheDigitalArtist from Pixabay

No quantum computer can break RSA-2048, elliptic-curve keys or Bitcoin today, but estimates of the hardware needed have dropped about 20-fold since 2019. NIST has set 2030 and 2035 deadlines, and Bitcoin developers have proposals that are not yet activated. DeepMind's RSA-1024 findings are a separate, classical result.

No quantum computer can break Bitcoin's signatures or the RSA encryption behind much of the internet today. Every figure describing such an attack is an estimate of what a future machine would need, not a demonstrated break. Those estimates have fallen steeply, though: from about 20 million noisy qubits to factor a 2048-bit RSA key in a 2019 analysis by Gidney and Ekerå to fewer than 1 million in a May 2025 preprint by Gidney, a Google researcher.

The shrinking estimates explain the current urgency. According to secondary reports, NIST finalized a schedule on October 27, 2026 that deprecates RSA and elliptic-curve algorithms after 2030 and disallows them after 2035. Bitcoin developers, meanwhile, are debating how to protect coins whose public keys are already visible on the blockchain, including a November 4 proposal that would make quantum thieves risk their own funds.

What a quantum computer actually threatens

The threat centers on one quantum method, Shor's algorithm, which could break the public-key schemes that protect most secure web traffic and digital signatures: RSA and elliptic-curve cryptography. Public-key schemes let two parties who have never met agree on secrets and verify who signed a message. Break them, and both confidentiality and authenticity are at risk.

Hash functions are a different matter. According to a Cointelegraph analysis carried by TradingView, quantum risk to Bitcoin falls mainly on exposed public keys rather than on the SHA-256 hashing that underpins mining. The sources gathered for this article do not examine symmetric ciphers such as AES, so this piece makes no claim about them.

Two kinds of qubit

A physical qubit is the raw, noisy hardware element. A logical qubit is an error-corrected qubit assembled from many physical ones. Attack estimates are usually stated in physical qubits, while hardware milestones are increasingly reported in logical qubits, and the two numbers are not interchangeable.

Harvest now, decrypt later

The informal term "Q-Day" refers to the day a quantum computer can break public-key cryptography. Security planners also worry about a slower threat. An adversary can record encrypted traffic now and decrypt it once quantum capabilities mature, a threat model described in an arXiv survey of post-quantum standards. For secrets that must stay private for decades, the deadline arrives before any such machine exists.

How big a machine an attack would need

Resource estimates calculate how many qubits and how much time a given algorithm would need on hardware with assumed error rates. In his May 2025 preprint (arXiv 2505.15917), Gidney estimated that a 2048-bit RSA key could be factored in under a week with fewer than 1 million noisy qubits. The 2019 estimate by Gidney and Ekerå was about 20 million qubits and eight hours. The brief did not confirm that the 2025 paper has been peer reviewed.

Estimated physical qubits needed to factor RSA-2048, 2012-2026 (estimates, not demonstrations)
Year Source Physical qubits Runtime Status
2012 Earlier work, as summarized by postquantum.com About 1 billion Not given in source Secondary summary
2019 Gidney and Ekerå About 20 million 8 hours Cited in Gidney's 2025 paper
2025 Gidney (Google) Fewer than 1 million Under 1 week Preprint
2026 "Pinnacle" architecture About 98,000 (alternative: 1 million) About 1 month (alternative: about 10 hours) Unverified; reported by an industry site

The drop from 2019 to 2025 came from better algorithms and error-correction methods rather than a leap in hardware. That distinction matters because the figures depend on assumptions about error rates, gate speeds and decoder performance. Gidney himself saw no way to cut the qubit count by another order of magnitude without changing the physical assumptions, and said he could not plausibly claim that 100,000 noisy qubits would suffice.

The unverified "Pinnacle" claim

A 2026 design called Pinnacle claims to need under 100,000 qubits, reportedly 98,000 qubits running for about a month or 1 million running for about 10 hours, according to the industry site postquantum.com. The same report says Gidney offered the most technically pointed assessment of it. The original paper, its peer-review status and the content of the critique were not available for this article, so the claim should be treated as unverified.

The missing Bitcoin number

Bitcoin relies on elliptic-curve signatures, not RSA, so RSA-2048 figures do not transfer directly. The Pinnacle comparison table lists a 2023 estimate for breaking 256-bit elliptic-curve cryptography using cat qubits (Gouzien 2023). No current, verified resource estimate for secp256k1, the curve Bitcoin uses, was found for this article. That is the most important missing number for the Bitcoin question.

Where the hardware stands

Recent milestones are real but come mostly from company press releases. On July 30, 2026, IBM and the University of Chicago reported running 70 logical qubits through 2,415 logical two-qubit operations and 468 logical T gates, with effective logical error rates 10 times lower than physical ones. The underlying paper was not retrieved.

On October 19, 2026, Quantinuum reported 99.9995% logical fidelity on its Helios machine, equal to 5 errors per million operations, with logical errors falling by a factor of 12 for each step up in code distance. These claims are the company's own and were not independently verified, though an outside Sandia researcher called the result a significant milestone. Quantinuum also targets fault-tolerant computing by 2029, which is a vendor roadmap rather than an independent forecast.

Why logical-qubit counts disagree

Headline counts are hard to compare. One tracker, entangledfuture.com, lists 8 as the most logical qubits held under error correction with no runs discarded, by Quantinuum. A second aggregator says QuEra holds up to 96 verified logical qubits, citing Nature in 2026. Both are low-quality aggregators, and the gaps likely reflect different rules on error detection versus correction, post-selection and circuit depth.

Scale remains the obstacle. An arXiv review of neutral-atom computing, which is not peer reviewed, says going from tens to thousands or tens of thousands of logical qubits still faces enormous challenges. For comparison, a LinkedIn analysis (a secondary source, to be checked against the paper) says Gidney's design uses 1,399 logical qubits running for days, far above the tens demonstrated so far.

Bitcoin's exposure and the proposed fixes

What is exposed

The weak point is the public key, not mining. Some early Bitcoin transactions have exposed their public keys since inception, so coins linked to Satoshi Nakamoto are potential targets, according to the Cointelegraph analysis. CoinCentral reports that about 7.1 million BTC, roughly 35% of supply, sits in wallets with exposed public keys. That figure comes from a single source with no stated methodology, although another outlet independently describes roughly one-third of Bitcoin as sitting in quantum-vulnerable addresses.

Reports distinguish two exposure types. Long exposure covers coins whose public keys are already on-chain and could in principle be attacked slowly. Short exposure covers a spend sitting unconfirmed in the mempool, which would require a very fast attack. Several outlets treat short exposure as harder, but no primary source quantifying it was found.

BIP-360 and BIP-361

BIP-360, authored by Hunter Beast of MARA, Ethan Heilman and Isabel Foxen Duke, was merged into the BIP repository as a Draft on February 11, 2026. A merge does not signal endorsement or guarantee activation. As a Forbes contributor describes it, the design builds on Taproot but removes the key-path spend, which exposes a public key on-chain.

It is explicitly partial. BIP-360 does not stop the short-exposure attack, and Heilman called it "step one." Full safety also requires a post-quantum signature algorithm, which version 0.8.0 deferred to a companion proposal in July 2025. One explainer says BIP-360 supports specific post-quantum schemes directly, but that appears to describe the older P2QRH version. BIP-360 also leaves untouched the roughly one-third of Bitcoin already in vulnerable addresses.

BIP-361, published in April 2026, proposes a phased sunset of old vulnerable signatures on an aggressive timeline, according to Quantum Zeitgeist, a lower-quality outlet. Neither proposal has been activated, and the community has not agreed on a post-quantum signature scheme. BTQ Technologies, a quantum company with a commercial interest, said it ran BIP-360 on its own testnet in March 2026.

  1. June 2024: the P2QRH design is first posted on Delving Bitcoin.
  2. July 2025: version 0.8.0 defers post-quantum signatures.
  3. February 11, 2026: BIP-360 is merged as a Draft.
  4. March 2026: BTQ Technologies runs it on a testnet.
  5. April 2026: BIP-361 is published.
  6. July 2026: a first Bitcoin Core regtest implementation is reported by a single source, spark.money.
  7. November 4, 2026: Cory Fields posts the landmines idea.

The "quantum landmines" thought experiment

Cory Fields, a Bitcoin Core developer described as a researcher at MIT's Digital Currency Initiative, outlined three thought experiments on the bitcoin-dev mailing list on November 4, according to CoinDesk. In the first, a thief stealing 1 BTC would have to post another 1 BTC from a quantum-safe wallet as collateral, with both locked for a waiting period such as 30 days. In that window the rightful owner could reveal previously hidden proof of ownership, recover the coins and claim the thief's collateral. Thieves could not tell which addresses hold such traps before committing funds.

The other variants are Staked Auctions and Sats Auctions. The Staked Auction targets coins whose keys may be lost, which landmines cannot protect. Fields noted a soft fork could implement these ideas, but key design questions remain open.

The limits are significant. By one outlet's account, an owner with savings in an old 10 BTC wallet would need 10 more BTC, already in quantum-secure form, to move their funds. Reports also raise questions about security and property rights, and no independent developer review had surfaced. Every report found traces back to CoinDesk or to Fields's post, so this is effectively one underlying source, and the idea is neither deployed nor formalized.

A skeptical view

Not everyone expects catastrophe. On-chain analyst James Check argued earlier this year that quantum computing would not trigger a catastrophic market collapse, and that even the theft and sale of Satoshi-era coins would not threaten the network's survival. That is an analyst's opinion, not a finding, and it sits alongside the developers' view that preparation should begin early.

Everyday encryption: the official clock

For the rest of the internet, the standards already exist. An arXiv survey reports that NIST published its first post-quantum standards in August 2024: FIPS 203 (ML-KEM), FIPS 204 (ML-DSA) and FIPS 205 (SLH-DSA). The NIST pages themselves were not retrieved for this article.

NIST's transition document, IR 8547, tightened between draft and final. The November 2024 draft deprecated 112-bit algorithms such as RSA-2048 after 2030 and disallowed all quantum-vulnerable public-key algorithms after 2035, but spared algorithms at 128 bits and above until 2035. According to three secondary sources (CryptoNext, pk-sharma.com and postquantum.com), the final version released October 27, 2026 deprecates RSA, ECDSA, EdDSA, Diffie-Hellman and MQV after 2030 and disallows them after 2035, regardless of key size.

NIST IR 8547 treatment of RSA and elliptic-curve algorithms at 128-bit strength or more, draft (2024) versus final (2026)
Version Deprecated Disallowed
Draft (November 2024) Not deprecated After 2035
Final (October 27, 2026) After 2030 After 2035

The schedule applies within NIST standards and guidelines, which govern FIPS 140 validation of cryptographic modules. It is not a law for the private sector. Gidney's own view is instructive: he backs deprecating vulnerable systems after 2030 and disallowing them after 2035, not because he expects big enough machines by 2030, but because he prefers security that does not depend on progress being slow.

The DeepMind RSA-1024 result is a different story

On November 6, 2026, Google DeepMind announced that human-supervised AI agents had found two weaknesses in legacy 1024-bit RSA, described in working papers dated November 5 and 6. This is classical cryptanalysis of an obsolete key size, not a quantum result. It does not involve Bitcoin, which uses elliptic-curve signatures rather than RSA, although no source in the research states that last point explicitly.

According to secondary summaries, one weakness concerns forging signatures. A theoretical estimate puts the improvement at about 1,600 times over the previous best method for 1024-bit keys, in cases where the attacker controls a small part of the signed data, such as a 32-byte fingerprint. In a measured reproduction, two agents with about 20 short messages from a human reran a recent 1024-bit signature attack in under four days, using about 3.5 times less CPU time than the academic team. The first root took 391 core-years.

The second weakness is batch factoring. A new algorithm reportedly cuts the average cost of each additional 1024-bit key by about 30 times relative to the multiple number field sieve. Another summary gives about 1/58 of a general number field sieve run per extra key, so the baselines differ. Reports also give different titles for the papers.

DeepMind's Pushmeet Kohli said, "In practice, these results only affect obsolete 1024-bit RSA systems," and that modern protocols using 2048-bit or larger keys remain secure. One report quotes the authors as saying even 4096-bit RSA does not appear to meet a 128-bit security level in this specific attack model. That is a narrow technical statement and not a claim that RSA is broken. The papers are vendor-published working papers, not peer reviewed, and no independent expert reaction was found.

What to watch

The honest answer is that nobody can currently date Q-Day. Demonstrated machines hold tens of logical qubits, while the estimated need is about 1,400 logical qubits running for days. Yet the estimates have fallen about 20-fold in six years, and Gidney himself says "attacks always get better."

Several markers will show whether the gap is closing: independent replication of fault-tolerant computation at hundreds to thousands of logical qubits, a verified resource estimate for Bitcoin's curve, and whether BIP-360 and a post-quantum signature scheme win broad agreement. The hardest Bitcoin question may be governance rather than math: how to treat old and lost coins that no upgrade can protect.

Frequently asked questions

Can a quantum computer break Bitcoin today?

No. All sources frame attacks as future estimates, and demonstrated machines are far smaller than the estimated need. The risk centers on exposed public keys rather than Bitcoin's SHA-256 hashing.

How many qubits would it take to break RSA-2048?

A May 2025 Gidney preprint estimates fewer than 1 million noisy qubits and under a week, down from about 20 million qubits and eight hours in a 2019 estimate. A 2026 design claiming under 100,000 qubits is unverified.

What is the NIST deadline for replacing RSA and elliptic-curve encryption?

According to three secondary sources, the final IR 8547 deprecates RSA, ECDSA, EdDSA, Diffie-Hellman and MQV after 2030 and disallows them after 2035, at all key sizes. The schedule applies to NIST standards and guidelines, not to private-sector law.

Did DeepMind's AI break RSA encryption?

No. DeepMind reported classical weaknesses in legacy 1024-bit RSA, and its Pushmeet Kohli said the results only affect obsolete 1024-bit systems. Modern 2048-bit or larger keys remain secure, and the work is unrelated to quantum computing and to Bitcoin.

What are Bitcoin developers doing about quantum computers?

BIP-360, a Draft merged in February 2026, removes the key-path spend that exposes a public key, but it is only a first step. BIP-361 proposes phasing out old signatures. Neither is activated, and Cory Fields's landmines idea is a mailing-list thought experiment.

Sources (24)

  1. Bitcoin developer proposes 'landmines' to make quantum thieves lose their own BTC , CoinDesk (2026-11-05)
  2. Bitcoin Developer Proposes Quantum Landmines to Make BTC Thieves Risk Their Own Funds , Cryptometer
  3. Bitcoin Developer Proposes Three Defenses Against Quantum Attacks , KuCoin News
  4. Bitcoin Developer Floats Collateral Trap That Could Cost Quantum Thieves Their Own BTC , CoinCentral
  5. How to factor 2048 bit RSA integers with less than a million noisy qubits , arXiv
  6. Reading Gidney's 2025 RSA-2048 resource estimate , XDC Network Labs
  7. Pinnacle Architecture: 100,000 Qubits to Break RSA-2048, but at What Cost? , postquantum.com
  8. Q-Day Revisited - RSA-2048 Broken by 2030: Detailed Analysis , LinkedIn
  9. IBM and The University of Chicago Demonstrate Quantum Advantage, Establishing Trusted Quantum Computation on Logical Circuits , IBM Newsroom (2026-07-30)
  10. Quantinuum Breaks Error-Correction World Record by 10x, Validating Its Leading Path to Fault-Tolerant Quantum Computing by 2029 , PR Newswire (2026-10-19)
  11. Logical Qubit Record: 8 Held by Quantinuum (2026) , entangledfuture.com
  12. Neutral Atom Quantum Computing: Principles, Routes, Progress, and Challenges , arXiv
  13. Bitcoin's quantum upgrade path: What BIP-360 changes and what it does not , Cointelegraph via TradingView News
  14. Bitcoin Took Its First Step Against Quantum Computers , Forbes (2026-02-23)
  15. BIP-360: Bitcoin's Quantum-Resistant Address Format Explained , spark.money
  16. Quantum Resistant Bitcoin, The Complete 2026 Guide , Quantum Zeitgeist
  17. Bitcoin Developer Proposes Quantum Landmines Report: BTQ Technologies Implements BIP 360 Quantum-Resistant Bitcoin Transactions on Testnet , The Quantum Insider (2026-03-20)
  18. Study of Post Quantum status of Widely Used Protocols , arXiv
  19. NIST IR 8547 Final: What Changed From the Draft , postquantum.com
  20. NIST's final IR 8547 deprecates all RSA and ECC after 2030 , pk-sharma.com
  21. New NIST Report: 2030 Becomes the Deadline for Post-Quantum Migration , CryptoNext Security
  22. DeepMind AI Agents Find 2 Weaknesses in Legacy 1024-Bit RSA , FourWeekMBA
  23. Google DeepMind's AI Agents Find Two Weak Spots in Legacy 1024-Bit RSA , WebProNews
  24. Google DeepMind announces AI-assisted research into two weaknesses in legacy 1024-bit RSA , Digg (2026-11-06)

Share this article

Back to list