Insights/ Threat Landscape · Post-Quantum Cryptography, Part 1 of 4
The encryption you trust today will be broken
Breaking RSA-2048 was estimated to need 20 million qubits in 2019. A 2025 paper revised that to under one million. Why quantum is a different kind of threat.

The mathematics underpinning today's encryption was designed to be computationally unbreakable, for classical computers. Quantum computers operate on fundamentally different principles, and those principles make RSA and elliptic curve cryptography solvable. In 2019, breaking RSA-2048 was estimated to require 20 million physical qubits. A 2025 paper revised that to under one million. The timeline hasn't been extended. It's been compressed.
Data sourcesGidney, arXiv 2505.15917 (2025) · NIST FIPS 203/204/205 (August 2024) · ASD Information Security Manual (2024) · NSA CNSA 2.0 (2022) · Google Willow announcement (December 2024)Figures reflect publicly reported research as of October 2026. Findings evolve. Verify current studies before citing externally.
01Not just a faster computer
Quantum computing is a different kind of threat
Classical computers process information in bits: zero or one. Quantum computers use qubits, which can exist in superposition: representing zero, one, or any combination of both at once, until measured. Combined with entanglement, quantum computers can explore an enormous number of possible solutions in parallel rather than sequentially.
For most computing tasks, that provides no meaningful advantage. But one class of problem becomes tractable on a quantum computer that is intractable on any classical machine, no matter how many processors you run in parallel, and breaking RSA encryption is one of them.
In 1994, Peter Shor published an algorithm that solves integer factorisation on a quantum computer dramatically faster than any known classical method (in polynomial time, against the super-polynomial time of the best classical algorithm). The question isn't whether RSA falls. It's when.
noisy physical qubits estimated to break RSA-2048 in about 8 hours.
Gidney & Ekerå, 2019revised estimate (under a week of run time): a 20-fold reduction from algorithmic improvements, under the same hardware assumptions.
Gidney, arXiv 2505.15917, 2025qubits in IBM's Condor processor, the first to pass 1,000. Today's machines hold hundreds to a few thousand physical qubits, not hundreds of thousands.
IBM, 2023Google's Willow chip, announced December 2024, demonstrated meaningful progress on the error-correction problem that has historically limited quantum computing's practical utility. The estimates keep falling: 2026 preprints, not yet peer-reviewed, put RSA-2048 within reach of around 100,000 physical qubits on hardware designs that have not yet been demonstrated at scale, and Google Quantum AI estimated in March 2026 that 256-bit elliptic curve keys could fall to fewer than 500,000. No expert consensus exists on an exact date. Estimates range from 2030 at the early end to 2040 and beyond at the conservative end. What isn't in reasonable dispute: the trajectory is toward capability, not away from it.
02Not all encryption, equally
Which algorithms are at risk, and which aren't
The distinction matters for prioritisation. The urgent problem is asymmetric cryptography. Symmetric key length is a medium-term concern. Hash functions need the least immediate attention.
RSA & Elliptic Curve Cryptography (ECC)
Vulnerable to Shor's Algorithm on a sufficiently powerful quantum computer. Used pervasively: HTTPS, TLS, SSH, digital certificates, code signing, VPNs, enterprise authentication. All key sizes in current use, RSA-2048, RSA-4096, ECC-256, offer no meaningful additional protection.
AES symmetric encryption
Threatened by Grover's Algorithm, which roughly halves effective key length. On paper, AES-128 drops to about 64 bits of quantum security and AES-256 to about 128. In practice Grover's Algorithm parallelises poorly, so NIST still treats AES-128 as acceptable, while the NSA's CNSA 2.0 requires AES-256. Moving to AES-256 is the cautious choice.
Hash functions (SHA-256, SHA-3)
Not significantly threatened by currently known quantum algorithms. NIST considers SHA-256 adequate; the NSA's CNSA 2.0 requires SHA-384 or SHA-512 for national security systems.
03Not theoretical
The threat is active, not future
The assumption that there's no urgency because quantum computers don't yet exist misses something fundamental about the threat model. Data encrypted today with RSA or ECC, and captured today, can be stored and decrypted later, once the capability exists.
Harvest now, decrypt later: nation-state adversaries are already collecting encrypted data, health records, legal documents, financial data, intellectual property, for decryption once a cryptographically relevant quantum computer exists.
The NCSC, CISA and ASD have all issued guidance naming this risk explicitly. The NSA's CNSA 2.0 guidance, published in 2022, sets transition deadlines on the same logic: the data these systems protect will still need protection for decades after the systems themselves are retired. For any data requiring confidentiality beyond ten to fifteen years, the threat is present now.
04The standards exist now
The first NIST standards are final
In August 2024, after an eight-year evaluation, NIST finalised the first three post-quantum cryptographic standards. Organisations don't need to wait for further guidance to start planning. The algorithms are already available in major cryptographic libraries.
ML-KEM
Key encapsulation mechanism: the primary replacement for RSA and ECDH in protocols like TLS.
✓ Already in Chrome & CloudflareML-DSA
Digital signature algorithm: replaces RSA and ECDSA for code signing and authentication.
✓ Available in major crypto librariesSLH-DSA
Hash-based signature algorithm: a conservative choice where long-term trust is critical.
✓ Conservative, high-assurance useFN-DSA
A compact lattice-based signature scheme, useful in bandwidth-constrained environments.
⟳ Still in draft (October 2026)05Already moved
The Australian position
ASD's Information Security Manual recommends ceasing the use of traditional asymmetric cryptography, including RSA, Diffie-Hellman, ECDH and ECDSA, by the end of 2030. Its planning guidance sets the milestones: a refined transition plan by the end of 2026, the transition under way by the end of 2028 (critical systems and data first), and complete by the end of 2030.
The guidance is consistent with the broader Five Eyes posture: the NSA, NCSC, CCCS, and GCSB have all published aligned guidance directing organisations to begin transition planning now. Organisations that haven't started their cryptographic inventory are already behind the guidance curve: the transition isn't a single change, it requires identifying every system using quantum-vulnerable algorithms, prioritising by data sensitivity, and executing across often-complex infrastructure. That takes years, not months.
·Key takeaways
What to take into your next risk conversation
RSA and ECC, protecting most encrypted communications and authentication, are vulnerable to Shor's Algorithm. AES-256 and hash functions are not equivalently threatened.
The qubit requirement to break RSA-2048 dropped from 20 million to under one million, a 20-fold improvement from algorithmic advances, not hardware.
Data encrypted today is being collected by nation-state adversaries for future decryption. Anything requiring confidentiality beyond 10 years is exposed right now.
NIST published ML-KEM, ML-DSA, and SLH-DSA in August 2024. ASD targets the end of 2030. Organisations that haven't started their inventory are already behind.


