IBM says its next quantum chip will run 10,000 operations on 520 qubits, and it plans to link two into a 1,040-qubit computer in 2027
Customers report up to 19 times faster runs on IBM’s current Nighthawk chip. The new chip is a design goal, not a finished product. The planned 1,040-qubit system would still be well short of even the newest, much smaller estimates of what it would take to break today’s public-key encryption.
IBM said Nov. 11 that its next quantum processor, Nighthawk r3, is designed to accurately run 10,000 quantum operations across 520 qubits, and that it plans to connect two of the chips in 2027 into one system with 1,040 qubits.
Quantum computers could one day solve problems that ordinary computers can’t finish in practical time, such as simulating molecules and new materials. Running more operations without errors means longer, more complex calculations. Nighthawk r3 is a design target, and IBM did not say when it will be available.
Q-day and today’s encryption
As of Nov. 11, 2026, IBM’s planned chips remain well short of the scale researchers estimate would be needed to break today’s widely used public-key encryption, a moment often called Q-day. Those estimates have fallen sharply as researchers find more efficient ways to correct errors. Physical qubits are easily disturbed by noise, so many of them must work together, using error-correcting codes, to act as one reliable “logical” qubit. The counts below are physical qubits.
Estimates for factoring a 2048-bit RSA key, a common public-key size, have dropped from millions of qubits to tens of thousands:
- 2019: Craig Gidney of Google Quantum AI and Martin Ekerå estimated about 20 million noisy physical qubits running about eight hours, using a standard error-correction method called the surface code.
- 2025: Gidney cut that to less than a million noisy qubits running less than a week.
- February 2026: Researchers at Iceberg Quantum, a quantum computing company, described a design they call the Pinnacle Architecture, which uses a more efficient family of error-correcting codes called quantum low-density parity-check (LDPC) codes. They estimated fewer than 100,000 physical qubits with an expected run of about one month, or about 139,000 qubits to finish in a week.
- October 2026: Quantum hardware company Alice & Bob and the French research institute Inria estimated 22,593 “cat qubits,” a type of superconducting qubit built to resist one kind of error, running 3.14 days. With about 611,000 cat qubits, the same job could take about 1.25 hours, they said.
All of these are paper estimates, not demonstrations. Each assumes future machines with error rates around 0.1% per operation or lower, an error-correction cycle every microsecond and fully working error correction, which IBM and other companies have not reached. The 2026 estimates were posted as preprints on arXiv, a research-paper server. Even the smallest, about 22,600 cat qubits, is more than 10 times the 2,034 physical qubits planned for IBM’s Nighthawk r3, which is designed for 520 programmable qubits, and assumes a different kind of qubit. IBM plans a 1,040-qubit system in 2027.
The National Institute of Standards and Technology says encrypted data is already at risk from a “harvest now, decrypt later” threat, in which adversaries collect data now and try to decrypt it later if a powerful quantum computer arrives. In August 2024, NIST published three post-quantum standards meant to resist such attacks: FIPS 203 (ML-KEM) for establishing secret keys, and FIPS 204 (ML-DSA) and FIPS 205 (SLH-DSA) for digital signatures. NIST’s October 2026 transition report says it will deprecate quantum-vulnerable public-key methods, including RSA and elliptic-curve algorithms, after 2030 and disallow them under its standards after 2035.
What today’s chip does
- Nighthawk r2, introduced in September, runs accurate calculations of more than 7,500 operations on 120 qubits and can take up to 100,000 samples a second, IBM said.
- Customers including German energy company E.ON and quantum software firms Algorithmiq and Q-CTRL have run 34 billion quantum programs on it and report jobs up to 19 times faster than on IBM’s earlier Heron chip. E.ON reported a 15-times speedup for machine learning used to plan the power grid.
- An Algorithmiq simulation of a quantum material ran 8 times faster than before. Matching it on ordinary computers would take an estimated 240 times more memory, IBM said.
How it works
A qubit is the basic unit of a quantum computer, the counterpart of a bit in an ordinary one. Each run of a quantum program returns one sample answer, so useful jobs need millions to billions of runs, IBM said. That is why the company stresses speed as well as accuracy.
Nighthawk r3 will have 2,034 physical qubits, 520 of them programmable, arranged in a square grid. It adds 10 chip-to-chip links that IBM calls l-couplers, meant to join chips into bigger machines. This year IBM linked earlier Nighthawk chips with four l-couplers and passed quantum information between them with 99.3% accuracy, short of its 99.9% goal. It also joined two of the super-cold refrigerators that house the chips into one system, cooled below 15 millikelvin, a tiny fraction of a degree above absolute zero.
IBM is also linking its quantum computers with ordinary supercomputers on its cloud service, IBM Quantum Platform, so users can split work between the two.
Limits
The speedups are reported by IBM and its customers and compare IBM’s own chips, not quantum computers against the best ordinary supercomputers. Today’s quantum chips still make errors. IBM describes Nighthawk as a step toward “fault tolerance,” meaning machines that can correct their own errors, which it has not reached.
Source: IBM news release, Nov. 11, 2026. NIST, “What Is Post-Quantum Cryptography?”. NIST news release on the post-quantum transition, Oct. 27, 2026. NIST IR 8547, Transition to Post-Quantum Cryptography Standards. NIST FIPS 203, FIPS 204 and FIPS 205. Craig Gidney, “How to factor 2048 bit RSA integers with less than a million noisy qubits,” arXiv:2505.15917. Webster et al., “The Pinnacle Architecture,” arXiv:2602.11457. “Shor’s algorithm compiled on cat qubits: 22 593 qubits for RSA-2048 in 3.14 days,” arXiv:2610.37839. Alice & Bob blog on the cat-qubit estimate.
- IBM: IBM Quantum Nighthawk Demonstrates New Levels of Speed and Quality, with Plans to Scale Beyond 1,000 Qubits in 2027 (Nov. 11, 2026)
- NIST: What Is Post-Quantum Cryptography? (updated Feb. 27, 2026)
- NIST: NIST Report Provides Guidelines for Transition to Post-Quantum Cryptography Standards (Oct. 27, 2026)
- NIST IR 8547: Transition to Post-Quantum Cryptography Standards (final, Oct. 27, 2026)
- NIST FIPS 203: Module-Lattice-Based Key-Encapsulation Mechanism Standard (ML-KEM) (final, Aug. 13, 2024)
- NIST FIPS 204: Module-Lattice-Based Digital Signature Standard (ML-DSA) (final, Aug. 13, 2024)
- NIST FIPS 205: Stateless Hash-Based Digital Signature Standard (SLH-DSA) (final, Aug. 13, 2024)
- Craig Gidney, How to factor 2048 bit RSA integers with less than a million noisy qubits, arXiv:2505.15917 (2025)
- Webster et al. (Iceberg Quantum), The Pinnacle Architecture: Reducing the cost of breaking RSA-2048 to 100 000 physical qubits using quantum LDPC codes, arXiv:2602.11457 (v2, May 5, 2026)
- Shor's algorithm compiled on cat qubits: 22 593 qubits for RSA-2048 in 3.14 days, arXiv:2610.37839 (Oct. 27, 2026)
- Alice & Bob: Compiling Shor’s Algorithm on Cat Qubits: 22 593 Qubits to Break RSA-2048 in 3.14 Days
- Iceberg Quantum: Pinnacle 1.2