Anthropic says its Claude AI model has designed more efficient quantum circuits across eight computing tasks, potentially helping researchers reduce the resources needed to run complex quantum algorithms.
In research published on October 16, 2026, the company described how Claude developed circuit designs for chemistry, optimisation, quantum simulation, number theory and numerical computing. Several designs reportedly improved on previously published approaches.
How Claude improved quantum circuit designs
Anthropic gave Claude plain-language descriptions of eight computational tasks and used an automated evaluator to test the proposed circuits for correctness and resource efficiency. Claude then used the feedback to revise its designs over repeated rounds without human input between attempts.
Quantum circuits use sequences of operations called gates to process information stored in quantum bits, or qubits. Reducing the number of gates and qubits can help make complex calculations more practical on future quantum hardware.
One design used 17 times fewer gates
One result involved modular inversion, a computationally expensive operation used in elliptic-curve calculations. Anthropic reported that Claude developed a design using 12% more qubits but requiring 17 times fewer gates than the smallest previously published design.
The finding illustrates a trade-off in quantum computing: using additional qubits can sometimes reduce the number of operations needed to complete a calculation.
AI reduces the cost of a quantum decoder
Claude also redesigned a computationally expensive step in a quantum optimisation method. According to Anthropic, the new design reduced the decoder’s gate count by a factor of 12 compared with an earlier construction developed by Google researchers.
These results suggest that AI-assisted quantum circuit design could help researchers identify mathematical improvements that reduce computational requirements.
Why efficient quantum circuits matter
Quantum computers could eventually help scientists investigate complex molecules, develop improved materials and explore industrial catalysts. However, today’s quantum hardware remains limited by qubit availability and sensitivity to errors.
More efficient circuits could help future machines perform useful calculations with fewer resources. Anthropic’s results remain specific to the tasks evaluated and do not demonstrate that current quantum computers can solve these problems at scale.
The research highlights a potential role for AI in scientific discovery, allowing models to explore mathematical approaches, test designs and improve specialised computational methods.
Read more: Anthropic has also warned about self-replicating AI threats.
