Acentric Research is produced with AI tools under human review. The authors are not financial professionals, and nothing here is investment advice or a recommendation to buy or sell any security. AI analysis can be wrong; check every claim against its cited source.
This Brief is an initial scan, not a full investigation. The method and scores behind it are in AI × Quantum in the Landscape Report.
Overview
A quantum program is written as a circuit: a sequence of simple steps, called gates, applied to qubits. Writing a good program, and then rewriting it so that a particular machine can run it with as few gates as possible, is difficult, and the cost grows quickly with size. This Brief covers artificial intelligence (AI) used to discover quantum algorithms and to compile them, meaning to translate a program into the gate steps a machine runs.
What stands out is the distance between ambition and published results. The U.S. Department of Energy (DOE) asks for AI that can discover new quantum algorithms. [1] A review of the field describes AI methods that synthesize circuits for a few qubits and cut the count of expensive gates, and lists scaling limits. [2] A quantum software company raised $110 million in 2025, but its announcement describes its compilation technology as algorithmic and does not say that AI does the work. [3]
Current state
The federal ask. DOE’s Genesis Mission opportunity has a challenge titled “Discovering Quantum Algorithms with AI”. It states that discovering new quantum algorithms is exceptionally difficult because of the number of possible quantum operations, and that AI could automate and optimize design and translate high-level problem descriptions in natural language into executable circuits. A focus area asks for AI and formal verification methods to develop quantum algorithmic primitives and compilation tools. Another asks for AI to find efficient error correction for specific scientific applications. DOE says new algorithms would have broad applications in fusion, high energy physics, nuclear physics, materials science and chemistry. These are requests, not results, and the call states that scientific utility and a provable quantum advantage are what the algorithms must eventually show. [1]
What the research literature reports. A 2025 review in Nature Communications, written mostly by NVIDIA staff, divides compilation into four stages: synthesis, optimization, mapping and routing, and gate decomposition. It reports AI approaches for synthesis and optimization: reinforcement learning, in which software learns by trial and reward, and neural networks for synthesizing circuits, with one approach reaching up to three qubits; diffusion models, the technique behind many image generators, applied to circuits of 3 and, in a later result, up to 5 qubits; and a Google DeepMind reinforcement-learning method, AlphaTensor-Quantum, that reduces the count of costly T-gates; this Brief did not open that paper. [2] The review lists limits: the models need large computing resources, training often requires simulating quantum circuits on classical computers, which is exponentially hard, and graph-based models strain memory as qubits increase. [2]
Two terms help here. A T-gate is one kind of quantum gate, which the review calls notoriously expensive. And the review names transformer-based approaches, called GPT-QE and QAOA-GPT, among those that need large and diverse training sets that are costly to generate. [2]
Software companies. Classiq, an Israeli company, announced a $110 million Series C round on 2025-05-12, which it says brings total funding to $173 million. It describes its product as a high-level quantum development platform that automates quantum programming, and its technology as “proprietary algorithmic quantum circuit compilation”. The announcement does not say that AI performs the compilation. [3]
Public funding. The July 2026 list of Genesis Mission projects chosen for negotiation includes titles on AI-assisted application search and algorithm-level optimization at IBM’s research center, AI-enabled optimization of quantum circuit design at Argonne National Laboratory, and AI-driven discovery of early fault-tolerant computation primitives at the University of California, Davis. [4]
How far it has come. Laboratory and simulation results on small circuits; a commercial compilation product exists, but the source opened does not describe it as AI. This Brief did not search for other vendors’ tools, so AI compilation products from other companies are not covered.
Opportunities
DOE states that there is strong evidence that quantum computers and algorithms will offer solutions to computational problems with high impact for science. It also states that new algorithms could bring commercial applications in drug, material and chemical discovery. [1] The review points to smaller, cheaper circuits as the benefit of AI optimization, which matters most for today’s noisy machines. [2] The test is whether AI-found circuits keep their advantage at the sizes real applications need.
Challenges
- Physics and engineering. Compilation cost grows quickly with circuit size, and the AI synthesis results the review describes cover circuits of 3 to 5 qubits. [2]
- Training cost. Training often needs classical simulation of quantum circuits, which the review says limits scalability. [2]
- Evidence. The source opened for the software company does not attribute its compilation to AI. [3]
Data sources
- DOE Genesis Mission opportunity and the list of chosen projects, for federal programs in this area. [1][4]
- Research literature, through Europe PMC or arXiv, for compilation and algorithm-discovery results. [5]
- Company announcements, as company claims. [3]
- No registry covers quantum software or records which compilers use AI.
Changelog
- 2026-11-18: First published.
Sources
- US Department of Energy, “The Genesis Mission: Transforming Science and Energy with AI”, Notice of Request for Application DE-FOA-0003612, issued 2026-03-17, Amendment 000003 of 2026-04-20; challenges 7 and 8. science.osti.gov (accessed 2026-11-18)
- Yuri Alexeev and others, “Artificial intelligence for quantum computing”, Nature Communications 16, 10829, published 2025-12-02 (a review; the first author and most listed authors are at NVIDIA, which sells AI computing hardware). doi.org (full text read through the Europe PMC API, PMC12673084, accessed 2026-11-18)
- Classiq, “Classiq Raises $110M in Largest-Ever Quantum Software Funding Round”, press release dated 2025-05-12 (company statement). classiq.io (accessed 2026-11-18)
- US Department of Energy, “Energy Department announces first Genesis Mission projects to accelerate new era of AI-driven scientific discovery”, list of selections for award negotiations, posted 2026-07-22 (the list states that selection is not a commitment to issue an award or provide funding). energy.gov (accessed 2026-11-18)
- Europe PMC, REST API (search and full-text endpoints), queried 2026-11-18. ebi.ac.uk