Quantum-Software Startup Quantoptima Publishes First Peer-Verified Benchmarks Showing QAOA Outperforming Classical Solvers on Industrial MaxCut Problems
Boston, MA – December 1, 2025 Quantoptima, a two-year-old quantum-software company spun out of MIT, today released the first independently reproduced optimization benchmarks that place the Quantum Approximate Optimization Algorithm (QAOA) ahead of state-of-the-art classical heuristics on industrially relevant MaxCut instances. The data—executed on IBM’s 133-qubit “Torino” processor and cross-checked on a cloud-based classical cluster—show a statistically significant advantage in approximation ratio on random d-regular graphs with 19–27 nodes, the sweet spot for supply-chain, circuit-layout and financial-index partitioning tasks.
“We took the extra step of open-sourcing our circuits and graph decks so the community could rerun the tests in under an hour,” said Dr. Maya Chen, Quantoptima co-founder and CEO. “Seeing a 5 % edge over the Goemans–Williamson algorithm at circuit depth p = 3, while holding latency flat, tells us that shallow QAOA is entering the zone of practical utility—not just academic curiosity.”
The benchmark suite, posted on arXiv and submitted to the ACM Transactions on Quantum Computing, compares five solvers across 250 weighted MaxCut instances derived from S&P 500 correlation matrices and semiconductor net-list graphs. All quantum jobs were executed with 1 000 shots and error-mitigation based on zero-noise extrapolation, keeping total compute cost under USD 4 per instance. Classical baselines included Gurobi 11 mixed-integer programming, simulated annealing and the Goemans–Williamson semi-definite relaxation; each was given a 5-second timeout to mirror near-real-time decision windows used by Quantoptima’s pilot customers.
“Optimization is the quiet work-horse of global industry—every 1 % improvement in network cuts or portfolio risk translates into billions in annual savings,” said Leo Marquez, senior analyst at Boston Consulting Group, citing the firm’s 2025 market survey that pegs addressable spend on combinatorial-optimization software at USD 7.8 billion by 2028. “If these numbers hold at 100-plus qubits, we are looking at the first credible threat to the classical optimization stack.”
Quantoptima’s error-mitigation layer reduced two-qubit gate infidelity from 0.83 % to 0.21 %, pushing median approximation ratio to 0.942 versus 0.891 for the best classical heuristic. Importantly, total job latency—including network queuing—stayed below 1.7 s, satisfying the sub-second SLA required by the company’s logistics partner, AmeriJet International, which this month began routing weekly cargo-loading problems through the startup’s cloud API.
About Quantoptima
Quantoptima delivers cloud-native quantum optimization software for logistics, finance and electronics design. Founded in 2023 and headquartered in Boston, the company has raised USD 14 million in seed financing led by The Engine (MIT) and is a member of the IBM Quantum Network. Its flagship product, Qaptain™, integrates with AWS Braket, Azure Quantum and IBM Quantum Network to provide drop-in MaxCut, QUBO and knapsack solvers that automatically select the best hardware-backend or classical fallback.
Media Contact
Sarha Al-Mansoori
Director of Corporate Communications
G42
Email: media@g42.a
Phone: +971 2555 0100
Website: www.g42.ai
