Quantum Software

Physics / mechanism

Quantum software encompasses the layers above quantum hardware: gate-level circuit compilers, error-correction code implementations, variational algorithms (VQE, QAOA), quantum simulation frameworks, and SDK/middleware stacks (Qiskit, Cirq, PennyLane, Braket). It operates by mapping computational problems onto qubit operations, managing decoherence budgets via transpilation and noise-aware scheduling. Current NISQ-era devices (50–1000+ physical qubits, gate fidelities 99.0–99.9% for two-qubit gates on leading superconducting and trapped-ion platforms) force software to do heavy lifting: error mitigation, circuit depth reduction, and classical co-processing. Fault-tolerant regimes—requiring ~1000 physical qubits per logical qubit under surface codes—remain 5–10 years from commercial utility. Software stacks therefore define practical capability more than raw qubit count today.

Competitive landscape

The primary competitive axis is classical HPC/GPU acceleration for the same target workloads (quantum chemistry, optimisation, ML inference). NVIDIA cuQuantum and GPU-accelerated tensor network simulators increasingly erode near-term quantum advantage claims. Adjacent segments: quantum-classical hybrid middleware (Quantinuum TKET, Q-CTRL), quantum-native compilers targeting specific hardware (IonQ, IBM), and domain-specific libraries for finance or pharma. The moat question is whether software locks to hardware vendor or abstracts across it.

Companies using

Connected ideas

Sources

Frontier (open questions)

• To be added.