Coherence time is one of the most critical performance metrics for any qubit technology, determining how many quantum operations can be performed before the qubit's quantum information is lost to the environment. It is characterized by two related time constants: T1 (energy relaxation time), which measures how long before an excited qubit spontaneously decays to its ground state, and T2 (dephasing time), which measures how long the phase relationship in a superposition state remains well-defined.
The ratio of coherence time to gate operation time determines the number of operations possible before errors dominate. A superconducting transmon qubit with T2 of 200 microseconds and single-qubit gate times of 20 nanoseconds can perform roughly 10,000 operations before decoherence — but two-qubit gates are slower (200-600 ns), reducing the effective circuit depth to hundreds of layers. Trapped-ion qubits have coherence times of seconds but gate times of microseconds, yielding a similar operations-per-coherence window. This ratio directly constrains algorithm design in the NISQ era.
Improving coherence times is a relentless engineering effort across all qubit modalities. For superconducting qubits, progress comes from better materials (tantalum replacing niobium, reducing oxide defects), improved fabrication processes, better electromagnetic filtering, and novel qubit designs. Coherence times for transmon qubits have improved roughly 10x per decade since their invention, from microseconds in the late 2000s to hundreds of microseconds today.