Fault-tolerant quantum computing (FTQC) is the regime where quantum computations are performed on error-corrected logical qubits with error rates low enough to execute arbitrarily long algorithms reliably. In an FTQC system, errors are continuously detected and corrected during computation, and the error correction circuits themselves are designed so that faults in the correction process do not propagate uncontrollably. This is the ultimate goal of quantum computing engineering.
The key requirement for fault tolerance is that physical error rates must be below the error threshold of the chosen QEC code. For the surface code, this threshold is approximately 1%. Once below threshold, logical error rates decrease exponentially with code distance, enabling any desired logical error rate at the cost of more physical qubits. A fault-tolerant computation also requires that all logical gate operations — including non-Clifford gates like the T gate — are performed in a fault-tolerant manner, typically through techniques like magic state distillation and gate teleportation.
The timeline for practically useful FTQC remains debated. Optimistic projections place the first useful fault-tolerant computations (molecular simulation for drug discovery, materials science) in the late 2020s, while more conservative estimates suggest the early-to-mid 2030s. Key milestones include Google's below-threshold surface code demonstration (2024), Microsoft's topological qubit announcement (2025), and multiple companies' roadmaps for 100+ logical qubit systems by 2028-2030. The transition from NISQ to FTQC is not binary — hybrid approaches using partial error correction and error mitigation will likely bridge the gap.