Architectures for Quantum Computing

Benjamin says qubits are fragile and hard to control, especially when one qubit must operate on another. He describes two-qubit gate fidelity around 99.9% in Oxford and explains how error-correcting codes use extra qubits to detect and repair errors without measuring the stored quantum information directly. Better codes have lowered the estimated control threshold from about 99.9999% to roughly 99%.

Benjamin describes the surface code as a leading approach because qubits can sit in a two-dimensional grid and interact with nearby neighbors. Earlier approaches could require distant qubits to connect, adding many swap operations and opportunities for errors. More connectivity can help, but nearest-neighbor layouts offer a simpler hardware design.

At Oxford, one approach uses individual atoms with an electron removed, so electric fields can trap and manipulate the resulting ions in a vacuum. Benjamin says ion-trap qubits can maintain coherence for tens of seconds, while their high-fidelity operations support small modules. His proposed architecture links modules with optical connections, then uses repeated entanglement and purification to improve link quality.

Benjamin says roughly 50 qubits could exceed practical classical simulation, but that milestone alone would not establish a useful application. He sees chemistry and materials simulation as possible nearer-term opportunities, while many established fault-tolerant algorithms may require millions of physical qubits. He cautions that useful breakthroughs could take longer than the first machines and warns that exaggerated expectations could hurt research funding.

Benjamin says quantum computing now offers roles beyond academic physics. He points to opportunities for software engineers who can improve classical simulations and help verify quantum devices, as well as systems engineers who can integrate complex equipment such as vacuum chambers and optical interfaces.