One logical qubit, the kind that can hold information reliably, may take anywhere from 1,000 to 10,000 physical qubits to build. That is the estimate from current surface code models, and it depends on how noisy the hardware is. Under the hood, the technical reality is that qubits are fragile things.
A classical bit is a 0 or a 1, and if it flips by accident, a parity check can usually put it right. Qubits sit in superpositions. Stray heat, noise or a sloppy gate operation can collapse the state or scramble its phase. Errors come in several flavours: bit flips, phase flips, depolarizing noise, and leakage, where the qubit slips out of the computational space altogether. Quantum states also cannot be copied or measured without disturbing them, so the usual classical tricks do not carry over.
Spreading the information out
Quantum error correction gets around this by spreading one logical qubit across many physical ones. Helper qubits are measured to flag what kind of error has occurred, without collapsing the logical state. The surface code arranges qubits on a two-dimensional lattice where each talks only to its neighbours, and it has a fault-tolerance threshold of roughly 1%. Because it needs only local operations, it suits superconducting and trapped-ion hardware.
The price is overhead. A paper on Quantum Design Automation describes the surface code as prominent but demanding in qubits, and notes that researchers are testing alternatives such as the Bivariate Bicycle Code to cut the bill.
Who is trying
IBM’s roadmap includes logical qubits and error-corrected circuits on its superconducting processors. Google Quantum AI has run early surface code experiments and reached what it calls break-even error correction. Microsoft is betting on topological qubits built from Majorana fermions. Quantinuum works with trapped ions. The prize is practical: breaking RSA with Shor’s algorithm needs error-free gates across thousands of qubits, and simulating large molecules for drug discovery needs deep circuits that do not pile up mistakes.
The detail that matters
As ever, the hurdles are mundane next to the physics. Physical qubits must reach error rates of around 10-4 or better. Systems need hundreds of thousands of qubits. Classical decoding has to keep pace in real time, correcting errors as they happen. And superconducting machines still need dilution refrigerators that run near absolute zero.
The Quantum Design Automation paper argues that the trouble rarely sits in one component. Systems break at the interfaces between circuit design, error correction and hardware, it says, and tools borrowed from classical chip design struggle with the different error models and connectivity limits of superconducting qubits. Its authors want hardware, software and algorithm teams designed together, with fault tolerance built in from the start.
Sources
- Fault-Tolerant Quantum Computing: Key to Scalability
- Building Quantum Computers: The Automation Imperative
