Superconducting vs trapped-ion qubits
Also known as: superconducting qubits, trapped-ion qubits, ion trap quantum computer, transmon qubit, quantum hardware types
Superconducting qubits and trapped-ion qubits are the two most mature ways to build quantum computers.
Superconducting qubits, used by Google and IBM, are tiny electrical circuits cooled to near absolute zero, and they're fast. Trapped-ion qubits, used by Quantinuum and IonQ, are individual charged atoms held in place by electromagnetic fields and controlled by lasers, and they're slower but highly accurate. Neither has won, and the race is far from over.
Superconducting vs trapped-ion qubits, explained at three levels
At home
Think of two ways to build an instrument. One is made on a factory line, quick to produce but each slightly different, so each needs tuning. The other uses natural parts that are all identical, but it's slow to play. Superconducting qubits are the factory instrument: tiny circuits printed on chips that run very fast. Trapped-ion qubits are the natural kind: single charged atoms, all identical and highly accurate, but slower.
At work
Superconducting and trapped-ion qubits are the two most mature ways to build a quantum computer. Superconducting qubits, used by IBM, Google, and Rigetti, are electrical circuits on chips, cooled near absolute zero; they run operations very quickly. Trapped-ion qubits, used by Quantinuum and IonQ, are individual charged atoms held by electromagnetic fields; they're slower but more accurate. Cloud services offer both, so businesses can compare them directly.
At home
A superconducting qubit is a small circuit made of a metal such as aluminium, printed on a chip like ordinary electronics. Cooled to about 10 thousandths of a degree above absolute zero, current flows without resistance, and two of its energy levels serve as 0 and 1. A trapped ion is a single atom, such as ytterbium or barium, missing an electron. Electromagnetic fields hold it in a vacuum, and lasers or microwaves set its internal energy levels to 0 or 1.
At work
The trade-offs show up in daily use. Superconducting gates take tens of nanoseconds, so circuits run fast, but each qubit connects only to a few neighbours, so compilers add swap operations to move information around. Ions in one trap can often interact with any other ion, avoiding those swaps, and they reach some of the highest reported gate fidelities, above 99.9%. But ion gates take microseconds to hundreds of microseconds, so the same job can run a thousand times slower.
At home
Superconducting qubits rely on Josephson junctions, thin insulating barriers that make the circuit's energy levels unevenly spaced, so two can be singled out as a qubit. Because every chip is manufactured, each qubit differs slightly and needs individual calibration. Ions use energy levels that nature fixes, so every ion is identical and coherence can last seconds. In my view, ions have nature's consistency on their side while superconducting qubits have industrial chipmaking, and that trade-off is why neither has pulled ahead.
At work
Each approach hits different scaling walls. Superconducting systems need control wiring for nearly every qubit through a dilution refrigerator with limited cooling power, plus consistent fabrication across large chips. Ion systems struggle to control more than about a hundred ions in one trap, so designs such as Quantinuum's shuttle ions between zones, and others plan to link traps with photons. Benchmarks should compare full-system results, such as logical error rates and time to solution for a target task, rather than qubit counts.
Real-world examples
At home
Sophie tried two cloud quantum services for a class project. The superconducting system returned results quickly but with more noise, while the trapped-ion system took longer but gave cleaner results for her small circuit.
At work
Marcus runs an innovation team at a Canadian bank exploring quantum optimization. He tested the same problem on both superconducting and trapped-ion platforms, comparing accuracy and runtime, and found neither yet beat the bank's classical methods.
How it compares
| feature | Superconducting qubits | Trapped-ion qubits |
|---|---|---|
| Qubit | Tiny superconducting circuits | Individual charged atoms |
| Operating conditions | Dilution refrigerator near absolute zero | Ultra-high vacuum, laser cooling |
| Gate speed | Fast (tens to hundreds of nanoseconds) | Slower (microseconds or longer) |
| Accuracy | Good, improving | Among the highest reported |
| Connectivity | Mostly nearest neighbours | Often all-to-all within a trap |
| Key players | Google, IBM, Rigetti | Quantinuum, IonQ |
Superconducting qubits win on speed and manufacturing; trapped ions win on accuracy and connectivity.
For analysts
Don't bet on one quantum hardware type yet. I'd keep pilots platform-neutral, using software tools that run on multiple backends, so you can switch as the technologies mature.
History & origin
In 1995, Ignacio Cirac and Peter Zoller proposed a trapped-ion quantum computer, and a NIST team led by David Wineland demonstrated a quantum logic gate with trapped ions that same year. Wineland shared the 2012 Nobel Prize in Physics. Yasunobu Nakamura and colleagues demonstrated a superconducting qubit in 1999. Superconducting systems scaled quickly: Google's 53-qubit Sycamore claimed quantum supremacy in 2019, IBM unveiled the 1,121-qubit Condor in 2023, and Google's Willow demonstrated below-threshold error correction in 2024. Trapped-ion systems pushed accuracy, with Quantinuum reporting two-qubit gate fidelities above 99.9%, and its Helios system launched in 2025. In my opinion, the contest comes down to a classic trade-off: speed and manufacturability versus precision and connectivity.