A qubit is a physical system that can exist in two quantum states at once, but what that system actually looks like depends entirely on the technology. It might be a tiny loop of superconducting metal, a single trapped atom, an electron confined in silicon, or even a particle of light. Each approach encodes the “0” and “1” of quantum computing into a different physical property: electrical current, atomic energy levels, electron spin, or light polarization. Here’s what each one actually is and how it works.
Superconducting Qubits: Circuits That Act Like Atoms
The most widely used qubit today is a superconducting circuit, built from nanoscale metal components on a chip that looks surprisingly similar to a classical processor. The key ingredient is a component called a Josephson junction: two layers of superconducting metal separated by an ultra-thin insulating barrier. When cooled to extreme temperatures, electrical current flows through this junction without any resistance, and the circuit behaves like an artificial atom with distinct energy levels.
In a normal electrical circuit, energy can take on any value, like a ramp you can stop on anywhere. The Josephson junction changes the circuit’s physics so that energy comes in discrete steps, like rungs on a ladder. The two lowest rungs become the qubit’s “0” and “1” states. Crucially, the spacing between these rungs is uneven, which lets engineers target just those two levels with microwave pulses without accidentally pushing the system to a higher rung. This uneven spacing is what makes the circuit usable as a qubit rather than just an oscillator.
Because these energy levels are determined by the physical dimensions and materials of the circuit, engineers can design qubits with specific properties. That tunability is a major advantage. It’s also why superconducting qubits are sometimes called “artificial atoms,” since they mimic the discrete energy structure of real atoms but can be customized on the drawing board. IBM’s current processors use this approach, with roadmap chips scaling to 360 qubits capable of running 7,500 gate operations.
Trapped Ions: Single Atoms Held in Place
A trapped-ion qubit is literally a single atom, stripped of one or more electrons to give it an electric charge, then suspended in a vacuum using electromagnetic fields. The qubit’s “0” and “1” are encoded in two specific internal energy levels of the ion, typically hyperfine levels. These are natural quantum states that arise from the interaction between the ion’s electron cloud and its nucleus, separated by an energy gap that falls in the gigahertz range.
To manipulate the qubit, researchers shine precisely tuned laser light onto the ion. Different laser frequencies initialize the qubit, flip it between states, or entangle it with a neighboring ion. One common technique, called a stimulated Raman transition, uses two laser beams whose frequency difference matches the energy gap between the qubit’s two states, driving the ion from one state to the other in a controlled way. To read the result, a detection laser is aimed at the ion: if it’s in one state it glows brightly, scattering photons that a detector can count. If it’s in the other state, it stays dark. Ytterbium-171 is a popular choice, manipulated with ultraviolet laser light at 369 nanometers.
Trapped ions have a natural advantage in that every ion of the same species is physically identical, so there’s no manufacturing variation between qubits. The tradeoff is speed: laser-based operations tend to be slower than the microwave pulses used on superconducting circuits.
Spin Qubits: Electrons Locked in Silicon
Spin qubits store information in the quantum spin of a single electron, a fundamental property that makes the electron behave like a tiny bar magnet pointing either “up” or “down.” To isolate that electron, researchers create a quantum dot: a nanoscale pocket in a semiconductor (usually silicon) where voltages applied to metal gates trap exactly one electron in place.
Silicon is attractive because the semiconductor industry already knows how to manufacture it at enormous scale. Recent work has produced industry-compatible silicon spin-qubit cells by forming quantum dots against thin oxide layers, which helps create well-isolated quantum states and efficient control through gate voltages. To read the qubit, a technique called Pauli spin blockade is used: two neighboring quantum dots are tuned so that current can only flow between them if the electrons have opposite spins. Whether current flows or not reveals the qubit’s state.
Spin qubits are physically tiny, on the order of tens of nanometers, which makes them promising candidates for scaling to very large numbers. The challenge has been getting precise enough control over each dot to achieve high-fidelity operations, but recent results show single and two-qubit gate fidelities exceeding 99% in silicon devices.
Photonic Qubits: Information Carried by Light
A photonic qubit encodes quantum information in a single photon, the smallest possible packet of light. The “0” and “1” can be represented by the photon’s polarization (horizontal versus vertical), its path through a chip (taking one waveguide versus another), or its arrival time. Polarization encoding is the most intuitive: a horizontally polarized photon represents 0, a vertically polarized one represents 1, and a photon polarized at an angle is in a superposition of both.
Photonic qubits have a unique practical benefit. They naturally travel at the speed of light, making them ideal for transmitting quantum information over distances. They’re also relatively resistant to environmental noise because photons don’t interact strongly with their surroundings. On-chip manipulation is done using specially designed waveguides, including twisted waveguide structures that rotate a photon’s polarization in a controlled way, acting as single-qubit gates. The main difficulty is making photons interact with each other, which is necessary for two-qubit operations. Photons pass right through one another unless you engineer indirect interactions, which makes building a full photonic quantum computer more complicated.
Topological Qubits: A More Exotic Approach
Topological qubits aim to store quantum information not in a single particle’s property but in the collective behavior of special quasiparticles called Majorana zero modes. These emerge at the ends of semiconductor nanowires (made from materials like indium antimonide or indium arsenide) placed in contact with a conventional superconductor and subjected to a magnetic field. The first experimental evidence appeared in 2012 from a lab in Delft, where researchers detected a characteristic electrical signal, a zero-bias conductance peak, that matched theoretical predictions for these exotic states.
The appeal of topological qubits is their built-in error protection. Quantum information is stored in the relationship between pairs of Majorana zero modes that can be separated by a physical distance along the wire. Local disturbances, such as a stray electric field or a vibration, can’t corrupt the information because it isn’t located at any single point. Computations are performed by “braiding” the quasiparticles, moving them around each other in specific patterns. The result of the computation depends only on the overall pattern of the braid, not on the exact path taken, which makes it inherently resistant to small errors. This protection holds as long as the operation is performed slowly enough and the disturbance isn’t strong enough to close the energy gap that separates the topological states from ordinary ones. Topological qubits are the least mature of the major approaches, but they remain a major research focus because of their potential for fault tolerance.
Why Qubits Need Extreme Conditions
Quantum states are extraordinarily fragile. Any stray energy from the environment, whether heat, electromagnetic radiation, or mechanical vibration, can destroy the superposition that makes a qubit useful. This process, called decoherence, is the central enemy of quantum computing.
Superconducting qubits require the most dramatic environmental control. They operate inside dilution refrigerators that cool the processor to roughly 15 to 20 millikelvins, just a fraction of a degree above absolute zero. At that temperature, thermal vibrations are so feeble that the delicate quantum states in the circuit can survive long enough to perform calculations. The refrigerator itself is a complex, multi-stage system, with intermediate cooling stages at around 200 millikelvins supporting the coldest stage below.
Trapped-ion systems operate in ultra-high vacuum chambers at room temperature for the surrounding equipment, but the ions themselves are laser-cooled to near absolute zero. Silicon spin qubits also require deep cooling, typically in the tens of millikelvins range, for the same reason as superconducting circuits. Photonic qubits are the exception: because photons barely interact with their environment, photonic processors can operate at or near room temperature.
How Close Are Physical Qubits to Reliable?
No physical qubit is perfect. Every operation introduces a small probability of error, and the standard measure of quality is gate fidelity: how close an operation comes to its intended result. In early 2025, MIT researchers achieved a record single-qubit gate fidelity of 99.998% using a superconducting qubit design called a fluxonium, with a complementary two-qubit gate fidelity of 99.92%. Trapped-ion platforms consistently achieve fidelities above 99.5% for single-qubit operations. Silicon spin qubits have recently crossed the 99% threshold for both single and two-qubit gates.
These numbers sound high, but for practical quantum computing, error rates need to be pushed even lower, or compensated for by using many physical qubits together to create a single, more reliable “logical” qubit. Current estimates suggest that anywhere from hundreds to thousands of physical qubits may be needed to support one logical qubit, depending on the error rate of the underlying hardware. That’s why the race to build better physical qubits and to scale them up is happening simultaneously.