Much of the excitement in quantum computing comes from the “qubit count”, the number of qubits a device has. Instead, a traditional microprocessor’s number of transistors (whether a 6502* or the latest Intel Pentium) is often considered a proxy for power. Although not the whole story, it does provide a simple measurement that is easy and simple to understand. Other metrics, such as quantum volume, have been exposed by companies such as IBM and utilized by Honeywell and Quantinuum.
We outline the growth in the number of qubits showing the rising number of qubits with time across the range of qubit technologies. Naturally, the trend is towards increasing qubit count, but we plot the different technologies against each other to highlight the current and past developments.
Qubit Growth
We plotted the qubit count against multiple technologies. We can see that superconducting is leading, and specifically, this is IBM with its 433 qubits. Of course, we are simply comparing numbers and not concerned with comparisons such as fidelity. Still, despite different technologies, tracking qubit count provides an intelligent and easy-to-understand metric comparing qubit technologies.
The trend line shows the mean or average qubit count with time. The simple measure highlights the upwards trend; it averages across all technologies. Also plotted is the trend line for each particular technology, again showing the improving growth in qubit numbers. Hockey stick growth is observed with a familiar exponential growth curve when planning the maximum number of qubits in one year. The line plotted is an “expanding” maximum which always plots the maximum qubit.
Counting Qubits
By some standards, counting the number of qubits might appear crude and open to exploitation. Indeed, the count is not being exploited with gate-based quantum computers compared to annealers, and we see no outlandish claims of vast numbers of useless qubits. Qubits being demonstrated can be manipulated, offer connectivity between other qubits and are useable for NISQ-based quantum circuits.
So despite alternatives, we still see the qubit count garnering the most headlines. Qubit count still generates headlines, and it is a significant development target or aspiration for many companies, including IBM (the originator of alternative metrics such as Quantum Volume). IBM has created a roadmap that shows the route “big blue” will take in its quest to dominate the superconducting quantum landscape. Of course, there are alternative technologies, but the superconducting gate-based system rocks a sporty 433 qubits right now, with rivals such as Rigetti sporting only double-digit qubits. PsiQuantum, one of the best-funded pure-play quantum start-ups/scale-ups, has set the bar at one million qubits, as has Xanadu with its modular photonic system.
Types of Qubit
Superconducting Qubits
Superconducting qubits are the most widely used in quantum computing today, with notable examples being the ones used by IBM and Google. They create a superconducting circuit where the electrical current can flow without resistance. The state of the qubit (0 or 1) depends on the direction of the current flow. Superconducting qubits offer scalability and can be manufactured using techniques similar to classical computer chips. However, they require extremely low temperatures to function, which demands complex and expensive cooling infrastructure.
Trapped Ion Qubits
In trapped ion qubits, individual ions are trapped using electromagnetic fields, and these ions are manipulated using lasers to represent qubits. The advantage of trapped ion qubits is that they have long coherence times, meaning they can maintain their quantum state for longer. This makes them less prone to errors. Companies like IonQ and Honeywell are using this type of qubit. The challenge here is the complexity and precision required for trapping and controlling ions with lasers.
Topological Qubits
Topological qubits are a form of quantum bit that some researchers are looking at, notably Microsoft through its Quantum Development Kit. They rely on a field of mathematics called topology and the concept of ‘braiding’ of anyons (quasiparticles that exist only in two dimensions). Theoretically, these qubits will have more resistance to errors due to their topological nature. However, as of my knowledge cutoff in September 2021, practical topological qubits have not been fully realized.
Photonic Qubits
Photonic qubits use the quantum properties of photons, the fundamental light particles, to perform quantum computations. Photons have the advantage of being very robust against noise, and they don’t need to be kept at cryogenic temperatures as superconducting qubits do. However, manipulating photons in a controlled way can be challenging. Companies such as Xanadu work on photonic systems.
Silicon Spin Qubits
Silicon spin qubits utilize the spin state of an electron in a silicon-based device as the basis for information storage and manipulation. The advantage of silicon spin qubits is that silicon is already widely used in traditional computing, which could simplify the manufacturing process. These qubits also have long coherence times compared to superconducting qubits. However, controlling and reading the spin state of a single electron is a technical challenge.
Neutral Atom Qubits
Neutral atom qubits trap individual neutral atoms and manipulate their quantum state using lasers. Similar to trapped ions, but as the atoms are neutral, they interact less with the environment, which could lead to lower error rates.
A significant advantage of neutral atom qubits is that they can be arranged in 2D or 3D configurations, providing more connectivity between qubits. Companies like ColdQuanta and Atom Computing are exploring this type of qubit.
NV Centre Qubits (Diamond)
NV centre qubits are a newer and less common type of qubit. They use a flaw (or “defect”) in a diamond’s structure, known as a Nitrogen-Vacancy (NV) centre. The NV centre can exist in multiple states simultaneously to be used as a qubit.
The NV centre qubits have relatively long coherence times and can operate at room temperature, unlike superconducting or trapped ion qubits. However, they are still in the research phase and have not been scaled to larger systems. This field is being researched in academic settings and not as much in the industry as the others.
Quantum Volume, what is it?
Quantum Volume is a metric used to measure a quantum computer’s power and complexity. It doesn’t just count the number of qubits (often the main focus in quantum computing) and considers other crucial factors such as qubit quality, error rates, and how well the qubits can interact with each other. It considers the largest square-shaped quantum circuit (with the same number of input and output qubits) that can successfully run on the quantum computer.
To simplify, think of Quantum Volume as a “score” that reflects not just the number of players (qubits) in a soccer team, but also how skilled they are (qubit quality), how many mistakes they make (error rates), and how well they work together (qubit interaction).
Qubit growth continues
The numbers will be updated as new developments happen. Please let us know if you think we have missed a result, and we’ll update our data points. Stay tuned, and we’ll update this with new developments; we soon expect that IBM will be pushing the limit to over 1000 qubits in 2024, according to the IBM quantum roadmap.
A special shoutout to the Superconducting Qubit
By far, the most popular area is the superconducting qubit area with a vast array of points plotted and currently is the best performer in terms of an absolute number of qubits.
*The nod to the 6502 is for the Vintage Computer buffs.