Potential and challenges of quantum computing hardware technologies

by Martina Gschwendtner, Niko Mohr, Nicole Morgan, and Henning Soller

The quantum computing market may grow to about $80 billion by 2035 or 2040. For now, many qubit technologies are in the running to become the basis of the first fault-tolerant universal quantum computer,1 but there are currently no shared criteria for evaluating the technologies’ benefits, challenges, and progress. Comparing the technologies can be difficult, even for decision makers with technical backgrounds.

In this post, we focus on quantum computing hardware. First, we discuss technical considerations to keep in mind. We then outline the five major qubit technologies: photonic networks, superconducting circuits, spin qubits, neutral atoms, and trapped ions. We conclude with a preview of a few qubit technologies that have not yet matured but may bring benefits in the future.

None of the information in this post is intended to be a ranking or an endorsement; there is no clear winner among the technologies. Instead, we hope to help technical decision makers think through the implications of engaging with each quantum computing technology. Because the considerations are varied and cannot be captured in a standard set of metrics, the discussion is necessarily qualitative.

A consistent way to compare qubit technologies

No generally accepted approaches to assessing and comparing qubit technologies have emerged, and a combination of qubit technologies may ultimately produce a fault-tolerant quantum computer. Through a review of several hundred research publications on qubit technologies, we identified six key considerations—and challenges—to evaluate for each qubit technology.

  • Fidelity at scale. Fidelity is intimately connected to quantum computing’s defining hurdle, which is increasing qubit count and computational power for complex algorithms while maintaining high levels of qubit quality (in other words, high fidelity).
  • Computation speed. Individual qubits can retain their quantum state—what’s called “coherence”—for only a limited time. To compensate, gate operations (a quantum gate is a basic quantum circuit operating on a small number of qubits) should occur quickly enough to make complex computations possible before qubits in the system lose coherence.
  • Multiqubit networking. The more qubits that can be linked to one another to perform gate operations, the more readily they can implement quantum computing algorithms and the more powerful the resulting quantum computer would be.
  • Control over individual qubits at scale. Control over individual qubits is critical to quantum computing. As the number of qubits in a quantum computing system increases, control over individual qubits becomes increasingly complex.
  • Cooling and environmental control. For most qubit technologies, the required scale of the cooling equipment in terms of both size and power is beyond the feasibility of currently available equipment.
  • Manufacturing. Some qubit designs use existing production technology, while others require new manufacturing techniques. The production of eventual full-scale quantum computers will require automated manufacturing and testing for components at scale.

Five major qubit technologies

We applied our evaluation criteria to five major qubit technologies (exhibit). Because the industry and the technology are still relatively early in their development, this discussion is a snapshot of a moment in time rather than a ranking or endorsement.

Five main qubit technologies are competing to build a scalable universal quantum computer.

Photonic networks

In photonic networks, each qubit is encoded in a single photon. An initial cluster state (a highly entangled state involving multiple qubits) is prepared.2 Afterwards, gate operations are executed through a series of measurements of photons. Unlike with other qubit technologies, two-qubit gates in photonic networks are probabilistic, not deterministic.3

The advantages of the technology are its potential for massive quantum entanglement,4 its speed, and its ability to operate at room temperature. However, fidelity at scale is a significant hurdle for photonic networks. The largest source of error for photonic networks is photon loss during computations. Another concern is the source of photons, which needs to consistently create identical photons for any computation to be valid.

Superconducting circuits

Superconducting circuits work by encoding each qubit in the energy levels of Cooper pairs (pairs of electrons bound together at low temperatures) on opposite sides of a Josephson junction.5

The benefits of superconducting circuits are their coherence, processing speed, and ease of manufacturing. Indeed, companies using this technology have found ways to automate the manufacturing of the chips required for superconducting circuits. These manufacturing processes may continue to be usable as the technology matures and more qubits are added.

However, scaling, calibration, control electronics, and cooling are all challenging. Fidelity in these systems decreases as the numbers of qubits and successive gates increase. Current approaches related to control electronics and calibration for superconducting qubits are also not yet scalable to the degree experts currently consider necessary. Another technical hurdle is cooling at scale because superconducting circuit systems need to be kept at cryogenic temperatures.

Spin qubits

In spin qubit systems, each qubit is encoded in the spin of the electron of a semiconductor quantum dot. Two-qubit gates are set up between entangled electrons in a silicon chip, and qubits are controlled by microwave electronics.

Spin qubits benefit from their small size—spin qubit systems revolve around single electrons, which creates small quantum systems—and the relatively straightforward way in which they can be manufactured. Indeed, manufacturing for spin qubit chips could borrow knowledge from classical manufacturing methods.

But spin qubits share many of the challenges that bedevil superconducting circuits, particularly fidelity at scale and control electronics. As with superconducting circuits, error rates for spin qubits scale with size. And because spin qubits are so small, they require much more precise control electronics because the qubits are physically closer together in space. The challenge of cooling at scale is intensified in these small systems because the dissipated heat would be concentrated in a smaller area compared with superconducting circuits.

Neutral atoms

In neutral-atom technology, a qubit is usually encoded in two ground-state (the lowest-energy state of an atom or particle) hyperfine levels of an atom. Two-qubit gates are created by exciting two atoms until they reach the Rydberg state.

Neutral-atom technology has advantages in scaling, coherence, and cooling. Scaling the number of qubits for neutral-atom technology to about 1,000 physical qubits should be relatively straightforward because their neutrality minimizes interference between qubits. The qubits in these systems also have strikingly long coherence times compared to other technologies: minutes under experimental conditions.6

The challenges would be scaling to a million qubits, control electronics, and error rates. It’s not yet clear if scaling arrays beyond 100,000 atoms with individual qubit control is achievable in neutral-atom systems. Control electronics at scale is another missing piece. Finally, neutral-atom systems’ error rates are higher than those of other systems.7

Trapped ions

In trapped-ion technology, each qubit is encoded in two energy level states of an ion. Two-qubit gates exploit the coupling between the electron and phonon (a quantum of energy associated with a compressional wave, such as a vibration) between an ion’s excited electron state and the vibrational modes of the ion chain.

Trapped ions have the benefit of the lowest error rates among the technologies we examined for small two-qubit gate systems, as well as lower cooling requirements compared to other atom-based technologies. Fidelity at scale is also a lesser challenge because the qubits targeted for entanglement can be physically moved and manipulated.

Increasing the number of qubits in trapped-ion systems is the most significant obstacle for the technology. Indeed, creating entanglement across more than two qubits in trapped-ion systems has proved difficult.8 Furthermore, qubits’ maneuverability in this technology creates speed challenges because physically moving ions is slow compared to changing electronic states.9

Other difficulties are intertwined with challenges of coherence. Trapped-ion systems are limited in their achievable size because fidelity declines with the distance ions must travel.

Emerging qubit technologies

Because no one knows which technology will ultimately be used to realize a fault-tolerant quantum computer, it is worth noting emerging qubit technologies that are still theoretical. For example, electrons on solid neon and electrons over superfluid helium have recently received attention.10

The leading emerging technology, however, is Majorana fermions. Majorana fermions are produced at the phase boundary in a superconducting wire or at the core of vertices in strong magnetic fields. Majorana fermions could create high-fidelity, highly scalable qubits with a quantum state that is protected by an effect known as topological protection.11 For now, additional research breakthroughs are necessary to make this technology a reality.

The technological future of quantum computing remains wide open. We hope the six criteria we outline here can help stakeholders in quantum computing—and eventually the public—discuss hardware technologies using a shared set of premises.

Martina Gschwendtner is a consultant in McKinsey’s Munich office, Niko Mohr is a partner in the Düsseldorf office, Nicole Morgan is a consultant in the Prague office, and Henning Soller is a partner in the Frankfurt office.

1 A “qubit” is a quantum bit—a basic unit of information in quantum, akin to bits in conventional computers.
2 For more on fusion events, see Sara Bartolucci et al., “Fusion-based quantum computation,” Nature Communications, February 2023, Volume 14, Number 912.
3 For more on measurement-based quantum computing, see H. J. Briegel et al., “Measurement-based quantum computation,” Nature Physics, January 2009, Volume 5.
4 In quantum entanglement, the quantum states of a pair of particles are indefinite until they are measured, and measuring one affects the outcome of measuring the other.
5 For more on Cooper pairs, see Leon Cooper, “Bound electron pairs in a degenerate Fermi gas,” Physical Review, November 1956, Volume 104, Issue 4.
6 Shuoming An et al., “Single-qubit quantum memory exceeding ten-minute coherence time,” Nature Photonics, September 2017, Volume 11.
7 Dolev Bluvstein et al., “Hardware-efficient, fault-tolerant quantum computation with Rydberg atoms,” Physical Review X, June 2022, Volume 12.
8 Rainer Blatt et al., “Observation of entangled states of a fully controlled 20-qubit system,” Physical Review X, April 2018, Volume 8.
9 Reinhold Blümel et al., “Efficient stabilized two-qubit gates on a trapped-ion quantum computer,” Physical Review Letters, June 2021, Volume 126.
10 For more on electrons on solid neon, see Ralu Divan et al., “Single electrons on solid neon as a solid-state qubit platform,” Nature, May 2022, Volume 605. For more on electrons over superfluid helium, see Frederic Lardinois, “EeroQ bets on helium for its quantum chip design,” TechCrunch+, August 23, 2022.
11 For more on topographical protection, see Miguel A. Bandres et al., “Topological protection versus degree of entanglement of two-photon light in photonic topological insulators,” Nature Communications, March 2021, Volume 12, Number 1974; and Lucas Casparis et al., “Protocol to identify a topological superconducting phase in a three-terminal device,” arXiv, 2021.