Quantum computers face limitations not from competing technologies but from the laws of physics; qubits are highly sensitive to environmental conditions and can easily lose their necessary computational states.

The latest edition of the "Quantum & After" series delves into the complexities of physical and logical qubits, decoherence, and error correction.

Brief Summary

Quantum computers are not expected to replace traditional computers, fit in a pocket, or run standard software. This limitation stems not only from the current technological stage but also from the inherent sensitivity of qubits to external noise, leading to rapid error accumulation as computations become more complex. Without error correction, lengthy algorithms become inaccurate. To mitigate errors, quantum information is distributed across multiple physical qubits, resulting in a more robust logical qubit.

Understanding the Limitations

Referencing an example from a previous installment, imagine a coin that has yet to land; we cannot predict whether it will show heads or tails. In quantum systems, uncertainty operates similarly yet differently: a qubit can exist in a superposition of two states. Any external interference can disrupt this state and introduce errors. In quantum hardware, sources of such errors include thermal noise, vibrations, electromagnetic interference, and imperfections in materials and control. This is why superconducting processors from Google and IBM operate at temperatures around 10–20 mK, significantly below the cosmic microwave background radiation temperature of approximately 2.72 K. Such cooling reduces thermal noise but necessitates complex cryogenic setups.

A physical qubit represents a real quantum system, such as a superconducting circuit, an ion trapped in a field, a photon, or an electron spin. Due to environmental interactions, these qubits are noisy; for superconducting qubits, the error rate for two-qubit operations typically ranges from 10⁻³ to 10⁻² (0.1% to 1%).

Accuracy levels of 99.9% have already been demonstrated on the technological platforms of market leaders. Among ion-based platforms, Quantinuum achieved a two-qubit gate accuracy of 99.914% across all qubit pairs with its H1-1 system, while its flagship 98-qubit Helios reached 99.921%. In superconducting systems, comparable accuracy levels are exhibited by IQM, which recorded a 99.91% accuracy for CZ gates, and IBM, whose Egret and Heron processors achieve approximately 99.9% accuracy.

A logical qubit is an error-protected unit of information whose quantum state is spread across multiple physical qubits through quantum error correction (QEC). One of the most researched methods is the surface code, which offers advantages such as localized operations, a flat two-dimensional architecture, and a high theoretical error resilience threshold of around 1% (source).

In the past two years, QEC experiments have advanced significantly:

  • The Willow quantum processor from Google Quantum AI demonstrated operation below the surface code threshold, showing exponential suppression of logical errors as the code distance increases. As the lattice size of physical qubits grows, the error rate of the logical qubit consistently decreases more than twofold with each scaling step;
  • The Quantinuum Helios quantum computer utilizes 98 physical qubits based on barium-137 ions in a QCCD architecture. In one configuration, the company showcased 48 logical qubits with error correction, with a ratio of about two physical qubits per logical qubit.
  • A research team from Harvard, MIT, and QuEra employed a processor with up to 448 neutral atoms, executing circuits with dozens of logical qubits, and in some setups, up to 96.

Historically, achieving reliability required redundancy of up to ~1000 physical qubits for a single logical qubit, leading to estimates of a million qubits for commercially viable tasks. High-speed qLDPC codes potentially reduce these requirements by an order of magnitude—to dozens of physical qubits per logical qubit, rather than just a few.

Implications

Mass-market universal fault-tolerant quantum computers still exist only in company roadmaps. For instance, IBM has announced plans to build a 200 logical qubit Starling system by 2029, capable of executing 100 million operations. Even if these projects stay on track—despite regular delays—it will result in a specialized co-processor in the cloud alongside classical supercomputers, rather than a replacement for data centers.

The number of physical qubits alone does not indicate the capabilities of a quantum machine. Factors such as operation quality, error emergence speed, suppression efficiency with code scaling, and, crucially, the number of logical qubits that the system can maintain with a sufficiently low error probability are also vital.

Q-check ForkLog: Where Advantages Lie

Will a quantum computer replace a standard laptop?

No: quantum computers operate on a different principle, addressing a narrow class of tasks, and require cryogenic temperatures around 10–20 mK.

Where is the advantage formally demonstrated?

In specialized test problems like boson sampling (simulating light particle behavior) and random circuits (generating truly random quantum states), quantum devices have shown superiority over classical computations. Certain algorithms, such as Shor's algorithm for factoring and discrete logarithms, have proven theoretical acceleration.

Where is advantage expected?

In simulating nature and optimizing complex systems: pharmaceuticals (accurate molecular modeling for drug development), materials science (discovering superconductors and new battery types), and chemistry (creating effective catalysts). Optimization and finance are also under exploration, but practical advantages are not yet proven.

Where is there no advantage?

In everyday tasks like text processing, file storage, running interfaces, gaming, and standard data processing, classical processors remain more efficient and cost-effective.

Does more qubits equal more utility?

No. A large processor is ineffective for lengthy algorithms if errors accumulate faster than the system can correct them.

Looking Ahead

If the technology is so error-sensitive and requires complex infrastructure, the next question arises: why are corporations and governments still investing billions in it? The next installment will explore the investment rationale within the industry: what corporations and states expect, and where the line is drawn between betting on the future and a bubble.

Catch up on previous editions:

  1. Can one profit from quantum technologies?
  2. How blockchains are preparing for the "quantum" era.
  3. Can the quantum internet be hacked?
  4. Is the quantum computer a threat to Bitcoin?
  5. Why do we need a quantum computer?