Forecasting timelines of quantum computing

Jaime Sevilla, C. Jess RiedelarXiv 2020 · arXiv

Abstract

Predicting the trajectory of quantum computing (QC) has traditionally relied on expert intuition, which is often subjective and lacks statistical rigor. This paper introduces a quantitative approach to forecasting QC milestones by defining the Generalized Logical Qubit (GLQ)—a unified metric that synthesizes physical qubit count and gate fidelity. By extrapolating historical progress from a curated dataset of 52 quantum systems, the authors provide a probabilistic timeline for critical milestones. Their findings suggest that while a single functional logical qubit may emerge by 2030, a quantum computer capable of breaking RSA-2048 is unlikely to appear before 2039, with a median projection of 2050.

The Need for Quantitative Metrics

The field of quantum computing is currently in the "Noisy Intermediate-Scale Quantum" (NISQ) era. While hardware is advancing rapidly, progress is often measured along two conflicting axes: increasing the number of physical qubits and decreasing gate error rates. Pushing both simultaneously represents the primary engineering hurdle. Existing forecasts often fail to account for this tension, leading to overly optimistic or inconsistent predictions. To address this, the authors propose the GLQ metric, which uses the scaling laws of the surface code to estimate the number of "perfect" logical qubits a noisy system could theoretically support.

Figure 1: Contour plot illustrating how the Generalized Logical Qubit (GLQ) metric synthesizes physical qubit counts and average error rates into a single figure of merit.

Methodology and Data

The study utilizes a newly compiled database of 52 quantum systems reported between 2003 and 2020, spanning technologies such as superconducting loops, trapped ions, and silicon qubits. The researchers focus their final forecasting model on superconducting qubits due to the robustness of their historical data.

The forecasting approach involves:

  1. Multivariate Modeling: Using log-linear regression to model the joint evolution of qubit counts and error rates over time.
  2. Frontier Extrapolation: Fitting models to "world record" data points to track the bleeding edge of technological capability.
  3. Uncertainty Quantification: Employing a bootstrapping procedure to generate 90% confidence intervals for future milestones.

Key Results and Milestones

The authors operationalize two major milestones using the GLQ metric: proof-of-concept quantum fault tolerance (1 GLQ) and the ability to execute a quantum attack on RSA-2048 (4100 GLQs).

Figure 5: Operationalization of key quantum computing milestones in terms of Generalized Logical Qubits.

The model's projections are notably more conservative than many industry surveys:

  • 1 GLQ (Fault Tolerance): Unlikely (<5% confidence) before 2026, with a median estimate of early 2030.
  • 4100 GLQs (RSA-2048 Attack): Unlikely (<5% confidence) before 2039, with a median projection of 2050.

Discussion and Limitations

A key insight from the data is the existence of a "technological frontier" where designers must trade off scale for fidelity. The authors view their results as a "generally optimistic" upper bound, as the model assumes exponential progress in error reduction will continue without hitting fundamental physical floors.

Significant caveats remain, including the scarcity of high-fidelity data points and the exclusion of factors like qubit connectivity and coherence times. Future refinements could incorporate economic variables, such as R&D investment levels, and evaluate alternative error-correction codes to provide even more granular forecasts.

Summary updated

BibTeX
@misc{sevilla2020forecasting,
  author = {Jaime Sevilla and C. Jess Riedel},
  title = {{Forecasting timelines of quantum computing}},
  year = {2020},
  eprint = {2009.05045},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2009.05045},
}
Figure