The Road to Practical Quantum Advantage: Science Experiment or Reality?

by Dr.

George Schwartz Executive Summary One pervading question in quantum computing has been whether commercially useful quantum advantage can be achieved with early, near-term devices — referred to as the kiloquop era by Global Quantum Intelligence (GQI). Historically, the field has struggled to move beyond esoteric benchmarks to find real-world applications that demonstrate the useful quantum advantage defined by program Q4Bio. However, at GQI, we are observing a tangible transition. Recent notable demonstrations—such as Q-CTRL’s highly optimized 1D Fermi-Hubbard simulation, Q4Bio’s hybrid tensor-network pipeline for chemical simulation, and Pasqal’s modeling of non-equilibrium dynamics in frustrated magnets—provide encouraging evidence for those who believe near-term commercial advantage is possible. While the quantum industry is not there yet, the likelihood of useful commercial advantage in the kiloquop era has increased sharply. GQI continues to track these capability thresholds closely to keep a pulse on the industry’s trajectory. The landscape of quantum computing has undergone a fundamental shift since the landmark 2019 experiment by Google, which first demonstrated quantum supremacy by creating quantum states on a 53-qubit processor with a computational state-space of dimension 2⁵³. [1] That original experiment was a conceptual milestone, proving that a quantum processor could perform a specific random circuit sampling task exponentially faster than any classical supercomputer. However, the field is now migrating to a more mature phase focused on practical and/or useful quantum advantage. This new frontier moves beyond synthetic benchmarks to address real-world problems directly relevant to material science, engineering, and critical bottlenecks in biology and medicine. By integrating quantum hardware with sophisticated classical methods, researchers are beginning to solve problems that were previously intractable, particularly in understanding complex chemical reactions and the behavior of exotic materials. Q4Bio’s Lotus of Useful Quantum Advantage The pursuit of quantum advantage is no longer defined by a single metric but by a conceptual convergence that the Q4Bio team visualized as a “Lotus of Useful Quantum Advantage.” For a quantum achievement to be considered truly useful, it must simultaneously satisfy executability on current hardware, demonstrate relevance to biologically or chemically meaningful systems, and undergo rigorous validation against state-of-the-art classical methods. Progress in isolation fails to address underlying computational bottlenecks. True advantage emerges when these elements are integrated into a single workflow, recognizing that quantum computers will likely act as generators of correlated samples that augment, rather than replace, classical high-performance computing. Biological Simulation and Hybrid Pipelines One potential driver for useful quantum advantage is the simulation of drug candidates for Photodynamic Therapy (PDT). PDT is a targeted medical treatment where a light-activated drug, or photosensitizer, selectively destroys diseased cells, such as those in tumors. The efficacy of these drugs depends on light-triggered chemistry, where the drug molecule transitions from a ground state to an excited state to produce reactive oxygen species that kill cancer cells. Because classical techniques like Density Functional Theory (DFT) are unreliable for the transition-metal complexes used in PDT, and more advanced classical methods like the Density Matrix Renormalization Group (DMRG) become cost-prohibitive as molecular entanglement increases, the Q4Bio program developed a hybrid pipeline using 100 qubits. [2] This pipeline utilizes a combination of the ADAPT-VQE algorithm with Algorithmiq’s Majorana Propagation to prepare high-quality quantum states with polynomial resources, avoiding the orthogonality catastrophe that typically hinders Quantum Phase Estimation (QPE) for chemistry. By extracting informationally complete data to create unbiased estimators of the quantum state, researchers fed these “quasistates” into classical tensor-network optimizations. The resulting QB-DMRG approach demonstrated that quantum-generated data can boost classical methods, achieving lower energy states than classical DMRG at the same bond dimension—a crucial parameter indicating the complexity of the simulation. As illustrated by GQI’s framework for quantum chemistry algorithms, the hybrid QB-DMRG pipeline successfully pushes the boundaries of quantum correlation far beyond standard VQE. Pushing the Boundaries of Chemical Dynamics On a similar trajectory toward practical advantage in complex systems, Q-CTRL recently showcased how deeply optimized algorithmic implementation can push simulation boundaries. Targeting the 1D Fermi-Hubbard model, their team executed a 60-site, 120-qubit simulation for 30 second-order Trotter steps. This feat was enabled by an application-tailored compilation that reduced the number of two-qubit gates by approximately 61% and circuit depth by over 99% compared to standard IBM Qiskit baselines, leading to a remarkable 152 2Q-depth with 9,057 2Q-gates. [3] Furthermore, improvements by Q-CTRL—including but not limited to the integration of dynamical decoupling, Pauli twirling, and readout-error mitigation—subsequently improved the root-mean-square error by roughly 24% and 12%, respectively. These error reduction techniques are based on runtime error suppression, rather than error mitigation, resulting in little to no effective postprocessing overhead. This efficiency cements speed as a winning angle for quantum computing. For the interested parties, the application-tailored compilation will be available as a future Qiskit function. To validate these results, the quantum outputs were benchmarked against Time-Dependent Variational Principle (TDVP) simulations using the classical ITensor library. By systematically increasing the bond dimension, 𝜒, (a parameter the controls the simulation accuracy) up to 4096, the classical dynamics matched the quantum simulator’s output for progressively longer time durations, up to an evolution time of 5.2 (in units of inverse hopping, 1/𝑡ₕ). At this point, both the quantum and classical simulation’s correctness was indeterminate, as illustrated via the red zone for the above RMSE figure. [3] Scaling the classical simulation beyond this crossover point became computationally prohibitive, at which stage the quantum computer proved to be over 3,000 times faster in total wall-clock time, requiring only 2 minutes compared to 100 hours for the classical tool. This emphasis on conquering the complexity of the Fermi-Hubbard model proves that in this demonstration, quantum hardware can currently outpace state-of-the-art classical architectures in simulating strongly correlated materials.

Achieving Chemical Accuracy with Error Correction While the aforementioned work focuses on boosting classical methods using noisy hardware, another advancement is the demonstration of a fully error-corrected pipeline for molecular energy calculations. Quantinuum’s recent work with molecular hydrogen is an example of a stepping stone for the transition toward fault-tolerant quantum computing. For quantum chemistry to be truly predictive, it must reach high-precision chemical accuracy, which typically requires QPE. In this demonstration, researchers utilized the [[7, 1, 3]] color code to encode logical qubits across seven physical qubits, with real-time error detection and correction. [5] This end-to-end pipeline estimated the experimental ground-state energy of the hydrogen molecule to within 0.001(13) hartree of the exact value, successfully benchmarking the system near the threshold of chemical accuracy (1.6 x 10⁻³ hartree). The experiment also highlighted that memory noise, which accumulates while qubits are idling or being transported across a processor, is currently a more significant bottleneck than gate infidelity in encoded circuits. By utilizing logical qubits and dynamical decoupling, the system suppressed these errors, establishing a path toward the more complex simulations enabling future materials discovery.

Simulating Frustrated Magnets and Real Materials Moving beyond molecular chemistry, the analog simulation of frustrated magnets offers an equally compelling direction for practical advantage. In this context, quantum simulators act as microscopic probes for real-world materials where atomic arrangements prevent magnetic spins from settling into a simple order. Using a 256-qubit Rydberg simulator, Pasqal researchers successfully modeled the low-dimensional frustrated quantum magnet TmMgGaO₄. [6] By rescaling the physical properties of the crystal into a programmable array, the analogue quantum simulator generated magnetization measurements that showed excellent agreement with laboratory susceptibility measurements. Crucially, this quantitative correspondence allowed the researchers to probe non-equilibrium dynamics on picosecond material timescales. When observing the system following a sudden quench, the growth of entanglement became so rapid and complex that classical numerical methods reached their limits. This ability to predict the response of a real material at picosecond timescales represents a threshold of capability that moves quantum computing to a predictive tool for complex solid-state physics. Observations on the Road to Scalability Practical quantum advantage is not arriving as a single moment of total dominance, but rather emerging through a sequence of capability thresholds. As system sizes scale into the hundreds of qubits and error mitigation transitions into full error correction, quantum processors are becoming essential, high-value components of broader scientific workflows. From resolving the subtle energy levels of a cancer drug to tracking ultrafast entanglement in a frustrated magnet, quantum advantage is now defined by its growing ability to expand the boundaries of what we can reliably predict and engineer in the physical world. References [1] Arute, F., Arya, K., Babbush, R., Bacon, D., Bardin, J. C., Barends, R., … & Martinis, J. M. (2019). Quantum supremacy using a programmable superconducting processor. Nature, 574(7779), 505–510. https://doi.org/10.1038/s41586-019-1666-5 [2] Algorithmiq. (2024). The road to useful quantum advantage: Q4Bio perspective paper. Algorithmiq. https://algorithmiq.fi/publications/Q4Bio_Perspective_Paper.pdf [3] Q-CTRL. (2026). Practical quantum advantage in materials simulation: Technology briefing. Q-CTRL. [4] Hartnett, G. S., Najafi, K. S., Khindanov, A., Liao, H., Schutzman, M., Hush, M. R., Biercuk, M. J., & Baum, Y. (2026). Fast, accurate, high-resolution simulation of large-scale Fermi-Hubbard models on a digital quantum processor. Q-CTRL. Advance online publication. [5] Yamamoto, K., Kikuchi, Y., Amaro, D., Criger, B., Dilkes, S., Ryan-Anderson, C., … & Muñoz Ramo, D. (2026). Quantum error-corrected computation of molecular energies. PRX Quantum. Advance online publication. arXiv: https://arxiv.org/abs/2505.09133 [6] Leclerc, L., Julià-Farré, S., Silva Freitas, G., Villaret, G., & Albrecht, B. (2026). One-to-one quantum simulation of the low-dimensional frustrated quantum magnet TmMgGaO₄ with 256 qubits. arXiv: https://arxiv.org/abs/2603.20372 May 6, 2026