AstraZeneca: Quantum-Accelerated Chemistry Simulation for Drug Synthesis

The Problem

Bringing a small-molecule drug to market depends not only on finding a molecule that binds its target but also on being able to make it efficiently. Synthetic chemists assemble drug molecules through sequences of reactions, and choosing routes that are fast, high-yielding, and clean is a major part of pharmaceutical development.

The Suzuki-Miyaura reaction is one of the most important tools in that toolkit: a class of carbon-carbon bond-forming reactions used widely in the synthesis of small-molecule drugs. Understanding such reactions at the level of electronic structure, how bonds break and form along the reaction pathway, requires solving quantum mechanical problems that are expensive for classical computers when the chemistry is strongly correlated.

AstraZeneca, IonQ, AWS, and NVIDIA collaborated to demonstrate that a hybrid quantum-classical workflow could tackle a realistic instance of this chemistry.

Why Quantum Computers for Chemistry

The electronic structure of molecules is a quantum many-body problem. The Hilbert space describing N electrons grows exponentially with N, which is exactly why classical methods must approximate, and why simulating quantum systems was Feynman’s original motivation for quantum computers. A quantum processor’s state space also grows exponentially with its qubit count, giving a natural mapping between molecular states and qubit states.

Near-term machines cannot yet simulate large molecules exactly, so practical work uses hybrid approaches: a quantum circuit handles the part of the problem that is genuinely quantum, while classical high-performance computing handles everything else and orchestrates the overall calculation. The simplified snippet below illustrates the shape of such a hybrid loop. It is a teaching example, not the collaboration’s production workflow.

# Illustrative only: the shape of a hybrid quantum-classical chemistry loop.
# Teaching example, not AstraZeneca/IonQ's actual workflow or results.
import numpy as np

def quantum_energy(params):
    """Stand-in: a real run evaluates <psi(params)|H|psi(params)>
    by executing a parameterised circuit on quantum hardware."""
    return np.sum((params - 0.3) ** 2) + 1.0  # toy energy surface

def classical_optimize(energy_fn, n_params, steps=200, lr=0.05):
    params = np.random.uniform(0, np.pi, n_params)
    for _ in range(steps):
        grad = np.zeros(n_params)
        eps = 1e-3
        base = energy_fn(params)
        for i in range(n_params):
            shifted = params.copy(); shifted[i] += eps
            grad[i] = (energy_fn(shifted) - base) / eps
        params -= lr * grad
    return params, energy_fn(params)

best_params, best_energy = classical_optimize(quantum_energy, n_params=4)
print("Toy converged energy:", round(best_energy, 4))

The Workflow

The demonstrated workflow combined IonQ’s Forte quantum processing unit, a trapped-ion system reported at 36 algorithmic qubits, with NVIDIA’s CUDA-Q platform, running through Amazon Braket and AWS ParallelCluster, and using NVIDIA H200 GPUs to accelerate the classical portions of the calculation. CUDA-Q orchestrated the hybrid execution so that quantum and classical resources worked together within a single pipeline.

Using this stack, the team ran a large-scale, end-to-end simulation of a Suzuki-Miyaura reaction. IonQ described it as the most complex chemical simulation run on its hardware to date.

Results

The collaboration reported more than a 20x improvement in end-to-end time-to-solution compared with previous implementations, with the expected runtime for the workflow reduced from months to days while maintaining accuracy. The work was announced on June 9, 2025, and the results were showcased at the ISC High Performance conference in Hamburg, June 10 to 13, 2025, with a research paper subsequently posted to arXiv.

The honest framing is a research demonstration: an end-to-end proof that a hybrid quantum-classical pipeline can handle a realistic drug-synthesis reaction faster than prior approaches. It is not a claim of production use or of advantage over the best classical chemistry methods at this scale.

Why It Matters

Each partner contributed a distinct layer: AstraZeneca brought the pharmaceutical chemistry problem and domain relevance, IonQ provided the quantum hardware, AWS supplied the cloud orchestration and compute fabric, and NVIDIA provided the GPU acceleration and the CUDA-Q software that tied quantum and classical execution together. That integration is itself the point. Useful near-term quantum chemistry is unlikely to come from quantum hardware alone; it will come from workflows where quantum processors and classical supercomputers operate as one system.

For students, this case study illustrates the hybrid quantum-classical pattern, the role of reaction simulation in drug development, and how speedups in near-term work are measured against prior implementations of a pipeline rather than against an absolute quantum advantage.

Learn more: CUDA-Q Reference | IonQ SDK Reference

Sources

• IonQ Speeds Quantum-Accelerated Drug Development Application with AstraZeneca, AWS, and NVIDIA (IonQ)
• IonQ Demonstrates Quantum-Accelerated Workflow for Drug Discovery with AstraZeneca, AWS, and NVIDIA (Quantum Computing Report)
• IonQ Speeds Quantum-Accelerated Drug Development Application With AstraZeneca, AWS and NVIDIA (Business Wire)