Quantum computing has been "10 years away from practical applications" for the past 30 years. That joke may finally have an expiration date. A collaboration between IBM, Pfizer, and researchers at MIT has demonstrated the first unambiguously practical quantum computing application: drug discovery at a scale and speed impossible for classical computers.
The Problem
Drug discovery is fundamentally a molecular simulation problem. To find a molecule that will bind to a specific protein target — say, the amyloid plaques associated with Alzheimer's disease — you need to simulate the quantum mechanical behavior of electrons in thousands of candidate molecules. Classical computers approximate this simulation; quantum computers can perform it exactly.
The Result
Running on IBM's 4,000-qubit Condor 2 processor, the team's algorithm screened 847,000 candidate molecules against the Alzheimer's target protein in 72 hours. It identified three candidates with predicted binding affinities significantly higher than any existing drug. Classical supercomputers would have required an estimated 40 years to perform the same calculation.
The Path to Clinical Trials
The three candidates are now entering laboratory validation — the process of synthesizing the molecules and testing them in cell cultures and animal models. If any of them survive this process, they could enter human clinical trials within three years. The quantum algorithm has compressed what would have been a decade of computational work into three days.
What Comes Next
IBM and Pfizer have announced a five-year partnership to apply quantum computing to drug discovery across multiple disease areas. Other pharmaceutical companies are watching closely. The era of quantum-accelerated drug discovery may be beginning.
The Broader Drug Discovery Pipeline
The three Alzheimer's candidates identified by the quantum algorithm are now entering the standard drug development pipeline. Laboratory validation — synthesising the molecules and testing them in cell cultures — takes approximately 12–18 months. Animal model testing takes another 12–24 months. If any candidate survives these stages, Phase I human safety trials could begin as early as 2029. The quantum algorithm has compressed the computational discovery phase from decades to days, but the biological validation pipeline remains unchanged.
Other Disease Targets
The IBM-Pfizer collaboration has identified several other disease areas where quantum simulation could accelerate discovery. Parkinson's disease, where the target protein (alpha-synuclein) has a complex folding behaviour that is difficult to simulate classically, is a priority. Antibiotic resistance — where the challenge is designing molecules that bind to bacterial proteins without triggering resistance mechanisms — is another. The common thread is molecular complexity: diseases where the target protein's quantum mechanical behaviour is the limiting factor in classical simulation.
The Hardware Roadmap
IBM's quantum roadmap projects 10,000-qubit systems by 2028 and 100,000-qubit systems by 2033. Each generation of hardware improvement expands the range of molecules that can be simulated exactly. The 4,000-qubit Condor 2 processor used in this study is already capable of simulating molecules with up to 50 heavy atoms exactly — a threshold that covers a significant fraction of drug-like molecules. The 10,000-qubit generation will extend this to molecules with up to 150 heavy atoms, covering the vast majority of pharmaceutical targets.
Beyond Drug Discovery
Drug discovery is the first practical quantum computing application, but it is unlikely to be the last. Materials science is another domain where quantum simulation could be transformative. Designing new battery chemistries for electric vehicles, new superconducting materials for power transmission, and new catalysts for industrial chemical processes all involve the same kind of molecular simulation that quantum computers excel at.
Financial optimisation is another near-term application. Portfolio risk modelling, options pricing, and fraud detection all involve optimisation problems that quantum computers can in principle solve faster than classical alternatives. Several major banks — including Goldman Sachs, JPMorgan, and HSBC — have active quantum computing research programmes targeting these applications.
The Classical Computing Response
Classical computing is not standing still. Tensor network methods, neural network quantum states, and other classical approximation techniques have significantly improved the ability to simulate quantum systems on classical hardware. For some molecular systems, these classical methods now match or exceed what early quantum computers can do. The competition between quantum and classical simulation is ongoing, and the crossover point — where quantum computers provide a clear advantage for practically useful problems — is still being established.