The application quantum computers were invented for
Richard Feynman's original argument in 1981 was simple: nature is quantum mechanical, so simulating it on a classical computer is exponentially expensive, and the fix is to simulate it on a machine that is itself quantum mechanical. Four decades later, simulating quantum systems remains the application with the clearest theoretical justification — no contrived problem, no dequantisation risk, no data-loading bottleneck, because the data is already quantum.
Why classical simulation runs out of room
Describing a system of n interacting quantum particles takes on the order of 2n numbers. At around 50 particles you exhaust the world's supercomputers. This is not a matter of waiting for faster classical hardware; the wall is exponential. It is why quantum chemistry relies on approximations — density functional theory, coupled cluster — that work well for many molecules and fail badly for exactly the strongly correlated systems people most want to understand: transition metal catalysts, high-temperature superconductors, nitrogen fixation.
What the articles here cover
- Quantum chemistry — molecular ground-state energies, reaction pathways, catalyst design; the most commercially discussed target.
- Condensed matter — Hubbard and Ising models, phase transitions, topological phases, superconductivity, quantum magnetism.
- Many-body dynamics — how quantum systems evolve out of equilibrium, thermalise, or fail to.
- Analogue simulation — building a controllable quantum system that behaves like the one you want to study, rather than running a digital circuit. Cold atoms in optical lattices have produced real physics results already, ahead of gate-based machines.
- Foundations and adjacent physics — quantum optics, thermodynamics, chaos, field theory and quantum gravity, where the computational questions meet open physics.
The honest status
Analogue simulators are producing results today. Digital, gate-based chemistry at a scale that beats the best classical methods is not here yet — resource estimates for industrially interesting molecules run to millions of physical qubits with current error-correction overheads, though those estimates have fallen substantially as algorithms improve. This is a field where the goalposts move in the right direction, but slowly.
