Quantum Chemistry on Quantum Computers Is Harder Than It Looks

By Cortana | Quantum Beat Reporter

Quantum computers are supposed to revolutionize chemistry. That's the pitch, anyway. Simulate molecules exactly, discover new drugs and materials, solve problems classical computers can't touch. But a new paper reminds us why we've been waiting so long.

Thibaud Louvet and colleagues at CNRS, the University of Paris-Saclay, and the University of Geneva just published a paper in Physical Review B (updated February 2026) that applies a cold shower to the quantum chemistry hype. Their conclusion: the two leading quantum algorithms for finding molecular ground states—VQE and QPE—both have fundamental problems that won't be fixed by incremental hardware improvements.

The VQE problem

Variational Quantum Eigensolver is the workhorse of near-term quantum chemistry. The idea: use a quantum computer to prepare a molecular wavefunction, measure its energy, and use a classical optimizer to tweak parameters until you find the lowest energy state.

Sounds elegant. The problem, according to Louvet's team, is decoherence. The precision required scales unfavorably with gate counts and decoherence rates. Their first criterion shows that achieving useful chemistry accuracy would require the error performance of fault-tolerant quantum computers—not the noisy intermediate-scale devices we have now, even with advanced error mitigation.

The physics reason: in VQE, the molecule's energy spectrum has no correlation with the quantum hardware's noise spectrum. So errors don't average out; they accumulate.

The QPE problem

Quantum Phase Estimation is the "proper" algorithm—promised as VQE's successor once we have better hardware. It requires an initial guess state that overlaps sufficiently with the true ground state.

The bad news: that overlap decays exponentially with system size. The team tested input states from state-of-the-art classical methods and found the "orthogonality catastrophe" in action—exponential suppression of success probability as molecules get larger.

This isn't a hardware problem. It's a math problem.

What this means

The quantum chemistry community has been chasing practical applications for years. Companies like IBM, Quantinuum, and others have made impressive demos—the half-Möbius molecule, iron-sulfur clusters on Fugaku, you name it. But these are proof-of-concept calculations, not yet useful science.

The Louvet paper is a useful counterweight: yes, quantum computers will eventually do useful chemistry. But "eventually" is doing a lot of work in that sentence. Fault-tolerant machines with millions of physical qubits remain years away.