By: Shailesh Mishra and Uma Vatsa
There are problems in chemistry that classical computers cannot solve. Making ammonia — which is vital for fertilizers, refrigeration, disinfection, and certain manufacturing — is reliant on the Haber-Bosch process from 1910. Ammonia production alone accounts for 2% of global energy consumption. But a similar process involving nitrogen molecules takes place naturally in bacteria among iron and molybdenum proteins, using an enzyme called nitrogenase. This process cannot be calculated precisely despite being studied and described since the 1960s. Electrons in the molecule occupy a spread of possible arrangements at once, making calculations beyond the reach of even supercomputers after a few dozen atoms. Quantum computing — in which a qubit can be both 0 and 1 at once unlike classical computing bits of 0 or 1 — is therefore necessary. It is no surprise that nitrogenase has become a standing benchmark problem in quantum chemistry.
Ammonia may be the best-known case, but similar limits define the energy industry today. How long a battery lasts is decided mostly in a very thin layer where the electrode meets the electrolyte. Nobody can calculate what goes on there, so manufacturers find out by building cells and running them until they fail. That is one reason batteries improve by a few percent a year instead of in jumps. Superconductors are harder still. Nobody can work out which material will carry electricity without losing any of it at a temperature you could actually use, so the search for superconductors has been mostly trial and error. Catalysts for splitting water into hydrogen and solvents for capturing carbon dioxide fall into the same bracket.
However, where quantum computing has reached utilities so far, it has mostly been about something else. The early projects concern optimization, where classical computing solutions already exist. Simulation is a different claim entirely. It is what quantum computers were invented for, and the one area where classical computing fails for reasons of physics rather than of engineering. Estimates of quantum technology being valued at $1.3 trillion to $2.7 trillion globally by 2035 rest on the simulation case holding up for materials and energy.
None of that exists yet. Qubits are fragile, and useful chemistry will need thousands of error-corrected logical qubits, each stitched together from many physical ones. The verified record, published in Nature in January, is 96 logical qubits built from 448 atoms. DARPA is running a program to test whether any architecture reaches industrial usefulness by 2033, and its managers now say somebody probably will, without venturing to guess who. A minority of serious researchers think error correction will never scale at all and that the whole enterprise fails at size.
This leaves the energy industry with a less dramatic question than the headlines imply. The first decade will be limited by hardware, and also by something more ordinary. You cannot point a quantum computer at a goal like “making ammonia cheaper.” The question must arrive as something far more specific: what do the iron and molybdenum molecules actually do to a nitrogen molecule? Or, what happens in that thin layer inside a cell when one battery additive is swapped for another? Turning a commercial problem into a question that precise is slow work, and it needs people who understand both the chemistry and the machine. There are few of them. The fertilizer producers and battery makers likely to gain first are the ones already doing that translation.
Quantum computing might not do much for the energy system in the very short-term. What it could change, if it works, is the ceiling on what the decades after that can build, because the limits on ammonia, energy storage, and electricity transmission are set by chemistry we cannot currently calculate. Those limits are already known. Describing them precisely enough to hand to a machine is work that can start now.
Shailesh Mishra is a Climate Justice Fellow at New York State Energy Research and Development Authority and Uma Vatsa is the Head of Artificial Intelligence and Data Science at Nelumbium Capital.