NISQ Reality: Near-Term vs Long-Term

On current noisy devices no fusion computation beats classical; the honest case for quantum is fault-tolerant and years out.

What NISQ can actually do today

Noisy intermediate-scale quantum (NISQ) devices have on the order of tens to low hundreds of physical qubits, two-qubit gate error rates around 1e-3 to 1e-2, and coherence times that limit circuits to roughly hundreds to low thousands of gates before noise dominates the signal. This budget is far too small for any Kronos materials problem at useful accuracy.

The circuit depth reachable before noise swamps the answer scales roughly as the inverse of the per-gate error. If a variational chemistry circuit for a modest active space needs depth D and two-qubit count G, the probability the whole circuit runs clean is approximately:

P_clean  ~=  (1 - e2)^G

# e2  = two-qubit gate error (~1e-3 best hardware)
# G   = number of two-qubit gates in the circuit
# For G = 1e4, e2 = 1e-3:  P_clean ~= e^{-10} ~= 4.5e-5
# -> answer is buried in noise without error mitigation/correction

The three honest statements

• No published quantum computation has produced a fusion-relevant materials or plasma result more accurate or faster than the best classical method.
• Variational methods (VQE, QAOA) have no proven scaling advantage and face barren-plateau and measurement-cost obstacles.
• The credible advantage cases require fault-tolerant, error-corrected machines that do not yet exist at the needed scale.

Where the line is

We treat near-term quantum as a research and benchmarking activity, not a production tool. It runs in L0 next to classical solvers, always with a classical baseline computed in parallel. A quantum result is accepted only if it reproduces the classical answer on cases classical can reach, before we ever trust it on cases classical cannot. See error mitigation and error correction for what changes the picture.

This candor is a feature. Overclaiming quantum advantage would poison the same trust the honest breeder and burner gates are meant to build. The machines are design-and-simulation studies until FOAK first tritium around 2030; the quantum program is positioned on an even longer horizon.