Why people believe this
VQCs are already showing promising results on small problems. As hardware improves and circuits get larger, they will naturally scale to industrially relevant problems in optimization, chemistry, and ML.
The correction
VQCs face three simultaneous fundamental scaling challenges that do not diminish with hardware improvements alone. First, barren plateaus make training exponentially harder as qubit count grows. Second, noise accumulates faster than circuit depth can increase. Third, the classical simulation needed for gradient computation scales exponentially. These are not engineering challenges — they are algorithmic ones. Current VQC results on toy problems do not provide evidence of scalable quantum advantage.
Try it in the simulator
What to do
Place Rx and Ry rotation gates on multiple qubits with different angles. Run and inspect the state vector. Notice how sensitive the output is to angle changes. Now increase noise to 5% and run 10 times — the results vary dramatically. This sensitivity combined with noise and barren plateaus is why scaling VQCs is extremely difficult.
Research notes
Tags