Variational quantum algorithms dominate gate-based applications of modern processors. The so-called layerwise trainability conjecture appears in various works throughout the variational computing literature. asserts that a circuit can be trained piecewise, e.g., few layers sequence to minimize an objective function. Here, we prove this false. Counterexamples are found by considering functions e...