That quantum computing is not superior to classical computing except WRT specific problems for which quantum effects can be leveraged. And currently there are no quantum advantage algorithms known or proposed for ML-related computational tasks (though there are ML methods applied to quantum optimization). To put it in perspective, the computer in your pocket is thousands of times more capable WRT inference than the best quantum computer on the planet, and we don’t expect that to change, because quantum computing isn’t meant to replace classical (“legacy”) computing, but rather to augment it.