I also thought it was silly, but maybe they mean that frequentist methods still have analytical solutions in some settings where Bayesian methods must resort to Monte Carlo methods?