Recent work by the author has included two computation‐intensive models that suffer severe practical limitations if forced to run on commonly available computers such as a DEC VAX 11/780. Results of one of the models PERUSE were recently published [J. Acoust. Soc. Am.75,1415 (1984)]. The second model is a state‐of‐the‐art coupled mode model described by C. A. Boyles [J. Acoust. Soc. Am.73,800 (1983)] and implemented numerically by the author [J. Acoust. Soc. Am. Suppl. 173,S96 (1983)]. Results of both models will be presented and discussed in light of how computation time relates to the complexity of the particular problem being solved. Citing also earlier work by the author and F. D. Tappert [J. Acoust. Soc. Am.63,353 and 533 (1978)] involving expensive Monte Carlo calculations required to verify a theory of stochastic wave propagation, a forceful argument will be made that solutions to some of the most pressing problems on the frontier of ocean acoustics today require the use of supercomputers or dedicated array processors.