A perfect example is airports: they are designed to handle different types of planes landing on them, from small to relatively big.
Yes, but the bigger the planes become, like the A380 shows, the more expensive the infrastructure gets, so you need a realistic number of large planes per year for affording the infrastructure. For a small regional airport far away, it makes no sense, but for a large hub, it does - and such a large hub will not waste the infrastructure for serving smaller planes, it will prefer the larger ones.
The size and costs of rockets will more likely evolve - by evolutionary pressure from the market, by means which really mean survival of the fittest. A launcher that is too large for the market will not get far, just like a launcher that is too small for it. Instead any growth in size (which is reasonable, if you look at the container ship market), will come slowly, step by step, with gradual reductions in costs. Small launchers will fill niches, which the bigger standard orbit carriers will leave open, since it is hard getting a good launch rate for non-standard orbits.
This does not mean automatically, that the bigger ones must be expendable. Reusable launch vehicles are not automatically bad, because they failed to meet the promises of the Space Shuttle. It is again a matter of evolution.