Most animal experiments are performed only if an effect size large enough to be ascertained using a small number of mice (i.e. <10, typically) is predicted based on in vitro or other models. Grant funded researchers do not typically have the resources to maintain such large colonies as you propose.

Also, most mice used in experimentation are inbred strains. Most human studies rely on such huge sample sizes to to identify low effect sizes within a diverse population.