For long-tail adverse events? It's easy to engineer power distribution systems for the average case, but unless we accept occasional brownouts as a cost of doing business, we need to build robustness into our generation fleet. Let's say we get another Krakatoa-scale eruption that reduces insolation (and thus drops solar & wind output while increasing heating demand). Some amount of firm generation (hydro, gas peaker, nuclear) is insurance against correlated failure in the global climate system.