Utilities can't design just for the average month, but the worst ones, and the worst hours of the worst days. Building new peak demand (peaker) plants is very expensive and most consumers probably don't see the actual market prices during those times. Shifting load by a very small percent could result in a decent cost savings and/or avoid building new capacity.
In practice it’s a little more complex as maintenance cycles for example are lined based on seasonal demand, but that’s offset by much higher transmission losses when the temperatures are highest.
Really, the US electric grid is designed to share power between states, that’s just part of capacity planning. Further, sharing power is a cost saving measure and happens as soon as importing power costs less than using a peaking power plant. It still requires someone to have built the generating capacity, but it’s much easier to add 1GW of base load capacity at a time when your selling some of that to a different state.
As to minor differences 50GW is 67% larger than 30 GW that’s a huge swing over the course of a day. At 0.5 KW per person in CA x 4 miles per kWh x 12 hours x 365 days, that’s 8,760 miles per year which is almost enough on it’s own. Suggesting at most a minimal capacity increase.