The right answer is to do what many other cities across the world do, have stores close to where people live so they don't have to drive those 500 cars.
Even in Europe with public transport to the stores, I would bet delivery is many times more efficient.
The only question is whether people buy so much more due to convenience that the maybe 50x efficiency gain gets offset by 50x more items ordered. But I don't think that's the case.
Would be interesting to know the numbers. It's a bit like IKEA, their furniture requires less energy to produce and ship but does the resulting cheapness cause people to buy significantly more?
I don't think I buy significantly more stuff than I otherwise would just because it's convenient from Amazon, but perhaps I do.
This reminds me of when I needed to connect a screen 3 or 4m from the source. I needed to buy a hodge-podge of connectors and cables Because of local store stock-limitations, instead of 1 cable that did it all that I could buy online now.
That might be the most optimal solution, but I wouldn't take it as a given. You may have less individuals getting things delivered, but then you're giving up prime real estate and causing more decentralization of population, which brings its own inefficiencies.
You might still be imagining it in terms of how it is done in the US. You can have department stores on the first floor of dense housing, so people only need to walk to the store. The real estate is essentially what would be a bar or restaurant but serves more people and gives people what they need. You're right if it means turning a block that could be housing into parking and a walmart. It's better than a walmart being a 10 minute drive away from where you live, but it's only half way to how it can be (and often is elsewhere).
The only question is whether people buy so much more due to convenience that the maybe 50x efficiency gain gets offset by 50x more items ordered. But I don't think that's the case.