You could use the same math to show why cars are a terrible solution to transportation. And yet, we've got millions upon millions of cars, and relatively few t

You could use the same math to show why cars are a terrible solution to transportation. And yet, we've got millions upon millions of cars, and relatively few trains and buses (at least on my continent). People don't decide how to live their lives by sitting down with a calculator and figuring out the most efficient method to serve the entire population of their country. As the article pointed out, the biggest problem in this case is not technical. It's social. Now that you've shown that the big machines are more efficient overall, how do you get there from here? I would even say that if you know from the start that you want a bunch of $500K machines, then bootstrapping -- what he's doing now -- is in fact the easiest way. Google has some massive datacenters but they didn't start by spending $50M on hardware on day 1.