Software is, at its core, an intellectual exercise. When you start up a new multi-billion dollar thing, there's very little physical manifestation. There isn't a sprawling factory, a company town, a bustling assembly line...
There's just a piece of software. Building it takes a lot of effort, but not like effort effort the way you have to move the ground and get machines in to build a factory.
So even though walking away from a massive software project means, in dollars, the same as abandoning a giant factory, it isn't legible in the same psychological way. If you're Intel, and your new process node isn't working the way you expected, do you board up the factory? No, of course not! It's a gigantic thing that you can look at and say "wow, we spent a billion dollars on this." But if you're Facebook, come monday morning, you can just tell all of your ICs -- "hey, you guys aren't working on this any more."
This is kind of related to how quickly money flows into software. Money come money go. Software come, software go. Not so in the physical world.
That's my theory.
That also means -- if you treat your software projects like a big factory, you can actually distinguish yourself. Like many on HN, I am very hesitant to sign up for a new google service, even if it's really really cool and I would pay for it! Because there is no factory, there is no machines, there's just information, information that can stop being improved tomorrow.