Let's say an organization decides to attack bitcoin by launching a 50% attack.
Can someone estimate how big the effort is?
- How expensive would such an attack be from scratch (buying enough ASICs)?
- How feasible is it for an organization with a lot of computers to do it using their existing computational power?
For example, can a company like google do it by, say, turning 5% of their computers towards attacking bitcoin? Can the US government do it (if they chose)? Can a few banks group up, buy some equipment and hijack bitcoin this way?
I understand that these are all possible in principle, so I'm looking for back of the envelope estimates of how realistically feasible such scenarios are.