So basically this is once more a case, where cloud providers promise you everything then corner you into unforgiving situations where you have to spent most valuable thing (engineers' time) to avoid associated costs with particular vendor lock in, in order to have service continuation.

60 TB is a lot, but we can all agree that is surely doesn't warrant -/+ 13000 USD one-way trip price tag.

This statement is in particular worrisome: "Google Compute Engine network rates (we’ll ignore these, as they get complicated)"

So you are basically buying a black-box that has an ability to drive your business into a ground if you are not careful enough.

Yes, I am aware you can argue and plead for mercy once you get the bill, but this is not the way: they wouldn't be so keen in forgetting your cost if it was making them a real expenditure: I was doing a napkin calculation for AWS and I found it to be around 10x more expensive than DIY solution, which means that from 10 suckers, only one has to pay for them to make profit.

The more I read about cloud providers the more I am convinced that their major business is based on the fact that they have to get hardware and software for their core business and since they cannot scale it precisely it is more lucrative for them to just try to sell "solution" to suckers for 10x and get rich in the process.