I've read about several companies that seemed to be doing well, shutting down. I've had the opportunity to start digging into one, and saw that the costs to keep their cloud running might have been what killed them. I'm in the process of trying to re-build the thing from the ground up on purchased, dedicated hardware, but not everyone could do that.
Are cloud costs, and the associated lax engineering practices that are being solved by throwing more cloud resources at problems causing problems for these companies?
Cloud providers seem like drug dealers. They make it easy to get hooked, then they jack up the price.
Once you buy, you are provided a dazzling array of APIs and platform features which do not have direct equivalents from provider to provider. The temptation to use them is strong- they're free, right?
Having eaten these metaphorical pomegranite seeds, you're stuck. Migrating to another provider becomes an uphill battle. This is the ecosystem working as intended. Of course, if you're a startup which intends to coast for 2 years before imploding, being bought out, or IPOing for billions, planning for the future might be sheer overhead! Smoke 'em (the VC funds, that is) while you've got 'em!
Sometimes the cloud is a cost-effective solution. Some use-cases benefit from the elasticity and low startup cost/time. Other times it would be much cheaper and easier to run a handful of physical servers. It varies. I suspect that if everyone was considering their options prudently and looking toward the future, the cloud would not be used as widely as it is today.
edit: more to the original question, perhaps the answer is that companies that are killing themselves anyway have an incentive to use cloud services, and do so recklessly!