People dig themselves in hard, with reliances on all kinds of proprietary services, and complex relationships.
My experience in helping people do cloud migrations are that companies also often quickly lose oversight over which cloud services they are even still running. Sometimes systems that should've been shut down years ago are still hanging around, or parts of them anyway, like S3 buckets etc. Most companies that use cloud systems underprovision their devops because they think they don't need much for a cloud system (in fact, they typically need more to do it well).
$300k/day for their revenue is very much crazy high.
During good times, many companies simply don't care about which services or how much. I've worked at several startups where they got a large funding round and the word was "we don't care about cloud costs, just get it done fast / make it scale." Unfortunately, one bad mistake (like storing all data in a proprietary service like DynamoDB) can make this difficult to unwind when things get bad...
My experience in helping people do cloud migrations are that companies also often quickly lose oversight over which cloud services they are even still running. Sometimes systems that should've been shut down years ago are still hanging around, or parts of them anyway, like S3 buckets etc. Most companies that use cloud systems underprovision their devops because they think they don't need much for a cloud system (in fact, they typically need more to do it well).
$300k/day for their revenue is very much crazy high.