These are the reasons that I refuse to build any of my projects on some cloud provider's special snowflake solutions.
EG: I can move HA PG to any provider but if I build around RDS specific tooling I'm boxed in and any migrations are OOM harder.
Before I moved reddit from a datacenter to AWS, I was "locked in" to my datacenter. It took weeks to get all the data out and into AWS.
Working around their special services isn't all that hard. Nothing AWS offers is that hard to replicate outside of AWS if you want to. The hard part is getting your data out.
So if you're going to put all your data into AWS anyway, you might as well use their services and save yourself some time and effort reinventing what they've already done. If you really need to move out later, then make the investment in rebuilding the tools they offer.
Terraform is "multi provider" not "provider agnostic".
And for not-infrastructure cloud services, aren't they pretty interchangeable? I recently evaluated four different cloud speech-to-text services. Sure the APIs differed, but you know, it's an API call. I see nothing to complain about.
https://build5nines.com/linux-is-most-used-os-in-microsoft-a...
Maybe the switch isn’t too bad for small customers, but in enterprise where people spend years specialising and obtaining and renewing certificates you can’t just transfer your staff members between the cloud services without a major retraining investment, in which you’ll likely loose your best staff because there is a reason they chose to spend all those years on your previous service.
Ideally, this particular challenge can be overcome using something like terraform, which allows you to build infrastructure in a relatively cloud-agnostic way. But even that is a risk, since it's a relatively new project.