I agree with this assessment. My philosophy as tech lead was this: every time we wanted to start using a SaaS product, up to and including cloud hosting itself

I agree with this assessment. My philosophy as tech lead was this: every time we wanted to start using a SaaS product, up to and including cloud hosting itself, we needed to have a Plan B to migrate to in case the service suddenly went away for any reason. Eg.: we needed a message queue, we picked up an Azure product because it was the most convenient and covered all our requirements at low cost. But we coded against an interface and we implemented a proof-of-concept RabbitMQ implementation as a backup plan (also came in handy for local testing!). If we had to drop the Azure service for any reason, we knew we could install RabbitMQ on a VM and tune it up to production standards relatively quickly.