But, the idea that you can just move workloads to a random VPS and get the same benefits of AWS/Azure/GCP is …naive and shows a lack of understanding of how any real company operates.
The last startup I worked for had 60 employees and we hosted microservices that aggregated public and private (to the company not PII) health care data for health care providers. They used them as backends for their internal processes and public websites and mobile apps.
We had to be scalable since any new provider could increase traffic by 10-20% and that helped when COVID came around in 2020 and traffic spiked
We had multiple MySQL databases, ElasticSearch, ETL processes using Lambda, CloudFront (CDN),S3 for storage, SQS (messaging), ECS (Docker orchestration), VMs for some workloads.
Should we have put those on some random VPS and spun up the equivalent for ourselves ?
Yes, I know how to do infrastructure outside of cloud. I started my career where half my job was managing all of this on prem - as in a raised floor in the same building - before AWS was a thing.
Everyone who throws together a website and hosts it on Hetzner shouldn’t be considered a “startup”.
And when that VPS goes down? Where are these projects?
In the scenario I listed about my last startup, it would be silly to host those on a bunch of VMs and not use managed services even if they are hosted versions of open source software. The management of infrastructure takes too much time once it gets necessarily complex and doesn’t add. Skid to the business