[threatofrain]

Many businesses start with single points of failure. The goal is to get enough money to where this is not true, but they all start this way. Ultimately the goal is managing risk, and that often means not building out your own data center as the first thing you do.

[ovao]

This, more increasingly, doesn’t mean building out your own data center. Often this just means you can deploy services to multiple platforms without a dramatic or upending burden of effort. I don’t think startups should necessarily burn capital on federation, but startups can at least make engineering decisions and trade-offs to ensure a future livelihood not contingent on single points of failure.

Unless you’re Apple or Google, you’ll have some single points of failure. But that’s not the same as the livelihood of what you’re setting out to do being reliant on a single vendor for over a decade.

[rglullis]

There is quite a lot between "building your own data center" and "building your tech stack in a way where you are not locked to one specific vendor".