I mean, imagine if all of AWS was just a single data centre labeled us-west-1. Imagine something happened to it. This is comparable to how Wuhan shutting down impacted the global supply chains.
(I lack a good web comparison for what happened afterwards - the EU, and recently the US, shutting down, hitting the global supply chains again, this time on the demand side.)
A dependency is something you own (a checked-out copy of). A better comparison is with services. So you may be using a payment gateway in your product, and maybe you need just a single one - but at the very minimum, you should have a plan in place to switch to a different one at a moment's notice. That's reasonable redundancy.
I think I've come up with an analogy for demand-side supply chain issues though. Imagine you have a web service that's constantly under 90% load during normal operations, and that eats almost exactly as much money in the infrastructure costs as it earns. Imagine that service gets mentioned on Hacker News. It almost immediately fails due to increased demand, you can't provision new instances because you don't have enough money in the bank, and then when suddenly the demand drops, you can't afford to keep it on at all.
Only a series of very bad decision would cause the above scenario to happen, and yet it's almost exactly like how supply chains operate in physical space.
Depending on services is where the fragility starts. They're single points of failure, unless you have a plan to quickly switch to an equivalent competitor.
--
[0] - Unless they have a time-limited commercial license. Then they're essentially a service.
We are also not getting there anytime soon.