That doesn’t seem right. For something so critical to many of their corporate mega-customers, there shouldn’t be any “third-party managed” infrastructure at all.

That’s not how the internet works. Every large provider has third-party connectivity. There’s no wire from Microsoft to your house.

You can buy connectivity from 5 diverse providers per data center, but if one of them continues to advertise your routes while not being able to actually pass traffic to your destination, you’re “down” to some people through no fault of your own.

I’ve had this happen with AT&T, XO/Verizon, and Cogent in recent years. It’s not uncommon. In all cases it was mis-configured ISP routers. Usually a support engineer admits a QoS or DDoS-protection configuration went wrong.

More so that there should be redundancy if it's a critical system.

"network infrastructure" - it'll be a fiber cut or failed router, together with a failover process that didn't work or was overlooked.