Their paid managed service works well for enterprise SaaS companies who need < 3-4 TB of storage max ever per customer cluster and have enterprise scale users (<100k users).

It also works well for early stage Consumer Internet startups that haven't yet achieved huge growth but care a lot about developer productivity while churning features at high velocity.

But it gets blamed for reliability issues suffered by highly successful Consumer Internet companies that have achieved scale and have DAU, MAU in > 10M and have lot more than 4TB of data in a single cluster.

But that's basically a good problem to have at that stage. Usually such a company would have many more problems – monolithic application with monolithic database with huge unmanageable schema and indexes gone wild etc. Usual solution at that stage would be some sort of rearchitecture towards Microservices with multiple specialized databases for different use-cases – usually cloud hosted managed databases with horizontal scalability dedicated for online user-path workloads and separate OLAP tech stack for offline ETL/analytics workloads.

At work I have a multi-petabyte cluster in MongoDB Atlas with no issues. The managed service is fantastic; I'm going to need some citations on how it degrades after a few terabytes. I don't really like MongoDB's document-model-everywhere approach but I have to respect how well their managed service works.

Grossly misinformed post. 4TB (compressed) per shard may be what you're thinking of?

There was many new node projects using MongoDB and I remember several failed, but details elude me.

There was a Slack competitor using MongoDB that was snappy and js hipster trendy that ultimately failed because of unreliable database IIRC.

How many crypto currency projects went down with $million+ losses due to using mongoDB as the backend?

they did their marketing right.