Devs working at the vast majority of shops:

Easily handle production load on 3-5 servers / VMs while reading all about how to script quickly spinning up 1,000 VMs if they really needed to.

There’s different kinds of scale. I work for a company that has “only” hundreds of thousands of daily active users, but we ingest and persist a lot of data for them, on the order of millions of events per day for some of the larger users. We definitely run into big data problems in select services, despite not having a crazy number of users. Many companies are similar.

In the comments the guy says the hardware the benchmark was running on had 3TB of RAM.

That's not average hardware. It's a couple hundred connections max running select statements as fast as possible on the best hardware they could afford to test against.

That doesn't translate into your CRUD app handling a million requests per second.