Not at all. Like I say above, right tool for the job. I don't doubt that all those companies use MySQL for something, indeed probably every large company in the world has at least one instance of it somewhere.

For example, taking the first company on your list that actually deals with real stuff in the real world rather than just running a website, DHL, here I see http://www.oracle.com/us/corporate/press/037809 - they're an Oracle shop. I could continue but I think I've proved my point.

PS the cost of not losing "ANY data for ANY reason" is effectively infinite. You make a call on what data loss would cost your business then you decide how likely each scenario is and that gives you your budget. Datacentre disruptions I have been personally involved with include workmen digging up fibre (several times actually), deliberate vandalism of cable (during May Day riots), fire in the generator room (a couple of times), large truck losing control and driving through the walls, flooding, and several more. No one of these scenarios was sufficient by itself to hurt us because we had mitigated them. But 2 happening together, then yeah, there is a significant possibility that some data loss would occur (e.g. replication to datacentre B is disrupted, then datacentre A is catastrophically lost). The cost of protecting against that however is prohibitive.

So before you get all starry-eyed about these amazing companies and their amazing MySQL installations, understand that you are starting from a false premise. Oracle - for a cost - provides some powerful capabilities (e.g. stretch-RAC[1]). Is it worth the money? Well, it comes back to what your data is worth. This incidentally is why some organizations cling to paper forms - it's cheaper to re-key the data than build out the infrastructure to do it all online...

[1] Tho' if you think you need this, what you probably really need is DB/2 and Sysplex