More to the point, those companies use lots of other software products, memcache among them, to improve the scalability of MySQL. It could be possible, if the numbers in the article are an indicator, that these companies are spending a lot more money on hardware than they would if they were using a more scalable database like PostgreSQL. They also may spend less money on engineering time if they didn't have to configure and manage all the software needed to get MySQL to scale.
People choose technology that they're more familiar with, not necessary the best out there.
Let's not forget that almost all of them use some sort of caching layer in front of MySQL, let's talk about that part of "scaling" shall we?
Last but not least, forking MySQL to suit their own needs do not necessarily imply that MySQL scales at their size, which is something you cannot claim.
Closing remark: let's say MySQL does scale, what if PostgreSQL scales even better?
I'm not a specific RDBMS fans since I use whatever dictates to me (with some exception) but really, the "name-calling" of who's using what should stop.
Facebook, Twitter, Yelp, LinkedIn, Flickr, Amazon, Craigslist, eBay, Etsy, Google, Groupon, Ticketmaster, Yahoo, 37Signals, DHL, Dropbox, Evernote, UPS, Kayak, LastMinute, Orbitz, Continental, Mint, Quora, Tumblr, Techcrunch, Slashdot, NYT, NBC, Reuters, Wotif, Zappos, Wikipedia, Youtube.
So that should calm any doubts about MySQL's ability to scale. Twitter and Facebook even have their own forks which are tailored to their individual use cases.