I hope it goes without saying that Yahoo!'s architecture is orders of magnitude more robust, complicated and scalable than Tumblr's. [1]

And of course they didn't "buy the architecture". They'll probably fold Tumblr into their own.

[1] Nothing against Tumblr but Yahoo operates at a scale much larger than Tumblr so their architecture isn't of much use.

It's not the first article on HighScalability named in the form of 'The service_name Architecture company_name Bought for a Cool some_amount Dollars'

Nevertheless, it's really good to get an insight in these stuff and I like these articles from them.

Given how they say it will stay independent, why do you reason they will ot simply keep using their current stack?

Ex-Yahoo. Yahoo has infrastructure for a bunch of things which are really beneficial. Abuse detection/prevention, internationalization into like 40 languages and a ton of scaling services - to name a few.

Tumblr might not need them but Yahoo! might feel like they'd benefit from all of it.

If I were the Tumblr team I wouldn't want to deal with the man hours to get onto Yahoo!'s platform but I'm not sure how much of a choice they have.