I worked at Rocket Internet and the main problem they have is like any big company; a few very good people and a lot of useless people. Any proper startup worth their salt, competing with rocket directly, will absolutely slaughter them.
I really enjoyed working in Berlin for a few months though, it was excellent fun and I had an enormous apartment all to myself for about a third of the cost for the same thing in London.
There's a business right there - doing the first part of expanding startups in Europe, India, Africa, China, etc. I'm sure you could scale it pretty well. I don't know what the costing/charging model would be. I suppose Y Combinator has access to some of these things already but not in the way programmers would think about it...
* Translation that actually tries to transfer jokes/tone of a site and explains said translations in English. Maybe you could crowd source/do multiple translations.[1]
* Tax Rules, Incorporation, Local Laws - as an API where possible.
* People who could be good from local markets that could be useful. Hiring abroad must be very difficult.
[1] http://gengo.com/ looks like it does this bit I suppose, but I'd like to see more of a market and people being able to discuss pros and cons of each translation.
I don't agree. I think it is very hard.
You even mention some of the problems! Law, tax, language, getting 4 translations done quickly enough for you to release new things on a regular basis