I think you can look back and see some things that Google has done successfully. Youtube turns out to have been a good buy. Google successfully managed the transition to mobile, which many questioned whether Google would be able to do ("Mobile first"). Most recently, Google has made a huge bet that their future is intrinsically tied in with machine intelligence ("AI first"). We'll get to watch over the next 5-10 years how that plays out.

(It's probably already clear that there's a lot of power in an AI-first approach, but it's worth noting that this change didn't just happen with Sundar's blog post last month. A better way to look at it is to realize that if Google has been running their custom machine learning ASICs for over a year now, they must have started designing them well before that, which, to me, was amazingly early. Google acquired DeepMind in 2014.)

Not all those bets will, or have to, work, as long as some of them work out really well. I'm curious to see what happens with VR, for example. But I think AI is the most obvious central bet, in the same that that mobile was beforehand.