I don't think he's planning to start a foundation, he's donating to existing ones, like the against malaria foundation.
It's just that it's sub-optimal if he can actually run an org to solve the problems. Especially the risk of not increasing donation size with increased conspicuous consumption.
You can get soft quickly if you're not thinking lean...
If you want to do ROI calculations on problems spaces - there's a group that publishes strategies for these:
Copenhagen Consensus [PDF]:
http://www.copenhagenconsensus.com/sites/default/files/Outco...