I've spent the last few days researching and creating this graph (which explains the lack of recent updates).
It's based on historical data from the World Bank, plus a Bloomberg article that forecasts out to 2050.
I looked everywhere, but nobody seems to have forecast the world electricity sources beyond 2050. I was curious to know how the energy landscape might look during the second half of this century – so I extrapolated the data for another 50 years to 2100.
And here's the result. We can see that fossil fuels may have already peaked, with a turning point having been reached by 2010. Oil is no longer a significant source of electricity and is likely to disappear from the mix entirely by the 2050s. Coal has recently peaked and is now entering a period of terminal decline that is likely to see its phasing out by 2060 (this is the worldwide date; richer nations will doubtless achieve it well before the poorer nations). Natural gas will maintain a significant percentage share for much of this century before collapsing and petering out in the 2070s.
The big story is obviously with renewables. They will achieve a 50% share by 2050 and a 90% share by 2100. This may even be a conservative estimate if their high growth rate can be sustained beyond 2040, although most technology adoption follows an S-curve, so I'd rather stick with a "safer" prediction. Nuclear energy peaked in the 1990s and is likely to continue a slow decline for the foreseeable future – reaching barely 1% by 2100. However, it may see a renaissance if you include fusion. Lastly, hydroelectric will decline as a share of the world's electricity, but is likely to maintain a significant percentage throughout the century, averaging 15% or so.
I'd be interested to hear your feedback, and whether any major changes are needed. This will be going in our Data & Trends section, plus I'll add a prediction on our timeline for 2078, which is the year that fossil fuels are removed from the global electricity supply (I had originally forecast this for 2090, but after some new calculations, I think it may happen sooner).
P.S. The forum layout might compress/distort the graph, so here's a direct link (or just click the embedded image to enlarge):