His actual prediction is beside the point - he laid it out in enough detail for the reader to intelligently disagree, if they had better information on the drivers he identified.
I quite clearly read his comments in square brackets as his predictions.
"The addressable market for modular solutions will have to grow much faster than that of inter-dependent vendors. [The chart shows in-line growth--again due to the overall growth of the market. The green guys above won't just roll over.]"
"The split of licenses between Android and Windows Phone within the addressable market must favor Android. [Again, this is unlikely]"
And his conclusion:
"In the end, the real limitation is that the brown area itself is not going to get disproportionately large. Android is limited by its addressable market and that market is a lot smaller than the whole market."
That's a clear, and clearly wrong, prediction. He lists a whole bunch of specific things that could go either way and predicts one consistent direction for all of them and they all came down the other way. That's a comprehensive failure.
He calls Android dominance "wishful thinking" and links that phrase to another of his pieces in which he says this:
"That simply means that no single platform can win a disproportionate share because it would threaten the balance of control the operators require. So talk of “dominance” of one platform or another is hyperbole. The most likely scenario is an even distribution of share between 4 or 5 competitors, so I expect iPhone and Android to get 20% share each."
So he was right about the iPhone (though they seem to have peaked), and wrong about the other 3/4 competitors in the market.