R is not built to be super performant at all. It can be very performant if used appropriately. R is also not built for analysing large amounts of data but for applying statistical methods to data - it is and can be used though for handling very large data sets.
I also don't see how this headline can keep its promise in general. The example simply features an R package which is not designed with high performance in mind.
If you need performance - f.x. for applying GLM methods - then I'd recommend to resort to something like h2o: http://0xdata.com/h2o/algorithms/