A word of warning, Raja Koduri, formerly of AMD Graphics, is the same guy who massively over hyped AMD's lacklustre and very much wanting Polaris and Vega GPU's.
This may not apply to hyping up Intel's CPUs.
However, while Koduri may a super-hyper, to be fair we don't know what the brief was for Vega.
For GPU Compute it was quite good. Problem being that probably made it poor for gaming and AMD didn't have deep enough pockets to take the design, strip out the compute and make a gaming card out of like Nvidia do with Gx100 vs Gx102 etc., and AMD doesn't have the sales in GPU Compute to displace Nvidia in many places and make Vega worthwhile.
So whatever happens with Intel's GPU ambitions will be more interesting. They do have deep enough pockets to adapt the designs to the various markets.
Intel's approach to graphic drivers is about as good (bad) AMD's approach to providing GPU Compute tools, but traditionally Intel have been strong on certain software.
Certainly ICC has been responsible for deliberate poor code on AMD CPUs, but their optimiser is considered very good.
The questions about Intel GPU is are the serious and willing to put in the resources? Or will this turn into another one of Intel's big (expensive) failures like Itanium, Larabee, McAfee, Atom panic and contra revenue, 4GB modem each of which cost $billions?