Ryzen gave them a nice bump which made things look promising, but it would appear that the pent up demand for anything red has been satisfied ... global market share has dropped for significantly over last year.
Q3 2016 was at 17.50%
Q3 2017 was at 22.30% ... a nice boost (almost 5%)
Q3 2018 was at 20.70% ... lost 1/3 of the gain
The improvements are quite remarkable but the offerings are still "silver medalists" in most market segments. The drop in market share over the last year indicates perhaps that consumers are recognizing that "more cores" don't necessarily mean more performance. And with MoBo costs figured in, AMD is losing some of the "value" advantage.
The problem is, it was never in Intel's interest to market significant performance increases and since Sandy bridge, if we saw 5%, it was a lot. I think Intel were caught unaware by the media's / consumers fascination with more cores when 'more cores' was not bringing significant application performance increases to the table outside small market segments. Younger consumers graduating from consoles don't know much but when asking mommy and daddy for new PC, all the know is they want "Intel Inside" and nVidia card.
I think for AMD to make a real dent, they need a win ... they need the gold medal cause three months after the Olympics, nobody remembers the name of the silver medal winner... the gold medal winner, they remember, they are on the wheaties box. So when we look at TPUs test results, the fact that AMDs new thing went from 82% to 99% since the lest generation, is less significant than the fact that they finished behind the 1st place finisher.
One of the best lessons the market has given us is the history of IBM laptops. IBM put out the A20p which every year was on the cover of every PC magazine in the country. The $5,000 machine topped the charts in all tests and was always editor's choice. No company exec or person wanted to be considered for a promotion walked into an office and whipped out a lappie w/o an IBM logo on it. Most were not the A20p, that was ridiculously expensive. But everyone wanted that IBM Logo "to be taken seriously". At some point, a bean counter within IBM decided that the A20p line was not profitable ... at that price, they didn't sell a lot and the overhead burden for such small sales didn't make sense in "dollars and cents". But the next year, IBM was no longer on the magazine covers and soon after, IBM was out of the laptop market.
AMDs marketing has been ... "Ooh we have this" ... and "ooh we also have this".... and last round it was "we have more cores". That will win pre-order folks but unless you capture the performance crowns, that excitement can't be sustained. To make significant inroads into the gaming market for example, they need to lose the "When and if this happens, then we will be better".... they need a "look we won this one".
So what will they respond with ? .... my thinking is ... if it's not a win, it doesn't matter.