The guy's analysis if so-so as he views pretty much everything AMD has done in the past years as horrible designs when that simply isn't true. No one can dispute that the FX CPUs are decent performing CPUs. Where he finds them to be terrible is in comparison to Intel's CPUs using a smaller node size and less power. Most consumers don't care what node size is used, they care about performance and cost. That's why AMD has sold a fair quantity of FX CPUs. He also has some misunderstanding of Zen in regards to core count and the 14Nm FinFET process. The one point he did get correct is that if AMD's Summit Ridge desktop CPUs are close in performance to whatever Intel is selling, AMD's going to hurt Intel sales significantly as AMD will be able to sell their CPUs for a tiny bit less and offer better actual CPU performance per $. THAT is a big deal for enterprise. At one time AMD owned a sizeable chunk of enterprise Biz and you can expect them to own a lot more with Zen cored products.
Personally I think AMD is playing their cards close because they have a excellent CPU design that will create a lot of sales and offer consumers a real performance alternative not just a less expensive alternative. AMD started from scratch with a clean sheet of paper with the goal to develop the best CPU possible. That's a daunting task but from all indications of the first batch testing, Zen has met all design parameters without issue. That is very telling, IME. AMD's biggest challenge may be supplying chips fast enough. Samsung can essentially double the output from GloFo so the good times may be not too far off for AMD and their customers.