This is why I think there is a very strong possibility (and I fervently hope it doesnt happen), that the next consoles will go ARM. The integration with mobile just seems too easy to pass up. Or maybe there will be two levels of console, a cheaper ARM one and a very high end x86 for VR type games.
I think the business model of Sony and Microsoft for gaming is fundamentally broken. Mobile gaming is growing faster, has lower barriers to entrance, a potentially much wider TAM and if some numbers are correct a higher ROI than traditional gaming. While Sony and Microsoft were eyeing integration with the PC ecosystem, which would help them to reduce costs of game development, they lost integration with the mobile ecosystem, while they were fighting to dominate the living room, where people spend 1-3 hours a day, they forgot about the people's phone, which is by their side every single hour of the day.
Sony and Microsoft built the XBO and the PS4 thinking "let's offer more of the same, graphics and fun on big TVs, but without the gruesome costs of the last generation" and not "let's offer new forms of fun to our customers", which is what the mobile gaming companies are offering.
In this context ARM or x86 is largely irrelevant, as Intel has proved that they can go low enough on costs and x86 Android tablets don't have performance problems if the opportunity arises. It's more a matter of which business model is going to win in the long run, e.g., which one is going to get a bigger slice of the investment pie, and it's quite clear that the dominant force in the future is more likely to be mobile gaming leveraging on off-the-shelf devices and on the cloud and not custom-built, closed ecosystem consoles.
It's not a threat for AMD also, I don't think other ARM companies are willing to go as low as AMD in terms of costs and margins, so if AMD is willing to swallow another round of 15% margins, the deal will be theirs.
The most existential threat to AMD today is their R&D budget, and what we are watching today with the current pathetic product stack is the results of Rory Read cuts in 2012. Rory crippled AMD product stack, he left AMD with the derp family for notebooks, 2012 leftovers on desktop, nothing on servers and the same old GPU architecture that wiped out their mobile GPU business. But the situation in 2012 was far better than it is today in terms of budget, so by the looks of R&D numbers, how do you think AMD product stack will look like in 2018 and how do you think about the odds of Zen and GCN2 making any splash on the market?