I tend to agree with this vision now. AMD could have the right product to get the x86 tablet market for themselves if they had developed a 28nm LP version of Brazos 2. AMD missed this opportunity and now they will have to compete with Quad Krait, Silvermont and Tegra 4. Not a nice prospect for their margins.
I saw this idea floated here and there and I think it is a good idea for shareholders, but only if you consider AMD FUBAR, and if you don't consider making a spin off, but a sale.
To make a spin off AMD would have to give ATI some cash before parting ways, and cash is something they don't have now, so we would need a buyer, someone to take a majority stake at ATI and pump money in the company.
The amount isn't exactly small, as ATI is in a very delicate position after entangling itself with AMD. Their GPU revenues are shrinking and too low, operating cash flows are low, their GPU are now overly complex because of the GPGPU burden AMD put in there, and they have no mobile presence, so they would need some serious cash to get back on their feet again, but I think this is feasible.
The other part of the problem is, how do you keep AMD alive after this spin off? Because right now AMD gets iGPU for free, something unacceptable for an independent ATI. AMD would have to pay royalties to this new ATI if not the costs for CPU + GPU integration at the die level and other specific designs they need for HSA. Given AMD current state, I doubt they could afford those new expenses.
It's easy to fathom this spin off when AMD goes for chapter 7, but not before that, and I understand you would like to see this spin off well before a chapter 7, so how do you think that this spin off would be feasible and how AMD could live without free iGPU?