Clearly Intel is concerned about losing market share. Maybe they should have thought about that after spending years twiddling their thumbs.
I get the sentiment, but their thumbs weren't entirely twiddling. The work that they did get accomplished is coming to fruition it seems, if rumors are to be believed, and unlike AMD who is currently relying on a tapped-out TSMC, Intel has fabs to spare, so to speak.
I don't expect that AMD has anything to worry about in the CPU space; I doubt Intel is going to be able to move to specifically dislodge AMD from the niches that they have created for themselves, for example, but in terms of AMD being nearly the universal superior option today, I don't expect AMDs time in the light to last.
The positive side of that is obviously that there really is competition, and it's intense, because AMD not only has to fend off Intel's newer releases, but they also have to fend off Apple for fab space just to get their products made!
Nvidia moving to Samsung, even with the disadvantages that that move has been speculated to incur, makes even more sense now; and Intel's move to push their nascent GPU line to outside fabs that have optimized for producing that sort of IC for others (TSMC, Samsung), also makes sense.
I used to think chip design was hard enough that it would limit interest... i guess not so, as everyone big enough seems to be making one. Then again, the market might shift yet again. I expect google to lead the way in cancelling projects, oh yeah after announcing an expansion, and swearing complete commitment. That's when you know there is one year left in the project.
Not sure about cancellation, but note that Intel has been a volume leader in 'GPU' shipments basically since they started putting them on their northbridges, and have been the undisputed marketshare leader since they integrated them into their GPUs as IGPs.
Intel has all the technology they need to target the lower end of the market, including technologies and ecosystems that AMD, their prime competitor in the cost-conscious side of things, cannot bring to bear. In particular, Quicksync is a monster. It's the most efficient means of enabling real-time transcoding on a desktop operating system today, and it is the most broadly-supported IP in the space. Even Nvidia plays second fiddle.
And when it comes to chip design, yes it is very hard, but remember that this is Intel we're talking about. In terms of resources that Intel has available to commit to graphics hardware design, they're second only to Nvidia and perhaps Apple. These three dwarf other competitors like Qualcomm and AMD and so on.
Intel getting into discrete graphics cannot help but be a joke, one that I've shared in (and will probably continue to), but a sober look at what they've done and what they have the potential to do does reveal interesting possibilities.
