You know, I started this thread because I found it interesting that nVidia had such a poor quarter considering their market dominance. Of course the usual suspects derailed the thread. I don't need to mention any names but everyone knows who they are. Anyways, it seems like there was a bit of an air of desperation by nVidia.
Ignoring the costs of the stock option buy, it's extremely surprising that nVidia even lost money. nVidia certainly didn't want to lose many millions in order to gain a paltry market share percentage increase. I think that nVidia ovestimated their GT200 GPU's and underestimated how competitive ATI would be. Perhaps nVidia over contracted the amount of GPU's from the fabs and needed to do a bit of dumping. I certainly doubt we'll know.
One thing is for sure, ATI has gained a good deal of mind share, though that hasn't shown up in market share. Yet. If ATI is competitive on the same level at the mid and lower ranges then nVidia has a true fight on its hand again. While a process shrink to 40nm certainly helps nVidia cut costs, it also benefits ATI as well. Keeping in mind that CPU's and GPU's are planned years ahead of time and one can reasonably state that nVidia's new GPU is going to be a huge (if powerful) GPU while ATI will continue the lean but adequate route they took with the 4xx0 series. ATI doesn't want the single card crown, they want to be the "Value per $" winner. Especially among the lower and mid range GPU's where the most video cards are sold.
I think nVidia may have a very very tough fight ahead of it so long as ATI doesn't make a huge blunder in the short term, basically within the next year. In the long term of three years and further, and if they can stem the bleeding as they've done and even turn a profit (huge if's), they can be competitive once again but at the expense of nVidia in more ways than one. As I
posted here, I believe the future of ATI and Intel are multi-core CPU/GPU's that will take away most of nVidia's strengths in the GPGPU segment. They'll be taking away nVidia's discrete GPU business by essentially shrinking the discrete GPU business. ATI still requires a sound discrete GPU solution for this to work.
The strategy is sound. ATI might not gain market share percentage wise but they can generate more revenues by stealing from their own discrete GPU business as well as nVidia's share of the discrete GPU market. Intel also will steal some market share with their own CPU/GPU. The biggest loser potentially could be nVidia. Imagine a CPU/GPU hybrid chip on a 25nm or 32nm process that has a four core CPU as well as a four core GPU that is equal to today's top end CPU's and GPU's.
OEM's would also love the one stop shop for parts which will result in savings related to inventory management and training/hiring workers to work on custom ordered PC's.
A thought did occur to me on what is nVidia's likely response to a CPU/GPU threat besides the obvious ones of pushing CUDA and PhysX on the industry. They could build their own CPU/GPU. It's not as outlandish as it sounds. nVidia already sees the writing on the wall. Hence the failed nVidia and VIA merger, even if it is just rumors, where there's smoke there's fire. The nVidia & Via merger talks are making the rounds once again this year. Failing that, they have already licensed technology from Transmeta and likely cna make an x86 emulated CPU. Doesn't have to be great or beat AMD & Intel's CPU's. Just good enough for browsing and office related tasks as well as moderate gaming. A strong GPU coupled with a Transmeta styled CPU in a multi-core chip is certainly intriguing.
The problem with that is that AMD and Intel aren't likely to license their CPU sockets just so nVidia can compete with them (AMD & Intel). nVidia would have an insanely tough time convincing OEM's to adopt yet another motherboard standard for use in systems. Either way, things are set to look really interesting in the CPU as well as GPU markets in the next couple of years.