Excellent post, and generally I agree with you. I think nVidia is making the right strategic decisions and it is well positioned in those markets where it can grow substantially
At the low end, you have mobile gaming (phones, tablets, handheld gaming devices). This is a market that will probably explode in the next decade and nVidia is better positioned than anyone else to benefit from it. Qualcomm & Co are very capable SoC providers, but lack nVidia's experience in the gaming market, both on the software and hardware side. AMD is not even in the game, and intel can only offer a very mediocre solution on the cpu side. I believe nVidia''s tegra line has huge potential here, both short and long term.
A bit further up you find consoles. If you are Sony, Microsoft or Nintendo, what are your options for your next gen console ? Currently they all depend on IBM power for the cpu and mostly on AMD for the gpu side. IBM is still a credible provider for cpu tech and chips, but it has halted Cell development and has no solutions for the graphics side. AMD is good on the GPU side, but can only provide x86 on the cpu side. While x86 has some appeal as it would make game development across PC and consoles much easier, microsoft, despite owning directX and desktop gaming, still dumped it as x86 comes with a too serious price/power/performance penalty and furthermore, PC gaming market is very likely to become much less important than other gaming markets. Intel is in no position to compete here at all. Like AMD they could supply x86 cpu's but its trackrecord on gaming grade gpu's is simply appalling. Which leaves.. an ARM based solution, possibly with integrated GPU. If one is not powerful enough then perhaps several, or integrated GPU "crossfired" with a more powerful external one. That also would make downsize to compatible handheld designs like next gen PSP, DS and a Zune gaming device a whole lot easier. Who's better positioned to provide this than nVidia?
On the desktop gaming side, nVidia's short term future looks less than stellar. Its all you read on this forum, how much fermi will suck, and probably it wont be the best gaming gpu ever, but fermi will be untouchable in the highend x86 market in its quadro and tesla forms or for gpgpu. Thats where the big bucks are and were growth potential is still huge. Up until now, both nvidia and AMD designed gaming cards and then made derivatives that "happened" to be useful for highend CAD, GPGPU, HPC etc. Fermi turns this upside down, its a gpgpu product that can be used for gaming. If you look at margins and untapped potential (especially HPC), changing that priority may not have been a bad bet. Especially not when you factor in the software side where nVidia has a huge lead on AMD and intel.
IIRC, nVidia made about 1/3 of its revenue with geforces. As you noted, this market is in peril, and likely to drop further; so being able to expand and even dominate the markets where you make those 2/3 sounds like a good plan to me.
I may not like JHH's "style", but as a CEO I have to respect his vision.