Perhaps I was a bit too base in my comments. Perhaps I wasn't clear, or perhaps it simply needs to be reiterated. Nvidia cannot keep building gigantic, physically gigantic that is, chips. Yields will always go into the toilet with them making them an unsustainable investment. Anand did a wonderful article on this shortly after the HD4000s came out and did a comparison between the approaches Nvidia and ATI, and in turn AMD, took towards graphics development. This is the lesson that Nvidia is learning hard this round. It's not just going to 40nm that has shot them in the feet, it's doing so while building the most massive piece of silicon ever engineered for graphics processing. Wafers aren't getting larger, and as such, when die sizes increase but wafer sizes do not it takes fewer and fewer failures to lower yields dramatically. That's what I've been saying and why I feel it may very well be the most powerful chips ever made, but I'm questioning the costs involved. It's a massive investment on Nvidia's part for this architecture and it is only getting more expensive every day they don't deliver product to the market. In the end though, it's not the highend market that makes either company money. It's the mainstream market.