Originally posted by: BenSkywalker
Maybe I hit my head and didn't realize it, but reducing inventory levels before being required to clearnace them is normally considered smart business.
In order for the next gen to replace the current gen it has to be available to board partners the same time the previous gen is EOLed. If there's a 3-4 month lag between ordering GPUs from Nvidia and boxes hitting retail shelves then there will be at least a 3-4 month lag between whenever Fermi ships to board partners and when Joe Sixpack can buy them brick and mortar. There has been a deafening silence regarding when Fermi GPUs are shipping to board partners, so it's reasonable to assume no parts in the pipeline between the time part A is discontinued and part B is available.
The parallels to sudden socket 939 starvation of the retail channel in 2006 are too strong to ignore. That didn't go well for AMD.