There seems to be a bit of a chasm developing between RampRate's analysis of YouTube's cost base and that analysed by Credit Suisse earlier this year.
I can't help feel that both are wide of the mark. There are a number of costs they have not focused on:
Legal costs - YouTube must have sky high legal costs considering they have to defend their service in practically every country on earth against an increasingly litigious content industry.
Power costs - the cost of powering and cooling YouTube must be huge; a back of a fag packet calculation based on January figures indicates that YouTube need around 5,000 servers in the US alone to cater for their service, and this does not bear in mind storage.
Peering asymmetry - sure there are peering agreements in place, but ISPs would have to be dumb to accept the asynchronicity of traffic from Google, so I suspect RampRate have applied industry norms to a very specific instance.
One of the biggest issues I encounter in business is that companies fool themselves by looking at costs of sales and not factoring in overheads, and I think there's a case of this here.
At the same time, estimates of YouTube's revenues are, IMHO, vastly exaggerated. I'm not a heavy user of the service, but I can never remember seeing an ad beyond the usual Adsense slots, and the initiative to build corporate subsites seems less than successful.
There is little doubt, in my opinion, that YouTube remains hugely loss making.
It's disappointing that the company that made monetising the long tail has failed to apply this model to video. Indeed, it confounds me that the ostensibly brightest company on earth is so out of its depth..