The problem in a nutshell:
It seems safe to assume that YouTube’s traffic will continue to grow, with no clear ceiling in sight. Since the majority of Google’s costs for the service are pure variable costs of bandwidth and storage, and since they’ve already reached the point at which no greater economies of scale remain, the costs of the business will continue to grow on a linear basis. Unfortunately, far more user-generated content than professional content makes its way onto the site, which means that while costs grow linearly, non-monetizable content is growing geometrically as compared against the monetizable content that YouTube really wants and needs to survive. This means less and less of YouTube’s library will be revenue-contributing, while the costs of delivering that library will continue to grow.