Forbes has published a rather dense (and by dense, I mean packed with information, rather than mentally deficient) article by Kenneth Brill examining the explosion in data center construction and the associated cost increases from electrical consumption and cooling requirements.
The increase in energy consumption, despite the recent advances in "green" technologies, aren't exactly news, nor is the fact that many IT departments launch these initiatives with a less than complete understanding of total end-to-end costs for construction and operation. Brill does a good job of examining the reasons behind these disconnects; on the other hand, it's not clear to me why I'm reading this article today rather than a few years ago, before the terrain shifted beneath the corporateworld.
It isn't clear from the article whether the data center construction boom Brill cites is being driven primarily by enterprises building out their internal operations capacity, or by the ground swell in new, consolidated cloud computing service providers. If the former, then Brill's warnings and recommendations make a great deal of sense… but one would expect to see some discussion, at least, of the trend toward outsourcing services, whether it be overblown in his perspective or under-utilized. If the latter, then the arguments around efficiency, hidden costs, and the like seem to be overplayed… after all, the entire intent is to increase utilization and efficiency by consolidating the capacity for many businesses into a unified operation which can be run closer to capacity, and the cost factors are part of the cloud provider's business model, not an afterthought likely to be disregarded by an IT department with other irons in the fire.
I am genuinely curious, however, what the number breakdown looks like between individual enterprise data center projects and cloud provider driven projects. I suspect that much of the recent growth (certainly all of the individual projects which I am aware of) has been driven by businesses ramping up to provide services to other businesses, and that seems to me to require a different sort of contemplation altogether, something accounting for the difficulties in predicting total utilization, technical problems balancing such uneven loads, and the possible effects of so much capacity coming available on the broader market in such short order. But I'm no economist nor a particularly deep thinker, so you'll have to hope someone who is picks up on the idea and writes such an article… let me know if you see one.