One question I often get is why companies don't make efficiency update to their datacenters more regularly. Part of the answer is the complexity and associated fear of modifying an operating IT infrastructure, which is especially challenging in 24x7 environments. However, there's another set of issues.
Here's a typical scenario:
- A company buys energy efficient capital equipment now, which has a certain tax/accounting treatment
- The resulting savings come down the road in the form of reduced operating expenses, which have a different tax/accounting treatment
There are three natural problems with this inside companies:
The most basic issue is that the company needs to be in a position to be able to spend money today in order to save it in future years. Company cash flow challenges and investor pressure frequently lead to arguments against doing this, so it often comes down to a leadership decision.
The second issue is the split incentive, where the investment in efficiency is spent in group A's budget, and the savings come to group B's budget. Clearly, Group A needs an additional incentive to do this project, either a mandate from above, or an accounting kickback of some kind. This is even trickier when Groups A and B aren't part of the same company, as is often the case in today's complex real estate ownership and management structures.
Given the differences in accounting between capital and expenses, organizations frequently manage these through separate and distinct budget mechanisms. For example, a corporate group typically gets an annual budget for capital and separate one for expenses, and you can't move money back and forth between them. Getting exceptions to this often involves the finance leadership of the organization, so is not something that is casually pursued. This is not just a corporate problem, but also frequently found in governments and projects that they fund. For example, a university IT leader complained to me a few years back about how the federal grants they were operating under prohibited making efficiency tradeoffs because of the different categories of funds and restrictions on how they could be used.
Each of these cases can be overcome by someone who sits high up in the organization deciding to make an exception, or changing a policy to simplify certain classes of efficiency investments. But that requires a combination of motivated and enlightened leadership coupled with empowered change agents in the organization, and personally I haven't seen that combination occur in many places. And as mentioned above, this is especially tricky when multiple organizations are involved (e.g. the government and a university) and there is no shared, common leadership.
In the case of datacenters, we see companies in very different situations. In the case of a Google or Facebook, the datacenter and the company's business are tightly coupled. In contrast many organizations see IT as an auxiliary service within their company that supports their main business, but is separate. This typically manifests itself in a CIO and IT organization reporting to the CEO, separate of the "business functions". It's not hard to imagine that the former organizations would be able to bridge common efficiency funding challenges more easily than the latter organizations.