The Rise of the North, data center edition

Facebook, the latest tech company to take the polar plunge, announced this week that it will build a data center just south of the Arctic Circle in Lulea, Sweden, where the average low in January is 3 degrees Fahrenheit.

If you had told me this fact and asked me the rationale, I would have said something about cheap electricity due to some combination of abundant hydro power and cheap Norwegian gas, but the real reason is more elegant.

Drawn by the promise of lower electricity costs, a growing number of tech companies are harnessing the region’s abundant cold air to cool their servers, cutting expensive air-conditioning out of the equation.

So the electricity savings, which Gartner believes could reach “tens of millions, if not hundreds of millions [of dollars], of savings per year”, are from quantity as well as (more than?) price. Who knew cold air was so valuable?

Hat tip MR, for both this news and an earlier pointer to The World in 2050, which I apparently never blogged about, but was fun to read and also put considerable meat on the bones of the case for why northern countries will rise in importance in the coming decades (better weather, increased agricultural productivity, abundant water, wealthy countries with good institutions, etc.).

Electric power supply-demand is nuanced

This post over at Marginal Revolution is a great example of why industry-specific knowledge is important, and why extrapolating from general economic principles can be dangerous.

Can you discuss whether [energy efficiency via smarter thermostats] can possibly work? As I understand the power industry, such a high percentage of the costs are upfront (with nuclear plants in particular, but with carbon burning plants as well) and the marginal price of producing energy (up to plant capacity) is so low, that falling demand would mostly cause plants to cut prices until they were again operating at capacity.

So “saving” energy at the consumer level won’t really reduce total energy consumption or gas emission.

As people familiar with the electric power industry already know, baseload power (e.g. nuclear) has low marginal cost but is also not supplying marginal supply - that role is played in most regions by gas turbines (and in some places by older coal or oil), which sell pretty much at marginal cost. A quick way to check this is by multiplying the price of natural gas by the heat rate of the marginal gas plant, and comparing that to the price per MWh of electricity - they will usually be pretty close.

The best commenters do a good job of explaining this. Ignore the first half dozen or so, Alfred and valuethinker are strong.

For what it's worth, U.S. industrial and commercial load is still below 2007 levels; energy efficiency standards from EPAct and EISA may be one factor, although obviously very difficult to parse out the effect of energy efficiency from other drivers like, say, the economy.