Wouldn’t you know it, the day after I review Mark Mills’s analyses (in 1999 and 2011) of the digital economy as a key driver of demand growth for coal-fired electric power, I receive an EnergyFactsWeekly in my email box featuring new analysis by Mills on that very topic. It also contains links to two other related commentaries by Mills.
In The Efficiency Wall and the Future of the Internet’s Energy Cost, Mills reports that “the historic gains in computing energy efficiency started slowing down in 2005″ due to the “inherent physics” of existing chip technology. During the same period, however, ”the growth in global traffic on the Internet has continued rising at the same old staggering exponential rate.” The upshot? “This combination arithmetically guarantees a higher growth rate now in the total energy consumed by the Internet.”
Computing efficiency gains are rapidly approaching an “asymptotic wall” much as the power of jet engines and cruise speed of jet aircraft did in 1960.
Jet engine power (measured in terms of the critical aviation metric, power per unit of weight) rose exponentially for the 20 years after invention, then hit a wall dictated by the inherent physics of the engines and materials. Consequently, the average cruise speed of jet aircraft also hit a wall.
“But,” notes Mills, ”there is a critical difference between aviation and digital traffic: the former rises linearly with population and wealth, while the latter grows exponentially as new applications continue to explode for Big Data.”
New materials and technologies are improving the energy efficiency of computing, but, says Mills, not enough to halt the growth in aggregate demand. He concludes with two predictions and a policy recommendation:
- Digital energy consumption will rise, locked into the physics of supply and economics of demand, and
- Energy costs will be increasingly dominated by factors external to the Internet, especially the cost of electricity. Cheap power will matter even more in the future.
We return to a familiar refrain? Dig more coal.
In Will the Cloud’s Efficiency Cut Energy (and Coal) Use?, Mills reviews Green Cloud Computing, a study by researchers at the University of Melbourne, in Australia. Key finding: Although “concentrated, shared and optimized use of massive computing assets is very efficient,” ”more energy is consumed in the Cloud, than on a PC, when the user accesses the Cloud frequently, or does high ‘intensity’ tasks.”
In Welcome to Earth Where Networks Gobble Power — and Coal, Mills examines whether improvements in the energy efficiency of cell phone networks will reduce overall energy demand and coal consumption. He looks at what has happened in China, where network energy efficiency has improved by 50% since 2005.
“As the above graph illustrates,” writes Mills, ”as the number of users expands, and the amount of traffic on the network expands even faster, energy use grows rapidly even with – or, as I have argued elsewhere in these Commentaries, because of – improving energy efficiency.”