Estimating the amount of energy the Internet uses is no small task. We have to take into account everything -- from the embodied energy of Internet-connected devices like smart phones, laptops, e-readers, desktops, cables and wires and of course the servers themselves, as well as the energy consumption of the servers and devices and more. It's a huge task, but two researchers from University of California, Berkeley, Justin Ma and Barath Raghavan, came up with an estimate they think is reasonable.
The study, called "The Energy and Emergy of the Internet" (PDF) looks at a whole slew of information. The team's estimates include 750 million of each desktops and laptops; 1 billion smartphones; 100 million servers; 1 million routers and router-like devices; 100 million LAN devices; 5 million cell towers; 75 million telecom switches; 1.5 billion km of fiber optic cabling; and 3.5 billion km of copper cabling for global telecommunications. Estimates on embodied energy of devices are based on previous studies.The team concludes that the Internet consumes between 170 and 307 GW, which is equal to between 1.1% and 1.9% of the 16 TW used by humans worldwide. Embodied power is responsible for about 53% of that total.
The team concludes that there are some obvious ways to reduce the energy consumption of the Internet, which includes reducing the embodied energy of new devices, as well as reducing the number of new devices by keeping older devices functional longer. They state that doubling the replacement timespan for all components they listed in their estimates could reduce the Internet's embodied power by 43-82 GW. That's a hefty amount of the total.
But again, the team estimates that the Internet is roughly less than 2% of the total energy use. They note that transportation takes up a far more substantial amount, and that there could be a beneficial tradeoff of increasing the Internet's use in reducing the need to travel as much. A little boost in the Internet's energy consumption could mean a big reduction in energy consumption in planes, trains and automobiles.
The team writes:
First, suppose we replace some fraction of business air travel worldwide with teleconferencing. Each year there are 1.8 billion air passenger (one-way) trips; suppose 25% of those trips are eligible for elimination and are replaced with video conferencing. This yields 400 million passenger trips eliminated yearly, each of which uses roughly 20 GJ, saving 285 GW total. Thus, by replacing one in four plane trips with videoconferencing, we save about as much power as the entire Internet, and in particular we save a lot of oil.
Their conclusions are not particularly novel or surprising -- we've been aware for a long time that reducing the consumption of devices, keeping old devices useful, and using recycled materials in the construction of new devices are key parts of keeping the carbon footprint (and therefore energy footprint) of electronics low. And Greenpeace has been on the IT industry for years to acknowledge its impact on the environment and move toward using renewable energy for data centers and improving the energy consumption of the data center as well as the servers it houses. Not to mention boosting the use of the Internet for communication and virtual meetings to help reduce how much we travel. But the study's methods and results still spark critical thought about the impact of the Internet -- something most of us couldn't imagine our lives without.
One thing will catch your eye in the paper -- a sentence that simply states, "Although we are certain our answer is wrong, we hope to raise awareness on the study of this important topic."
If there's two things this study leaves us with, it is gratitude for a solid guess and curiosity for what the real answer actually is. New Scientist reports that the research will be presented next month at the Workshop on Hot Topics in Networks in Cambridge, Massachusetts.
Follow Jaymi on Twitter for more stories like this