February 1, 2001

Berkeley Lab Science Beat

Lab website index

Lawrence Berkeley National Lab home page

Search Lab science articles archive
 Advanced Search  
Search Tips

How much electricity is used to power information technology in the U.S.? Berkeley Lab's Environmental Energy Technologies Division's (EETD) End-Use Energy Forecasting Group, led by Jonathan Koomey, has released a new study estimating office and network equipment energy use as of the end of 1999. The last major study examining energy consumption by this group of products was also conducted by this group and was published in 1995, just before the explosive growth of the Internet and the Web. The new study, "Electricity Used by Office Equipment and Network Equipment in the U.S.," was coauthored by Kaoru Kawamoto, Jonathan Koomey, Bruce Nordman, Richard Brown, Mary Ann Piette, and Alan Meier, all of Berkeley Lab.

Researchers conclude that office and network equipment consume about three percent of the total U.S. electricity use.  Their findings challenge widely reported claims that this equipment uses 13 percent of the total.

The EETD team divided office equipment into 11 types. For each type, they estimated the annual energy consumption for residential, commercial, and industrial use by combining estimates of stock, power requirements, usage, and saturation of power-management technology. They also estimated annual energy consumption for six types of network equipment.

The latest analysis shows that the total electricity used by office and network equipment is about 74 TeraWatt-hours (TWh) per year, about two percent of the total electricity use in the U.S. When telephone switching equipment and manufacturing energy for semiconductors and computers is included, the total goes up to about three percent of all electricity use in the U.S.

More than 70 percent of the 74 TWh is used by office equipment in the commercial sector. Network equipment (excluding telecommunications equipment) uses less than five percent of this total, or about 3 TWh.

Power-management features found on many devices including computers, printers, and fax machines currently save about 23 TWh/year. Complete saturation and proper functioning of power management would save an additional 17 TWh/year, and shutting down all equipment not required to operate at night would save an additional 7 TWh/year beyond this. The table below summarizes the results for the 11 categories of office equipment.

Best estimate of annual electricity used by U.S. office equipment in 1999, TWh/year:

Equipment Type Residential Commercial Industrial Total
Portable Computer 0.14 0.13 0.02 0.29
Desktop Computer 2.67 10.21 1.46 14.34
Server 0 1.60 0.23 1.83
Minicomputer 0 8.86 2.95 11.81
Mainframe 0 5.62 0.63 6.25
Terminal 0 1.83 0.61 2.44
Display 3.13 9.82 1.40 14.35
Laser Printer 0.10 5.36 0.77 6.23
Inkjet/Dot Printer 1.10 1.56 0.22 2.88
Copier 1.10 5.71 0.82 7.63
Fax 0.44 2.26 0.32 3.02
Total 8.7 53 9.4 71

Results Challenge Current Thinking

In 1999, Mark P. Mills published a report for the Greening Earth Society, "The Internet Begins with Coal" (later summarized in an article in Forbes magazine), that attempted to calculate the "Internet-related" portion of electricity use. The Mills study differs in a number of ways from the EETD study. One significant difference is its attempt to isolate the electricity use associated with just the Internet, not the total electricity use of office equipment such as PCs and laptops which may or may not be connected to the Internet.

The Mills report claimed that electricity use associated just with the Internet totaled about eight percent of all U.S. electricity use in 1998 and that it would grow to half of all electricity use in the next two decades. The EETD research team examined the assumptions and conclusions of the Mills report in a technical memorandum, and raised a number of questions about its methodology and starting assumptions. One of the questions focused on possible overestimates of the power used by particular kinds of equipment. Mills assumes, for example, that the active power of a personal computer plus monitor is 1000 watts, when the measured data for a Pentium III PC with a 17" monitor show total active power use of 135 watts. Of course, most PCs and monitors bear the Energy Star label, which further reduces the power use when the computer is in standby mode.

The EETD group concluded that Mills significantly overestimated electricity use, in some cases by more than an order of magnitude. Adjusting the Mills estimates to reflect measured data and more accurate assumptions reduced Mills' overall estimate of total Internet-related electricity use by about a factor of eight. "The explosive growth in electricity demand that Mills alleges simply does not show up in the data, either at the micro level as estimated for each type of equipment, or at the macro level as measured by aggregate statistics on total electricity use," says Koomey.

Various other researchers, including Jay Hakes, formerly of the Department of Energy's Energy Information Administration, the Rocky Mountain Institute's Amory Lovins, and Joseph Romm, Executive Director of the Center for Energy and Climate Solutions, have also raised serious questions about the Mills study's methodologies and results.

Beyond estimating the direct energy use of office and network equipment, Koomey's group is beginning to assess the indirect effects of the Internet on resource use in the U.S. economy. These effects include structural changes (where new institutional arrangements and technological capabilities become possible because of the Internet) and substitution effects (where the new technologies substitute for established energy uses). An example of a substitution effect would be the use of Internet shopping in place of routine trips to the grocery store or the mall, which might reduce personal transportation energy use. A structural change scenario might involve a decline in brick-and-mortar retail stores in favor of warehousing of retail goods for direct home delivery.

The End-Use Energy Forecasting Group is currently analyzing energy use of emerging technologies such as Web phones, handheld computers, and Internet terminals, which are not currently included in the Berkeley Lab estimates.

This article was reprinted from Berkeley Lab's Environmental Energy Technologies Division Newsletter.

Additional information:

Media contact:

Percentage of the total annual U.S. office and network equipment electricity use of 74 TerraWatt-hours attributable to each sector at the end of 1999