Uclue has now closed (more details). So long and thanks so much for everything.
Ask a Question | Browse Questions

ANSWERED on Sat 18 Aug 2007 - 12:17 am UTC by David Sarokin

Question: Energy Use of Internet

Please carefully read the Disclaimer and Terms & conditionsT&Cs.
Priced at $101.00

Actions: Add Comment



 17 Aug 2007 01:03 UTCFri 17 Aug 2007 - 1:03 am UTC 

What is the current total global energy consumption of the internet?

Included in this figure is the energy consumed by all the PCs in the world (when they are on!), as well as all the local hubs, regional data centers, company servers, the transmission lines run by telecoms and intermediate routers, etc. I'm not concerned with printers and other non-communicating devices. It should not include the energy needed to manufacture this infrastructure; but it should represent the energy required to run it and -- this is important -- the energy required to cool it. What is the utility bill for keeping the internet on? I would like this figure in the form of kilowatt hours, or ergs per time, rather than percent of total energy use, although that is nice too. And I need it for the planet.

Nearly a decade ago there was some attempt at this answer, but I am looking for current figures no more than a year or two old.  For instance this article (Emerging Technology: Energy Consumption And The New Economy http://www.networkcomputing.com/showArticle.jhtml?articleId=8703032) which was issued in Jan 5, 2001, refers to an article written in 1999 which used equipment surveys from 1995 in its calculations. There was a second attempt to re-calculate this in Feb 1, 2001, in this article, Research Finds Computer-Related Electricity Use to be Overestimated http://www.lbl.gov/Science-Articles/Archive/net-energy-studies.html. This was for US use only. Of course the internet is more global now, and more ubiquitous, more 24/7, more vital, so I am looking for a more current appraisal. It may have to be cobbled together. There is a lot of controversy over any kind of large-scale measurement like this, so I'll take a range if there are more than one estimate. And show your work!


David Sarokin 


 18 Aug 2007 00:17 UTCSat 18 Aug 2007 - 12:17 am UTC 


I've estimated the electricity consumption for the internet as follows:

US:    350 billion kWh per year

World: 868 billion kWh per year

These numbers represent 9.4% of total US electricity consumption, and 5.3% of global electricity consumption.


The breakout of the data is as follows:

Annual Electricity Use for the Internet--US and World

Category..........................US Consumption.......World Consumption
.....................................Billion kWh........Billion kWh

(1) Data Centers (includes cooling)......45.................112.5

(2) PCs&Monitors........................235.................588

(3) Modems/routers/etc...................67.................167

(4) Phone network.......................0.4...................1.0

TOTAL ELECTRICITY DEMAND        ~350 billion kWh       ~868 billion
OF THE INTERNET ............................U.S.................World                 


The data were derived as follows:

(1) These data come directly from a recent report on data centers:

Jonathan G. Koomey, Ph.D.
Staff Scientist, Lawrence Berkeley National Laboratory and
Consulting Professor, Stanford University
February 15, 2007

Note that the above report estimates global power consumption for servers as 2.5 times the US consumption...I used this scaling factor throughout my own estimates to "translate" data between US-scale and global-scale.

Also note that the report includes details on cooling and other data center needs, which are about 50% of total demand. 

(2) Desktop computer power consumption was estimated as follows:

There are approximately a billion PCs in the world:

Worldwide PC Adoption Forecast, 2007 To 2015
June 11, 2007
...Forrester's 2007 worldwide PC adoption forecast shows that there will be more than a billion PCs in use by the end of 2008 and more than 2 billion by 2015 — a 12.3% compound annual growth rate (CAGR)

and the average PC and monitor uses about 588 kWh or electricity in a year:

How much energy do computers and their monitors use?
...The average PC/monitor combination consumes 588 kWh of electricity every year

Therefore, global electricity consumption for PCs is 588 billion kWh per year (dividing by the 2.5 scaling factor gives an estimate for the US).

(3) Networking equipment such as modems and routers also have a significant power draw.  Data shows that networking components in an office setting account for about 1/4 the electricity demand of computers and monitors (that is, for every 100 kWh of demand from desktop PCs, another 25 kWh is needed for networking components):

PC Energy-Efficiency Trends and Technologies
[See piechart in Figure 1] 

I assumed this ratio held across the board, so that the data in (3) are 25% of the values in (2)

(4) Most internet data is transmitted over telephone lines, and telephone system power draw is fairly small.  This report:

Network Electricity Use Associated with Wireless Personal Digital Assistants

includes a table showing that total phone system use in the US is 3.8 TWh (3.8 billion kWh):

Table 1. Network and Phone System Direct Electricity Use

and another table detailing that internet use accounts for about 5% of overall phone traffic:

Table 2. U.S. Data and Telephone Traffic in 2000

Since VOIP technology has likely increased the proportion of internet-related phone traffic, I upped the percentage to 10% for my calculations.  Still, the overall contribution is fairly small.

I did not see data on other transmission technologies, such as cable or satellite, though these are presumably small as well.

In addition to the above figures, I used electricity consumption data for the US and for the World from the following sources:

US Electricity - consumption: 3.717 trillion kWh

World Electricity - consumption: 16.33 trillion kWh 


From the links you provided in your question, it seems clear that you're well aware of the controversies surrounding estimates of total elecricity demand of the internet.  One of the real oddities of the controversies is the almost complete absence of any attempt to add up all major components of internet-related demand, as your question called for.  There are many studies of individual components of the system (most of them quite dated), but no real attempt to get at the big picture.

The data I've presented here is no less subject to challenge than any of the other estimates out there.  But it is a plausible attempt at just such a "big picture", using the most recent and well-regarded data I could find.

I hope this meets your needs.  But if there's anything more I can do for you, just let me know by posting a Clarification request.

All the best,





 20 Aug 2007 18:59 UTCMon 20 Aug 2007 - 6:59 pm UTC 


Your answer is clear, on target, and as far as I can see, your work is accurate and reasonable. I am happy to accept it.


David Sarokin 


 21 Aug 2007 02:57 UTCTue 21 Aug 2007 - 2:57 am UTC 


Thanks so much.  This was a very challenging and very interesting assignment.  Let us know if there's anything more we can do for you on this.





 18 Sep 2007 20:55 UTCTue 18 Sep 2007 - 8:55 pm UTC 

How much do you estimate the global consumption will be in five years?





 28 Sep 2007 14:34 UTCFri 28 Sep 2007 - 2:34 pm UTC 

Unfortunately, the calculations shown above by David are erroneous.  The server and telephone numbers are correct, because I calculated them, but the PC numbers are wrong.  The ADL study referenced by Verdium, which was good for its day, was based on data from 1999 and 2000.  The Verdium people appear to have given the electricity use for a desktop PC with a CRT screen.  Two major things have changed since then:  There is a large and growing number of laptops in use (which use about 10X less power per unit than desktops) and the CRT monitors that dominated the stock in 2000 are almost all gone now (and the LCDs that replace them use 2-3x less power than the CRTs).  I haven't checked the network numbers yet, but it is often dangerous to take one number out of context from a particular study and apply it the way that it is done here.

It is difficult to do these calculations correctly, and I commend David for being clear about his analysis and assumptions, but as usual, the devil is in the details.

This particular issue was a very contentious one back in 1999 through 2002, and I urge the interested reader to check out http://enduse.lbl.gov/projects/infotech.html for more info.

I will monitor this particular thread in case there are more questions.  I may not be able to answer all of them, but I can at least point people in the right directions.

Good luck,

Jonathan Koomey, Ph.D.
Project Scientist, Lawrence Berkeley National Laboratory
Consulting Professor, Stanford University
Author, Turning Numbers into Knowledge:  Mastering the Art of Problem Solving




 30 Sep 2007 18:34 UTCSun 30 Sep 2007 - 6:34 pm UTC 

You may be interested in <a href="http://www.dgen.net/blog/index.php/2007/09/24/web-00-the-zero-energy-web/">this post</a> by Gavin Starks, laying down the challenge for a zero-emission internet, and the work he and d:gen have done in creating <a href="http://wiki.co2.dgen.net/index.php/Main_Page">AMEE</a> (The Avoiding Mass Extinctions Engine).

AMEE enables a common standard for CO2 footprint profiling and measurement, whose data is <a href="http://wiki.co2.dgen.net/index.php/Main_Page">open sourced</a>. The very small aim of AMEE is to make available all global energy consumption data. The data they have captured so far, I believe, is now the standard being used by the UK government.




 1 Oct 2007 00:26 UTCMon 1 Oct 2007 - 12:26 am UTC 

It looks like Prof. Koomey has a good point, but I do hate the expression "10X less".




 1 Oct 2007 17:20 UTCMon 1 Oct 2007 - 5:20 pm UTC 

Professor Koomey,

I am the person who commissioned this question. You are very clear and adamant that the totals calculated above are flat out "wrong." Normally the only way someone could say that was if they know what the correct figure is, or if they had an estimate of what the  correct figure is.

Can you supply me with a better, more accurate estimate (that is all these are)of the total energy consumption of the internet? What is your gut feeling?

In  the absence of such a guess I am forced to use David's guessimate. His figures could still be right, despite flawed calculations. I am comfortable with his calculations because while it is true that many PCs have LCD screens or are laptops, his calculations made no attempt to include PDAs and other handheld email/web devices (which are usually powered by adapters running 24/7), which would reverse the gains, either in part or in whole.

How much do you believe his figures are off? What do you believe is the correct figure?




 1 Oct 2007 18:29 UTCMon 1 Oct 2007 - 6:29 pm UTC 

Hi Kevin,

Thanks for your interest in this question.

To make an accurate estimate of the total power used by all computers is a complex and difficult task.  We did that work back in 1999-2002 when people were making claims like this, and found that the total electricity used by all office equipment and network equipment in the US was about 3% of total electricity use.  These findings were validated by independent studies by Arthur D. Little, the Energy Information Administration, the Environmental Protection Agency, the California Energy Commission, and the Rand corporation.

To show that a logical statement is erroneous I can show that the underlying assumptions (the premises) are incorrect, or I can show that the logic leading to the conclusion is incorrect (See Koomey 2001, below). I have shown that the data used for David's calculation for PC + monitor electricity use (which is the most important term in his calculation) is based on a study that is 5 years old that used data that are 7 years old.  In addition, it is clear that the data were cited out of context--it is simply wrong to use the estimated energy consumption of a PC circa 2000 and a CRT monitor and apply that to the total number of PCs in use nowadays.  If you go back to the original Roth et al. ADL study (as I did) you will see that what I am saying is correct.

And yes, other devices exist now in greater abundance, but the PDAs you cite are battery powered devices that use little power.

I am unable to give you a correct and updated number for the total today, but I know that the numbers given above are not correct and are likely an overestimate.  If I had $300-400k to do another study, I could give you an accurate answer, but failing that, I can only tell you that the numbers cited above are not right.

There were many people back in the 1999-2002 period who lost a LOT of money betting that  computers used 13% of all electricity and that this total would grow to half of all electricity use in 10-20 years.  I would urge you not to go down that path again.  If the % of electricity use associated with computers is important to you, I urge you to read the documents below
and contact me directly if you have further questions

It's probably good to start by looking at http://enduse.lbl.gov/projects/infotech.html

Good luck,


Baer, Walter S., Scott Hassell, and Ben Vollaard. 2002. Electricity Requirements for a Digital Society. RAND Corporation. MR-1617-DOE, ISBN 0-8330-3279-8.   http://www.rand.org/publications/MR/MR1617/

Blazek, Michele, Huimin Chong, Woonsien Loh, and Jonathan Koomey. 2004. "A data center revisited:  Assessment of the energy impacts of retrofits and technology trends in a high-density computing facility."  The ASCE Journal of Infrastructure Systems.  vol. 10, no. 3. September. pp. 98-104.

Kawamoto, Kaoru, Jonathan Koomey, Bruce Nordman, Richard E. Brown, Maryann Piette, Michael Ting, and Alan Meier. 2002. "Electricity Used by Office Equipment and Network Equipment in the U.S."  Energy–The International Journal (also LBNL-45917).  vol. 27, no. 3. March. pp. 255-269.

Koomey, Jonathan, Mary Ann Piette, Mike Cramer, and Joe Eto. 1996. "Efficiency Improvements in U.S. Office Equipment:  Expected Policy Impacts and Uncertainties."  Energy Policy.  vol. 24, no. 12. December. pp. 1101-1110.

Koomey, Jonathan. 2001. Turning Numbers into Knowledge:  Mastering the Art of Problem Solving. Oakland, CA: Analytics Press. http://www.analyticspress.com

Koomey, Jonathan. 2003. "Sorry, Wrong Number:  Separating Fact from Fiction in the Information Age." In IEEE Spectrum. June. pp. 11-12.

Koomey, Jonathan, Chris Calwell, Skip Laitner, Jane Thornton, Richard E. Brown, Joe Eto, Carrie Webber, and Cathy Cullicott. 2002. "Sorry, wrong number:  The use and misuse of numerical facts in analysis and media reporting of energy issues."  In Annual Review of Energy and the Environment 2002. Edited by R. H. Socolow, D. Anderson and J. Harte. Palo Alto, CA: Annual Reviews, Inc. (also LBNL-50499). pp. 119-158.

Koomey, Jonathan, Huimin Chong, Woonsien Loh, Bruce Nordman, and Michele Blazek. 2004. "Network electricity use associated with wireless personal digital assistants."  The ASCE Journal of Infrastructure Systems (also LBNL-54105).  vol. 10, no. 3. September. pp. 131-137.

Koomey, Jonathan. 2007. Estimating total power consumption by servers in the U.S. and the world. Oakland, CA: Analytics Press.  February 15. http://enterprise.amd.com/us-en/AMD-Business/Technology-Home/Power-Management.aspx

Mitchell-Jackson, Jennifer, Jonathan Koomey, Michele Blazek, and Bruce Nordman. 2002. "National and Regional Implications of Internet Data Center Growth."  Resources, Conservation, and Recycling (also LBNL-50534).  vol. 36, no. 3. October. pp. 175-185.

Mitchell-Jackson, Jennifer, Jonathan Koomey, Bruce Nordman, and Michele Blazek. 2003. "Data Center Power Requirements: Measurements From Silicon Valley."  Energy–The International Journal (also LBNL-48554).  vol. 28, no. 8. June. pp. 837 - 850.

Roth, Kurt, Fred Goldstein, and Jonathan Kleinman. 2002. Energy Consumption by Office and Telecommunications Equipment in Commercial Buildings--Volume I:  Energy Consumption Baseline. Washington, DC: Prepared by Arthur D. Little for the U.S. Department of Energy. A.D. Little Reference no. 72895-00.  January. http://www.eren.doe.gov/buildings/documents

US EPA. 2007. Report to Congress on Server and Data Center Energy Efficiency, Public Law 109-431. Prepared for the U.S. Environmental Protection Agency, ENERGY STAR Program, by Lawrence Berkeley National Laboratory.  August 2. http://www.energystar.gov/datacenters


Roger Browne 


 1 Oct 2007 19:36 UTCMon 1 Oct 2007 - 7:36 pm UTC 

Although modern LCD displays use less power than CRTs, there is also a factor changing in the opposite direction. As processor speeds increase, the processor is using more and more power, as is the GPU on high-end graphics cards.

A desktop computer could easily include a 480 watt power supply, which would have been uncommon in 2000. Similarly, typical laptop power consumption is higher now than it was in 2000.

Certainly, there has been a trend away from desktops towards laptops, but it's not clear to me whether or not this is happening fast enough to counteract the increase in average power consumption of those devices.




 1 Oct 2007 23:30 UTCMon 1 Oct 2007 - 11:30 pm UTC 

As I implied in my previous comment, this is ultimately an empirical question.  To accurately characterize total power use of computers will take serious research effort. 

It is true that some factors have been pushing power use up for high end machines, but many users do not need and do not use such machines.  What's the distribution of high end machine vs. Imac-like all in one devices vs. laptop?  To truly assess this we need more than anecdotes, we need measurements and data, and that takes research money and serious effort.
We did this circa 2000, but that's the latest data that exist.

Please note that most computers actually draw only 1/3 to 1/2 of the rated power (that's the power of the power supply, like the 480V cited above). So be careful in how you use the rated power in your calculations.

There are additional factors pushing power use in both directions not mentioned in my comments (increased enabling of Energy Star pushes it down, more computers left on more hours pushes it up, and so on).  I don't have time to go into all the subtleties of these calculations, but I wanted to warn Uclue readers about drawing conclusions based on old data taken out of context.  I know everyone wants tidy and complete answers, but alas, life isn't like that oftentimes.

Good luck,


David Sarokin 


 3 Oct 2007 11:36 UTCWed 3 Oct 2007 - 11:36 am UTC 


Uclue did for $101.00!

Seriously, though, the points raised here are all good ones, as is a key point made elsewhere that computers and monitors do lots of stuff that isn't internet-related, so it's an open question as to how much of their power draw should be described as energy use of the internet.

To say this analysis is in error, though, is a bit silly.  Of course it's in error...as are the analyses of every single study cited above.  All data is imperfect.  All data is out of date as soon as it is published.  All models are, at best, faulty reflections of the real world.

We simply do the best we can with the information that is available, and have to side-step on information that's missing (such as Google's contribution to overall data center power use!).  It's guesswork, but it's not bad guesswork.  It's simply a place to start.

I hope someone will see fit to post a follow-up question on this topic.  It's a fascinating area.





 4 Oct 2007 05:31 UTCThu 4 Oct 2007 - 5:31 am UTC 

Dr. Koomey,

Thank you so much for your invaluable assistance! Your expertise and your willingness to monitor this question for follow-ups are greatly appreciated.

I hope you appreciate that we tackle many tough, rather obscure and difficut questions here at Uclue, and we do our very best to be thorough and accurate. We were very lucky in this instance that you learned of this matter and were willing to pitch in and help.

Thank you very much!

Best regards,

Uclue Researcher


David Sarokin 


 7 Oct 2007 12:45 UTCSun 7 Oct 2007 - 12:45 pm UTC 

A bit of an update for anyone following this discussion.

The estimate presented here of the energy use of the internet has generated a bit of controversy.  This isn't all that surprising since, as the question itself notes, "There is a lot of controversy over any kind of large-scale measurement like this..."

The weakest link in my analysis is most certainly the 588 kWh estimate for typical PC electricity consumption in the course of the year.  As Dr. Koomey has noted, this is a dated estimate from a study by Verdiem, and does not likely reflect the current mix of computers in use. 

Some factors, such as greater use of laptops, and a shift to LCD monitors, argues for lower overall energy demand.  Other factors, such as higher processor speeds and greater graphics demands suggest increased power use. 

I took another look for any more up-to-date figures on this, but, as Dr. Koomey noted, this figure does not appear to have been updated since the Verdiem estimate.

I also agree that it would be good for a new study to be funded that could provide a more comprehensive and more up-to-date overview of electricity use of the internet than was possible in the context of this particular Q&A. 

I hope Dr. Koomey is successful in securing such funding, and my thanks to him, and to everyone else, who took the time to contribute to this very intriguing discussion.



Roger Browne 


 7 Oct 2007 16:25 UTCSun 7 Oct 2007 - 4:25 pm UTC 

This page became the subject of a discussion at Slashdot:

   Internet Uses 9.4% of Electricity in the US

Slashdot readers posted 270 comments. Here are some brief excerpts:

So then I guess you are saying that since bittorrent comsumes about 50% of the internet bandwith it consumes perhaps half 4% of the power ... By the same token spam is also a major consume of world power.

Perhaps a server farm should be located beside a restaurant and the waste heat be used to heat the french fry oil, which will ultimately go into some fool's diesel car.

One factor is if you want your home heated or not. That waste heat from the edge servers is heating homes and thus is an equivalent savings on the energy needed to heat homes. The opposite is true if you had the AC on. On the backbone all waste heat is working against the AC.

You know someone over at the MPAA or RIAA is going to spin this in a way that pits pirates as harmful to the future of the planet on an environmental level now too.

Where's the derivative factoid about World of Warcraft, a fictitious "country", using 10x more electricity than a real country, Vanuatu?

Many things have changed since early 2000 lowering the amount of power needed for the average home PC to operate. Most users in early 2000 were using CRT monitors which use almost 3 times as much power than a modern LCD.

And the average cpu uses a LOT more juice. So does the average video card. Who's buying all those 550 watt PSUs?

They shouldn't count PCs, they have many more uses than just the internet.

more [people] are browsing the web *instead* of watching TV? That would mean that TV power is going to PC's instead

John Hasler:
That's simply 99 gigawatts. "kilowatt-hours per year" is silly.

Or according to Einstein (and Google):
868 billion kilowatt hours = 3.1248 × 1018 joules
(3.1248 × (10 ** 18) joules) / (c ** 2) = 34.768089 kilograms
So keeping the current Internet running requires turning nearly 35 kilograms of mass into electricity.

Paul Carver:
Don't forget the vacuum cleaners used to clean the carpets in the buildings where the network designers and operators work, or the stereos that play music while people are browsing the net, or the electric lights that let the non-touch-typists see their keyboards at night ... unless they're somehow able to measure electricity used only while a computer user is actively viewing Internet content ... it's absurd to attribute the electricity usage to "the Internet".

I wonder how much energy is actually SAVED because of the internet, quick example: email. How much energy is used shipping a letter across the country?

It would also be interesting to know how much energy the Internet saves. For example instead of people flying around they talk on VoIP or have a teleconference ... Music and movies are downloaded rather than people driving to the shops for a disk.

It looks like they are assuming that if a PC is connected to the internet, that all electricity consumed by that pc, monitor, etc., is directly attributable to the internet.

surely it is more than that. The vending machine in my building is on the Internet. My phone is on the Internet. My laser printer is on the Internet, and in a way, I believe my cable box is too. Between infrastructure, servers, telecommunications, and end systems, a huge fraction of the electricity-using devices we interact with are on the Net.

868 billion kilowatt-hours per year = 10^11W=100GW
Space shuttle liftoff: 100GW

What percentage of the US food supply is used up keeping humans alive to maintain the internet? My God... this thing is a monster!

Also consider that today's power supplies are often >80% efficient, which is probably doubled in the last five years. In addition, Windows now implements CPU Idle functionality

my two 22" flat-panels probably pull about as much as my old 20" CRT did, if the heat coming off of the screens is any indication

My P1 200 didn't need a cpu fan (it had one, the fan died, it kept working happily away). my ati all-in-wonder didn't need its own fan. my motherboard didn't need a fan either. Today, all these need fans. More juice being used on a constant basis.

5% of total world electricity isn't all that much, considering we're talking about a worldwide network of computers ... It's less than lights and it's even less than you'd save if everyone in the world simply switched to power saving lightbulbs

Sarokin estimates the U.S. energy consumption of data centers (including cooling) at 45 billion kWh. The EPA Report to Congress [slashdot.org] in August on IT energy efficiency estimated that U.S. data centers used 61 billion kWh in 2006, so that part of the report missed by about 35 percent on the low side. Sarokin used a slightly older estimate from AMD.


Roger Browne 


 7 Oct 2007 16:29 UTCSun 7 Oct 2007 - 4:29 pm UTC 

Here's the report referred to by miller60 in the comment above:

Report to Congress on Server and Data Center Energy Efficiency




 7 Oct 2007 20:23 UTCSun 7 Oct 2007 - 8:23 pm UTC 

Thank you to all who have responded.

I have read through the comments, and references, including those provided by Koomey (and Slashdot!), and I am happy to say that David Sarokin's original answer still comes out on top. Here is how I would phrase the results;

The best estimate of how much electricity the global internet uses (in 2007) is 868 billion kwhs, or 5.3% of world electricity usage. Given the many unmeasured variables in this calculation, the actual answer might be less, more, or exactly the same. Until a more current estimate is compiled, this estimate remains the best estimate. If you have a more current or differently calculated estimate, I will be most happy to seriously entertain it. Simply suggesting that this estimate is incorrect won't displace it as the best estimate we have.


David Sarokin 


 13 May 2008 20:00 UTCTue 13 May 2008 - 8:00 pm UTC 

A bit of an addendum for anyone interested:

According to a presentation by the US Air Force:

Information and Communications Technology (ICT) Global CO2Emissions

information and communications technology (ICT) accounts for 2% of all greenhouse gas emissions worldwide (see the slide on page 33).  Almost 2/3 of this amount is from PCs, monitors and servers.

Put another way, the power consumption of the computers and other equipment making up the internet account for more than 1% of global greenhouse gas emissions, at least according to this presentation.

Just something to chew on...




 13 May 2008 21:31 UTCTue 13 May 2008 - 9:31 pm UTC 

Thanks, David, for the additional data point. It's useful to me.




 13 May 2008 22:48 UTCTue 13 May 2008 - 10:48 pm UTC 

"Something to chew on ..."
I put it in my pipe and smoked it, a now outdated expression and very non-PC practice  - more greenhouse gases.

Trying to relate greenhouse gas to energy use, I found this site that says that 82.3% of greenhouse gases in the States are from energy production:

Ignoring the confusing of US and world data, if ICT accounts for 2% of greenhouse gases, 82.3% of which are from energy production (of all types), then it seems that ICT uses ca. 1.6% of all energy, obviously a larger percentage of electrical energy, from the figures in the middle of this chart, 3 times as much, since electricity production is the source of 1/3 of the greenhouse gases in USA:

That would suggest that ICT consumes 4.8% of electrical energy  -
remembering that I am confusing world and US data.

Someone else can guesstimate about US ICT use relative to "world" and how the world mix of sources of greenhouse gases differs from that of the US.

But it probably isn't worth the effort, since that 2% figure may have been calculated by picking someone's estimate of ICT energy use, and we have seen that this is subject to debate.
AND, as Toonol points out, all power used by computers is not internet related.

If I got my numbers wrong, sorry, Myo


David Sarokin 


 15 Jun 2009 17:16 UTCMon 15 Jun 2009 - 5:16 pm UTC 


Can't resist these little updates every now and then:


"...According to a report from an environmental consultancy...some 62 trillion unsolicited e-mails were sent in 2008, using 33 terawatt hours of electricity. That is equivalent to the energy consumed by 1.5m American homes or 3.1m cars over a year. If generated by coal-fired power stations it would release 17m tonnes of carbon dioxide, some 0.2% of global emissions of this greenhouse gas..."




 15 Jun 2009 18:00 UTCMon 15 Jun 2009 - 6:00 pm UTC 

According to a talk posted on SlideShare in March, 2009
Mr. Koomey estimates the total world data center consumption of electricity for 2005 to be 1%-2%. (Slide 33) Since his figure does not include personal internet electric use, it seems believable to me the total might be double, bringing it to 2-4%. And since the total was forecasted to double in only five years, if I had to bet, I'd go with David Sarokin's original estimate of 5%.




 30 Mar 2011 21:24 UTCWed 30 Mar 2011 - 9:24 pm UTC 

No one was considering the electricity distribution factor yet, which is related to location, power wire quality and voltage. Data center should have less power distribution loss in  than PCs at home.


David Sarokin 


 12 Apr 2011 14:37 UTCTue 12 Apr 2011 - 2:37 pm UTC 

Every now and then, something comes along that I just feel I need to add to this question, such as a Microsoft study that cloud computing can reduce CO2 emissions 90% from small businesses, and from 30-60% in larger operations:

Cloud Computing and Sustainability

Other Microsoft environmental docs on cloud computing are here:







 12 Apr 2011 16:26 UTCTue 12 Apr 2011 - 4:26 pm UTC 

Thanks, David.  Given the importance of this subject I am surprised no one else has tackled a more rigorous answer to date. Your estimate is still in place.




 7 May 2011 13:01 UTCSat 7 May 2011 - 1:01 pm UTC 

thank you for this interesting thread. but is there any new data on the internet's global energy consumption? only ones i can find was published in 2009 or earlier. but since green it really got started about 2 years ago i'd need the latest studies on that topic.


David Sarokin 


 7 May 2011 13:39 UTCSat 7 May 2011 - 1:39 pm UTC 

This recent report from Greenpeace:


covers the energy costs and environmental impact of cloud computing. The report itself (and some of its reference material) has a lot of interesting and fairly current data.




 10 Dec 2011 03:29 UTCSat 10 Dec 2011 - 3:29 am UTC 

Is there historical data on energy usage for the Internet over the past 20 years that can be directly compared?

And, is there there any measure of per-browser/workstation use?


Rob Bowler 


 12 Dec 2011 23:42 UTCMon 12 Dec 2011 - 11:42 pm UTC 

I think this question has used up 1 billion Kw  :)


Uclue Admin 

 13 Dec 2011 12:48 UTCTue 13 Dec 2011 - 12:48 pm UTC 

Bowler, we can estimate a reasonable upper limit for this page.

According to the webserver stats, this page is accessed 400 times per month. The page "weight" is 75 kB, so the monthly bandwidth for this page is 30 MB.

Uclue's website is hosted at
where pricing is proportional to resource usage.

Uclue pays $0.25 per gigabyte for data transfer, so bandwidth for this page costs just under $0.01 per month.

Storage is charged in "byte-days", which for this page is 75,000*30 per month, i.e. 2.25 megabyte-days. This costs $0.02 per month.

Other costs include the associated email notifications, backups, DNS, database access etc. This page's share of those costs is under $0.01 per month.

The web host has held these prices for some time, so is presumably turning a profit. Therefore, the power consumption must be less than these charges. I don't know how much they pay for electricity, but it won't be less than $0.04 per kW-hour. Therefore the energy use of this page is no more than one kilowatt-hour per month.

This article is 52 months old, therefore this question has used up no more than 52 kilowatt-hours.


David Sarokin 


 28 Dec 2011 18:59 UTCWed 28 Dec 2011 - 6:59 pm UTC 

k2k et al,

This is an interesting article:

Implications of Historical Trends in the Electrical Efficiency of Computing

The full article is subscription-only, but the abstract I linked to will give you the gist of it.





 28 Dec 2011 19:53 UTCWed 28 Dec 2011 - 7:53 pm UTC 

Thanks. It is worth noting that one of the authors of the article on historical trends is Jon Koomey, who did the scholarly assessments cited above in this thread, and also joined in this discussion.


David Sarokin 


 23 Sep 2012 18:48 UTCSun 23 Sep 2012 - 6:48 pm UTC 

Couldn't resist yet adding yet another updated piece of the puzzle:

Power, Pollution and the Internet



David Sarokin 


 27 Jun 2013 14:46 UTCThu 27 Jun 2013 - 2:46 pm UTC 

Seems like once or twice a year, something new comes along to lend additional perspective. This latest report:

Power-Hungry Devices Use $70 Billion of Energy Annually

provides a lot of comparative detail on computers vs other devices in terms of overall energy use (apparently, ceiling fans are a surprisingly large energy hog).



David Sarokin 


 2 Apr 2014 18:24 UTCWed 2 Apr 2014 - 6:24 pm UTC 

In the ongoing saga, Greenpeace reports that global cloud computing accounts for 684 billion kWh of electricity demand:

Clicking Clean:
How Companies are Creating the Green Internet
April 2014

In the same ballpark as our earlier calculations. Here's a Time magazine report on the work:





Actions: Add Comment


Frequently Asked Questions | Terms & Conditions | Disclaimer | Privacy Policy | Contact Us | Spread the word!

Sun 18 Mar 2018 - 6:06 am UTC - © 2018 Uclue Ltd