I’ve been hearing for a while now rumors to the effect that the internet uses a lot of energy, but I didn’t know how much “a lot” might be, how those figures compare to other emissions sources, or what we really ought to do about it. Today, I decided to do something about my ignorance and share the results with you.
Turns out, a lot of writers have had similar ideas–there are many articles about the carbon footprint of the internet, though they vary in their emphasis, their intended readership, and other factors. I’ve combined what I learned from them with the answers to some other questions I’ve gone looking for–I want to cover a bit more territory than the existing articles do.
The Carbon Footprint of the Internet
A carbon footprint is the total amount of greenhouse gas emissions associated with a given thing. Because there are multiple greenhouse gasses, each with a different potential for warming the planet, carbon footprints are usually expressed as tons (or tonnes or kilograms or whatever other unit of weight) of carbon dioxide equivalent (CO2e), that is, the amount of carbon dioxide necessary to cause the same amount of warming if it were the only greenhouse gas in play.
Carbon footprints are almost always estimates. That’s because in most cases, and certainly for anything as big and as intricate as the internet, actually identifying and measuring all associated emissions is impractical or even impossible.
Unfortunately, the articles I’ve found on the carbon footprint of the internet don’t explain how the estimates they quote were generated or how old the data they are based on are. Because these articles come from reputable sources (in the case of the following paragraphs, except where noted, the source is the BBC), I’m confident that they are at least in the right ball-park, which is good enough for our discussion here, but it’s worth noting the limitations of the information I’m presenting.
- Total global emissions for manufacture and use of “digital technologies” (a category that sounds somewhat larger than the internet as such) is 1.6 billion tons CO2e.
- So defined, the internet produces roughly 3.7% of humanity’s total greenhouse gas emissions. For comparison, the airline industry produces a similar share of emissions.
- Roughly 53.6% percent of humanity uses the internet–that’s about 4.1 billion people (probably far more than those who use airplanes).
- The average user’s personal internet-related carbon footprint is thus 14oz CO2e per year.
That last figure is interesting. After all, the same article gives an estimated footprint for a regular email (one without photos or large attachments) of 0.14oz CO2e. Does that mean that the average user’s internet use consists of sending just ten emails per year? Of course, one would expect some users’ footprints to be much larger than the average, but other users must have much smaller footprints, too. That’s what “average” means. Who are all of these people sending less than ten emails per year, yet somehow still counting as internet users?
This apparent discrepancy reflects differences in how the relevant estimates were calculated. Simply put, the per-email figure includes a lot of energy use that isn’t online, such as heating and lighting the building you are composing your email in. Such discrepancies don’t cause a problem for us so long as we remember these numbers are big-picture rough estimates. Treat them as precise (by, for example, attempting to calculate your online carbon footprint by counting the number of emails you send and multiplying by .14oz), and you’ll rapidly run into problems.
Onward we go.
When we’re talking about the carbon footprint of the internet, we’re talking about several different categories of emission sources.
Data Centers
Data centers are large collections of servers. They aren’t the only servers hosting the internet, but they have a large share of it, and are relatively easy to get information about (as opposed to tracking down everybody who has their own server for email and such in their closet). They use a lot of electricity, both to run the servers and to cool them–servers generate a lot of waste heat, and a group of them together could produce enough heat to damage themselves.
US data centers account for 2% of the country’s electricity. That doesn’t sound like a big number, but for a single element of a single industry, it’s pretty impressive. I don’t know how that number compares to figures from other countries or whether we have more or less than our fair share of server capacity based on how much Americans use the internet.
Globally, data centers account for 1% of total electricity usage.
There are two important things to remember. First, electricity usage alone doesn’t tell us the carbon footprint, since we don’t know how that electricity is generated. Are we talking coal-fired power plants, or are we talking wind turbines? We could make a rough estimate for each country’s data centers by looking up the carbon footprint for each country’s electrical grid, but a data center’s owner could well make a point of sourcing power sustainably, and some do. It’s also worth remembering that although internet usage is increasing, the energy efficiency of data centers is increasing rapidly as well.
So I don’t have a number for the carbon footprint of data centers. A reasonable estimate appears to be “considerable, but shrinking.”
Hardware
Building and shipping and then disposing of internet-related hardware (both servers and the various computers, laptops, smartphones and so forth using the internet) also has a carbon footprint (as well as other kinds of environmental impact). I have not found an estimate for this one, but you can find a more general discussion of the environmental impact of electronics, including certain aspects of their carbon footprints, here.
Use
In addition to the energy usage by servers, there is the energy usage by the various devices that are logged on to the internet or otherwise associated with it. At the moment, my laptop is one of them. The carbon footprint of its energy usage should be either zero or negative, depending on where our energy co-operative is buying its renewable power these days. We pay extra for the renewable option. Of course, the manufacture of my laptop had a carbon footprint, too, as will its eventual disposal.
In general, the user’s share of the internet’s footprint is harder to estimate yet easier to get a feel for because we see our own electric bills. Our bills don’t itemize which household devices used how much (though it’s fairly easy to figure that out), nor do they tell us how much electricity other people are using, but home and business electricity usage is not the black box that data centers seem to many of us. At least we know the right ballpark.
Based on my own home audit some years ago, unless a person is a serious electronics enthusiast, digital devices are likely a small but meaningful portion of home energy usage.
You can find information on at least American electricity usage here.
The Internet vs. the Alternative
It’s tempting to mentally compare the internet’s carbon footprint to zero, to look at its size (1.6 billion tons CO2e!) and think wow, that’s a problem, we’ve got to use the internet less! And it is a problem. We need to become carbon neutral as soon as possible. But not all problems are equal in severity, nor are all solutions equal in priority–and it’s possible the internet itself represents at least a partial solution to other problems.
To understand the real footprint of the internet, we’ve got to understand the footprints of what we’d be doing without the internet.
Communication
The internet is all about communication. Part of that communication is personal, or at least interpersonal, one person directly addressing another. These contacts might otherwise happen in person, by telephone, or by mail. So, let’s look at estimates for the carbon cost of non-internet means of connecting with each other.
- A letter has a carbon footprint of roughly 1oz CO2e. Yes, I know that figure is from a company that sells online greeting cards. A company that sells postal equipment had a similar figure–the two companies differed only in the importance they ascribed their figures.
- A text message has a carbon footprint of less than one-thousanth of an ounce, CO2e.
- A phone call has….Well, that depends on whether the call goes over the cellphone network or over the landline network. The former is about 4oz CO2e per minute, the latter is somewhat over 1oz CO2e per minute. What happens with a mixed-technology call I am not sure. Please note that a lot of phone calls are a lot longer than one minute!
- Meeting in-person has a very small carbon footprint, provided you don’t need to travel to see each other and don’t engage in any carbon-intensive activities while you are together.
So how does this compare to online communication? Online communication means email, video calling (though video calls can also go over the cell phone network, which does not count as the internet), and social media. I do not have good figures for any of these–yes, I have what looks like a per-email figure, 0.14oz CO2e, but remember that includes all of the emissions sources associated with sending an email, including heating or cooling your home, owning a computer, etc. Many of those carbon costs don’t go away simply because you’re not writing an email at the moment–and I don’t know whether the estimates for letter-writing and so forth include any of those costs. We might have an apples-to-oranges situation.
Video calling is even more confusing, for while figures for the carbon cost of these calls exist, they vary, and some are plainly wrong. It does seem reasonable to conclude that a video call is more carbon-intensive than an audio-only call made through the same equipment and the same network, since it requires more bits of information, and electronic information means electricity.
Social media use is confusing in a different way, for while there are apps that can estimate the carbon costs of using various social networks (here is a description of one, though I can’t tell how accurate the app in question is), it’s hard to know what non-internet activity to compare social media to. In many ways, these networks are truly new.
So how does online communication compare to its alternative? Most likely, a per-email carbon cost that could be fairly compared to the figures for letter-writing and so forth would be lower, maybe significantly lower, than 0.14oz CO2e, meaning email still has letters and phone calls beat, but is beat in turn by text messaging (not that all these formats are interchangeable–there are times when a text won’t do). Video is probably beat by phone and by letter (and by email), but video far and away beats traveling across country to meet in person.
Publishing
A lot of web content replaces printed periodicals, either directly (in the case of online issues of newspapers and magazines) or indirectly (in the case of a blog that might otherwise have been a magazine column, a newsletter, or a photocopied zine). Some sites replace printed books, such as encyclopedias, traveler’s handbooks, and phone books. Plus there are ebooks that can be read online. Of course, “replace” is questionable here, since it’s not clear in all cases that more stuff online really means less stuff in print, but let’s leave that aside for now.
I’ve compared print books and ebooks before. The latter came out ahead in some circumstances but not others. Since the greater part of a print book’s carbon footprint is its paper, magazines and newspapers probably have roughly the same footprint as books of the same weight. But the footprint of a printbook is spread out over a working lifetime that could easily last decades, maybe even centuries. Periodicals, in contrast, are basically single-use items (beloved collections of Natty Geos notwithstanding). I have not found a formal analysis of the subject, but unless I really intend to read a periodical cover to cover and keep it for reference, I’m probably better off using the internet.
Telecommuting
Telecommuting, considered very narrowly, obviously produces less CO2 than driving to work. Considered more broadly, it might not.
I’m working here off of an article that summarized an analysis of a large group of studies on the carbon savings involved with telecommuting. The analysis showed that the more rigorous and comprehensive a study was, the less difference it found between telecommuting and going to the office–and some studies found that in some circumstances telecommuting had higher emissions.
In general, the point of the article is that telecommuting isn’t just about what you stop doing (driving to work). It’s also about what you start doing. Additional emissions sources flagged by the authors include:
- Extra car trips because you can’t go shopping on the way home from work anymore
- Extra car trips because someone else in the household uses the car while you work now that you’re not using it
- More lighting and heating and cooling at home because you’re not out of the house all day (meanwhile, if the office isn’t closed, it’s also being lit and heated or cooled while you’re not there)
- Purchasing and using equipment, such as printers, for home use rather than using shared machines at the office
- Greater internet usage
- Lots of Zoom meetings (which may count as internet usage or may go over the cell phone network).
Whether the net effect is an increase or decrease in emissions depends on what you would have been doing if you were at the office, and whether the office is still open (and being lit and heated, etc.).
For further interesting discussions along these lines, try here and here.
Shopping
In general, if you buy one item from on online business that has no brick-and-mortar store and have it delivered, you will produce less greenhouse gas than if you drive to a brick-and-mortar store and buy the same single item yourself. But there are complications.
The trip to the brick-and-mortar can be made much more efficient by not driving and by buying more than one item. Online purchases get much less efficient if the item is returned or if it is miss-delivered the first time–most of the carbon footprint of an online purchase comes from the “last mile” part of delivery, the part where the delivery truck is headed to your house specifically, so anything that increases the number of trips that truck has to make increases the carbon footprint. Plus, returns are much more common in online purchases because you can’t try things on or otherwise examine them–and you won’t get lost on your way to your house, the way a delivery driver might.
Finally, whatever carbon advantage an online purchase may have is erased when shoppers select rushed delivery.
So on average online shopping has the smaller footprint, but there are circumstances under which shopping in-person may be just as good or even better.
Entertainment
Over half the world’s internet traffic is online video, accounting for about 1% of global emissions, or almost half the internet-related emissions total. Not all of that video is entertainment (and not all entertainment is video), but a lot of it is. Pornography alone is a third of all online video, plus there are music videos, online games, funny little video clips, and streaming movies and TV. Figuring out the carbon footprint for all of this is difficult.
For one thing, some internet video providers (such at Netflix) reduce and offset their carbon footprints. So while it may be possible to estimate the standard electricity demand of, say, an hour of video, there is no simple way to derive a one-size-fits-all carbon cost from that figure. For another, some of the popular estimates for the electricity demand of online video are very severely wrong. In fact, one oft-quoted group of figures both dramatically over-estimates the carbon cost of streaming video itself while under-estimating the electricity demand of the devices people use to watch the video (a demand which is the same when viewing video offline)–making it look like watching traditional TV is far and away better than streaming when it might not be.
Poking around on YouTube to pass the time clearly has a larger carbon footprint than, say, going for a walk around your block or having a conversation with your spouse. There are lots of forms of entertainment that are carbon-neutral. And it does seem reasonable to suppose that audio-only has a smaller footprint than audio with video, which means I should probably stop using YouTube as a music player (while ignoring the video entirely). If watching video isn’t important, it’s best not to do it and to choose some other pastime instead. But when watching video is very much the thing you want to do, it seems unclear whether watching online is better or worse than the alternative.
So How Does the Internet Compare?
In general, the internet seems to represent a carbon reduction over the offline activities it replaces, though there are some important exceptions (don’t watch online porn; have sex for real! Do it for the planet!). In context, the internet is mostly part of the solution, not part of the problem.
Can we make it into a better solution? Yes!
We can and we should.
Greening the Internet?
In researching for this article, I found a number of suggested ways to improve matters, including the idea that we ought to stop sending emails that say “thank you.” Apparently, these words are wasteful and unnecessary, and if everyone would just stop emailing them, it would be the equivalent of taking one car off the road, or some such.
No.
Not only is such advice based on a serious misunderstanding of the 0.14oz CO2e per-email figure, as noted earlier, but even 0.14oz really isn’t all that much. It’s possible to spend so much time and mental energy trying to shave off tiny personal environmental impacts here and there that one forgets to pay attention to the big picture, where the major impacts are. We are not going to end anthropogenic climate change by not saying thank you.
But it does seem reasonable to adopt a couple of climate-wise internet habits:
- Don’t waste time online. If you want to waste time, do it offline, where it won’t have as large a carbon footprint.
- Don’t waste other people’s time online. Don’t CC the whole group if you don’t have to, don’t send chain emails, don’t post stuff to social media just because you’re bored, and so on. While the carbon costs of online activity are mostly small, they are not zero.
- Don’t use video when you can use audio-only or text. This goes for both making calls and entertainment. Use video only when the video truly adds something to the situation, and use the lowest video quality that works for your situation.
- Don’t waste online storage capacity. Those 3,850 pages of emails hanging out in my in-box? The ones I’ve either read or identified as spam but haven’t gotten around to deleting or archiving offline? Yeah, I gotta do something about that.
- Don’t forget about the environmental impact of your devices. Wherever possible, use energy-efficient devices to use the internet and resist the pressure to replace your devices prematurely–the manufacture of a device could account for as much as 80% of its total lifetime carbon footprint, and the manufacture and disposal of electronics present other, serious environmental challenges.
- Shop wisely. When shopping online, choose slower shipping options when possible. Avoid having to return items by only buying what you really think you’ll keep. Shop through environmentally-responsible companies that are taking steps to minimize delivery-related emissions. If the package can’t be left on the doorstep, arrange to be home when it arrives so the driver does not need to try again.
But there are even better ways to address the problem.
Get Out the Big, Green Guns
The big guns, the heavy-hitters, for shrinking the carbon footprint of the internet involve policy and infrastructure.
It’s not that ordinary people can’t move the needle, it’s that sending fewer emails and other such changes isn’t the way to make a big difference. Even if every internet user on Earth, all 4.1 billion of us, took steps to make our internet use as efficient as possible, the electricity thus saved would likely be used for something else. Demand for energy rarely if ever goes down, provided the supply remains consistent (or increases!). But if even a small fraction of us demanded fossil-free electric grids and hyper-efficient electronic infrastructure, we’d get them. In fact, we are demanding, and we are getting, just not as swiftly as we must.
Demand better.
The large online companies have the power to improve the internet if consumers demand that they do so, and if government regulations and incentives support the change. If the big players want carbon-neutral data centers, they’ll get them.
Here are our marching orders:
- Find out the climate impact of the online companies you use. Reward those that are doing the right thing. Demand that others do better.
- Demand transparency and accountability from online entities, especially as regards energy usage.
- Demand similar environmental responsibility from manufacturers of electronic devices.
- Support climate legislation and the elected officials who enact it.
- Support regulations that favor the shift away from fossil fuels and the reduction of waste. Support the elected officials who enact such regulations.
- Vote.