How much CO2 emissions could a country save by deleting unused data

Have you ever wondered how much greenhouse gas emissions can you save by deleting your unused data? 💾💨And what if your entire country did it?

I recently discovered that Slovenia is running a new community initiative titled "Digital Cleanup Day", where the goal is to delete data that's sitting idle and unused on our devices, and in cloud services. They have some big backers from the industry and even the Ministrstvo za digitalno preobrazbo / Ministry of Digital Transformation is on board. It’s part of a wider Digital Cleanup Day initiative. If you’d like to take part, the day to mark on your calendar is 16 March 2024.

The idea behind the initiative is to boost awareness of how digital waste contributes to CO2 emissions as well, even though we don't regularly treat it as something physical, or even real (it's all in the cloud, innit).

While the website - https://digital.ocistimo.si - tends to liberally mix the costs of RUNNING a website (roughly 0.03g CO2 emitted per page view of their own site) and the initiative is all about deleting your data AT REST (and I’m bothered by such discrepancies), I find the idea behind the initiative to be generally sound. More importantly, it got me thinking about what savings we can reasonably expect to see, especially if everyone did in fact chip in.

While you can scroll to the end of the post to see my final figure, let's crunch some numbers! 🤓

You can also find my calculations in this spreadsheet: https://docs.google.com/spreadsheets/d/1CqD5QuKVwIVnkuZlSKtLY_Hoz5vxfLYYwPU_Q5GFb10/edit?usp=sharing

Now, before you grab your pitchforks, I have some notes and disclaimers:

Notes and disclaimers

This article is intended to entertain and to get you thinking. There are edge cases I haven’t considered, and I simplified a lot to make it short and concise. Think of it as Google’s ping pong balls in a bus question, but with CO2 emissions. At the very best, I used some quick and dirty back-of-napkin math.

I also ignored the fact that any operations we conduct such as deleting the data will undoubtedly also emit CO2.

We also don’t take into account how much of the savings will persist, and when will the data just be filled with some other digi-rubbish, most likely AI generated.

First, let's figure out how much data we can feasibly delete.

How much digital waste data can a nation delete?

Let's say that every single one of the 2.1M residents of Slovenia deleted 50 gigabytes (GB) of data.

Maybe that's a bit generous, but I have deleted movie libraries, games and LLMs way bigger than that, not to mention the collection of terrible photos I have stored on Google. 50GB is also roughly 10% of the entry-level Macbook Pro storage as of early 2024, so we’re going with that as our figure.

Let's also assume that all of that data is deleted from cloud services, and not from your local devices, so that we can calculate running costs for all that data in the cloud.
Local data has a running cost close to 0 because you're probably using your phones and computers regardless, and likely don't use your devices just for data storage. And if you hit the storage limit on your devices, you’d delete that digi-crap regardless of whether it was the Digital Cleanup Day or not.

2.1M residents x 50GB of data gives us a total tally of 105 petabytes (PB) or 105 million GB or 105,000 TB - you get the drift.
Now, we need to figure out how much energy a datacenter consumes per unit of storage.

How much energy does a datacenter consume?

On the site cloudcarbonfootprint.org it is estimated that 1TB of HDD storage in a datacenter uses roughly 6.5Wh. Let's multiply by 3 to cater for the replication factors (I’ve taken the value for AWS S3) which gets us to 19.5 Wh/TBh.

We're interested in yearly tallies, which, multiplied by 24 and 356 brings us to 170.82 kWh per TB-year. That's less than our house’s solar panels produced so far this year, by the end of February in the famously blue-sky-shy London.

Putting the two numbers together, and multiplying our yearly data center consumption by 105,000TB of data we'd hypothetically be deleting, gets us a whopping 17.49 GWh of energy that Slovenians would save in a year, were it not for that pesky digital waste.

For a single person deleting 50GB of data that would mean an annual saving of 8.33 kWh.
Finally, for the fun part, let’s calculate the CO2 emission savings from all that deleted digi-garbage.

Calculating the CO2 emissions savings for deleted data

Staying close to home, next to my hometown of Velenje there's a rather sizable thermal power plant - Šoštanj Power Plant. It’s the one in the header photo in all its glory. It burns lignite, which is like coal but less energy efficient and more polluting, and accounts for between 20-30% of Slovenian electricity production.

What if all of the energy saved came from that power plant, as a worst-case scenario. Coal is, after all, terribly polluting. Luckily, it belongs to a state owned utility company that produces reports. Here’s the annual report from 2021 (in Slovenian).

The TL;DR is that in a year, our power plant produced 3,137 GWh of electricity from burning 2,833 kt of coal, which produced 3,266 kt CO2 emissions in 2021. It also produces municipal heat from some of the coal, so I had to work out the CO2 emissions just related to electricity production.

Our projected savings of 17.49 GWh from deleting all that data would therefore translate to a reduction of 18,214t of CO2 emissions.

That figure shocked me. That seems like a lot of emissions, especially once you imagine how much CO2 by volume 18.2 kilotons must be. it’s 9,940,933.5 m3 or just under 3977 olympic swimming pools, if you like swimming. A lot. That's of course the best (or worst, depending how you look at it) case scenario, where all the electricity came from burning the most CO2-emitting source of energy.

Looking at a single person deleting 50GB, they would reduce the emissions by 8.7kg. To put it into perspective, it's a bit more than a large Thanksgiving Turkey in weight, or  roughly 1.4 Cybertruck cargo bed spaces worth’ of volume.

Conclusions (it’s not that bad, really)

Now, companies operating data centers are usually very proud of their green credentials, and, if they can avoid it, they will probably not buy energy generated from coal, let alone lignite.

Instead, they will purchase their energy (or generate it) from renewable sources, meaning the actual emission reductions are likely much lower, in some cases probably also close to zero.

AWS for example boasts that 90% of their data center consumption comes from renewables, including 100% for a lot of their US data centers (yes us-east-1 is there). https://sustainability.aboutamazon.com/products-services/the-cloud?energyType=true

Google on the other hand has a plan to go fully carbon neutral by 2030 - https://www.google.com/about/datacenters/cleanenergy/

So, here you have it. How much CO2 could a small nation save by deleting a handful of files each?

Somewhere between 18 kilotons and zero, I guess.

Still, if you want to participate in the digi-trash cleanup day and delete some of your e-junk - check out the site: https://www.digitalcleanupday.org to learn more.

And if you want to look at how I worked out the numbers - here’s my spreadsheet: https://docs.google.com/spreadsheets/d/1CqD5QuKVwIVnkuZlSKtLY_Hoz5vxfLYYwPU_Q5GFb10/edit#gid=0