Is cloud storage bad for the environment?

Storing files has a real energy cost, but it is smaller per file than most people imagine and larger in aggregate than most people notice — and the number that matters is duration, not volume. A gigabyte kept for a year costs vastly more energy than a gigabyte transferred once, which is why deleting things is the only intervention with meaningful leverage.

The honest state of the numbers

Before any figures: published estimates for data centre energy use vary enormously, and estimates per gigabyte vary by more than an order of magnitude depending on assumptions about hardware, utilisation, replication and what share of overheads you allocate.

Anyone quoting you a confident figure for the carbon cost of one gigabyte is being more precise than the evidence supports. What is well established:

• Data centres account for a low single-digit percentage of global electricity use.
• That share has grown far more slowly than data volumes, because efficiency improved dramatically over the same period.
• Efficiency gains have historically offset much of the growth, and there is genuine disagreement about whether that continues.

So: real, not apocalyptic, and dominated by factors other than how many photographs you keep.

Why duration is the variable that matters

A file being transferred consumes energy in networking equipment for a few minutes. A file being stored consumes energy continuously — drives powered, plus cooling — for as long as it exists.

The ratio is not close. Keeping a file for a year involves roughly fifty thousand times more elapsed time than the transfer took. Even though storage draws less power per second than a transfer does, duration wins overwhelmingly.

Two consequences follow, and both are counterintuitive:

A service that deletes your file after two weeks is greener than one that keeps it forever on renewable power. Duration multiplies everything, including the renewable version.

Compression barely matters. Making a file 20% smaller saves 20%. Deleting it saves 100%.

Replication multiplies it

Most large cloud storage keeps multiple copies, often across regions, for durability. Three copies is three times the storage footprint.

This is a genuine engineering trade rather than waste — it is what makes "we will not lose your data" true. But it is worth knowing that the durability guarantee you are being sold has a continuous energy cost, and that not everyone needs eleven nines of it for a folder of holiday photographs.

Dark data is the actual problem

The uncomfortable part is not that storage costs energy. It is how much of what is stored is never accessed again.

Industry surveys consistently suggest that a large majority of stored enterprise data is never read after it is written. Estimates of the proportion vary and the methodology is usually loose, but the direction is not in dispute by anyone who has looked at their own storage.

For creative work the effect is stark. A photographer keeps every frame, including the eleven near-identical ones. A video editor keeps every version of every export. None of it is opened again, and all of it is powered continuously for years.

That is the whole argument for deletion, and it is where individual action actually has leverage.

What to do about it

In rough order of effect:

Delete what you will not use. Cull footage and frames before archiving, not after. This is the entire ballgame.

Set a retention rule and execute it. Working files should have a shorter life than deliverables. Choosing a period.

Stop keeping redundant copies you did not ask for. Check whether your backup, your cloud sync and your archive are all holding the same thing.

Prefer providers with honest energy disclosure and short defaults. Smaller effect than the first item, but not zero. How to tell whether a provider's claim is substantive.

Compress when it is free. Worth doing, not worth agonising over.

Keeping perspective

For an individual, storage is a small share of a personal footprint. Flights, heating, driving and the manufacturing footprint of devices all dwarf it. Anyone telling you that deleting emails is a meaningful climate action is selling something, and the framing does harm by making climate action look trivial and performative.

Where it does matter:

At organisational scale, where petabytes of never-read data have a real cost in both money and energy, and where nobody owns the decision to delete.

As a cost discipline, because storage is one of the few places where the environmental answer and the financial answer are identical.

As a risk reduction, since data you deleted cannot leak — which is also what the GDPR requires rather than merely permits.

That three-way alignment is unusual and it is the reason to bother. You are not choosing between doing the right thing and doing the cheap thing.

For file transfer specifically

The choice is genuinely low-stakes. What to look for in a greener service takes five minutes to apply, and then the useful work is on your side of the line: send less, keep it for less time, and stop archiving things you will never open.

Disclosure: I run Yungle, which is one server in Germany on renewable power with one primary copy rather than three regional replicas — an efficiency choice, and equally an availability trade.