Design has a digital carbon footprint that all creatives can address

Designers are increasingly aware of the emissions associated with their offices, flights, tools, materials, logistics, but should digital activities get more attention? Do video calls, file transfers and cloud platforms quietly replicating work across multiple servers add up to something meaningful?

When URGE recently looked for an estimate of digital carbon specific to professional design workflows, we wanted to go beyond the standard government figure for powering an employee's computer and lights while homeworking. So we built something new.

What is the digital carbon of designing?

0.06 kg CO2e per person per hour based on a typical design working hour, including video calls, file uploads, cloud storage, backups, team chat and app usage, all on top of the device energy the government figure already captures.

The model draws on UK DSIT data on data centre energy intensity, Zoom's published bandwidth figures, the HTTP Archive's web page weight data, and Microsoft's figures on typical email and messaging volumes. Using likely energy demands and reasonable emissions factors for the UK grid, we got 0.06.

Standard carbon accounting uses the UK government's office equipment factor of 0.03144 kg CO2e per person per hour, covering the energy your devices consume every hour. In some jobs, maybe that's a reasonable approximation. But in design work, it's probably a significant understatement.

Add the government's device figure, and a design working hour likely sits above 0.09 kg CO2e and roughly three times the standard assumption. That's without AI queries, large asset rendering, or video-heavy workflows. It also excludes the embodied carbon in the hardware and software itself. More on both shortly.

The main culprit is files in transit, not files at rest. The vast majority of the estimated emissions come from data moving, not data sitting on a server. And a design file moves more than most people realise: each save or autosave, each replication to backup locations, each future download by a colleague is a separate transfer. Our model captures the headline movements; continuous sync traffic in cloud-native tools like Figma or Adobe CC would push the figure higher still, making our estimate likely conservative for active and collaborative workflows.

Typical emails without attachments were relatively negligible in carbon terms, even if removing images from email signatures is a worthwhile easy win.

Our estimate does not capture everything

Two significant impacts are outside the scope of the 0.06 figure.

One omission is hardware. Manufacturing a laptop carries substantial embodied carbon, emitted at the point of production, regardless of how efficiently it runs or how quickly the grid decarbonises. Our estimate assumes the devices exist. It doesn't account for the carbon cost of making them.

Another omission is AI. Every query to a generative AI model draws energy in the moment. But AI models also carry substantial embedded carbon from training, something most software doesn't necessarily have. Building a large model consumes energy at a scale with no equivalent in conventional software development. All software has to be created, but not all software requires a data centre running at scale for weeks to come into existence. In that sense, AI is worth thinking about more like a piece of hardware: there is an embodied cost before you ever use it, and a running cost every time you do.

Water is an underreported part of this story and many others. The cooling demands of AI inference and training are attracting increasing attention. Water footprints deserve their own URGE newsletter, and will get one.

How can creatives do something about it?

Digital carbon isn't a single problem with a single solution. The spectrum runs from infrastructure decisions affecting thousands of users to individual habits and devices. Most creatives will recognise themselves at more than one point on it.

At the high-leverage end, those working on server infrastructure, software or hardware used by many people, including those managing an organisation's O365. They make decisions that become defaults inherited by everyone downstream, often for years. How much data is stored, replicated and retrieved at every step will only grow in importance as the volume of data collectively created continues to rise.

Those working with digital assets: brand guidelines, websites, apps, campaigns. They make decisions with real impact, whether framed that way or not. A website carrying large images, looped video, multiple font files or unnecessary code sends more data to every device that visits it, multiplied across every visit. Simple things like screen brightness add up (the idea behind Blackle is that OLED screens use less energy displaying black than white).

Those working with large files: video, architectural BIM models, high-resolution assets, render pipelines. They will have a data footprint per working hour substantially higher than the URGE estimate, which assumed a general design workflow. Format choices, compression, render settings, retention policies and how many copies are made, kept and moved between machines.

And for just about every creative: the hardware. Manufacturing electronic devices carries substantial embodied carbon, emitted at the point of production regardless of how efficiently it runs or how quickly the grid decarbonises. The longer it is used, the more the upfront impact is amortised. Specifying hardware that can be upgraded, repaired, and supported long-term is worth extra thought. Maybe a high-end phone can replace a laptop for some people, and if not now, possibly soon.

Take this thinking further with Design Declares

On Friday 20 March, Design Declares is hosting Designing Digital for a Sustainable Future at the ustwo offices in Shoreditch, 4–8 pm.

Three speakers, three levels of the problem. Kenny Heard of Driftime on the framework they are developing to balance creative ambition with digital sustainability day to day. Sandrine Herbert-Razafinjato of The Responsible Design Studio on what it actually takes to shift the digital practices of 30,000 people during the GenAI transition. Thorsten Jonas on the importance of design choices in an algorithmic society.

As always with Design Declares, expect panels and workshops, not just talks.