The current forecasts look bleak: our information technology could account for around 50 percent of global electricity consumption over the next 10 to 20 years. At the same time, our data centres require ever-increasing amounts of water. Meanwhile, the climate is becoming more extreme and our soils ever drier. However, the point at which we could simply do without computers, mobile phones and AI chatbots has long since passed.
But the fact is, companies, organisations and the self-employed also need to make and use websites. Likewise, service providers can’t realistically avoid using cloud services and hosting, even if they cared deeply about their environmental impact.
Therefore, it simply makes sense to adopt green practices right from the start when designing a new website or service. Green coding or hosting can form part of a sensible Corporate Digital Responsibility strategy. It can also serve as a future safeguard should you wish to move away from US big tech companies.
How exactly does Green Coding work?
All programmes, services and websites have one thing in common: they consist of code. Whether written by a human or by generative AI is, for now, irrelevant – just as with its origins, modern information technology is based on simple programming code. Enter: green coding.
Max Westing from the ‘Green-Coding’ project explained to us in an interview that it is always a matter of ‘translating code that is readable by humans into code that can be executed by machines’. And depending on how this code is designed, the energy consumption of the systems that execute it rises or falls.
On the other hand, however, energy consumption is also affected by how efficiently people or other systems can work with the tool that has been developed. “If you don’t have to spend ages clicking around in a piece of software to find things, you use it for less time – and that in itself makes it more efficient,” says Max.
Energy-efficient programming reduces the carbon footprint of software and websites. Use the most efficient libraries possible
During our conversation, however, Max also made it clear that new programmes or websites are rarely written from scratch. So-called ‘libraries’ function like standard components that you fit into your car. Modern software, for example, is made up of several libraries that provide specific basic functionalities. It therefore makes sense for programmers to seek out the most efficient libraries possible.
To find them, Max Westing cites the Green Software Foundation as a general point of contact for “green coding”. Developers can find more detailed information, which we will not go into in detail here. To optimise one’s own software for CO2 intensity, the specifications of the Software Carbon Intensity Score are also helpful.
That brings us to the next tip.
Use Carbon Tracker right from the coding stage
Modern software, apps and websites nowadays operate via “distributed systems”. Our devices do not run them entirely on their own; instead, they offload computational tasks to servers or cloud services. Researchers working on the Eco:Digit project are demonstrating just how difficult it has become to analyse the resource consumption of programmes.
This complexity is also the reason why websites and software are often inadequately optimised for energy efficiency or low resource consumption. Daniela Grau from the think tank Climate+Tech explained to us that such optimisations are all too common. This makes optimisation particularly difficult, as a great deal of money and time has already been invested in developing the programme code.
The “Carbonara” tool from Climate+Tech comes into play at a very early stage. It can be used as a plug-in in VSCode-based editors such as VSCode or Cursors. There, it shows – even whilst you are coding – how large the application’s carbon footprint will be later on.
Carbonara also offers a database of “Green Software Patterns”, which is provided by the Green Software Foundation. This means that programmers receive guidance on sustainable programming practices and the most efficient libraries available right from the coding stage.
Grid awareness and CO₂-conscious load shifting
Our next tip may sound complicated, but it’s based on a simple concept: websites designed to be ‘grid-aware’ adapt their content to the current energy mix. If, at any given time, this mix consists largely of fossil fuels, users will see only downscaled, colour-reduced versions instead of large images. So, if the electricity required to display the content is particularly carbon-intensive, the energy-efficient website compensates for this by reducing data traffic.
Carbon-aware load shifting works in a similar way – but in the software sector: if a programme is due to perform a computationally intensive task, it checks the task’s priority and the current electricity mix. If the task can wait – for example, because the results aren’t needed until the next day anyway – the programme will only execute the task when the electricity mix is ‘green’ enough.
In detail, of course, both concepts require certain information. Grid-aware websites retrieve the geographical location of visitors, according to Fershad Irani of the Green Web Foundation. Using a programming interface provided by the Green Web Foundation itself, the grid intensity value for the location is then retrieved and compared with regional averages.
Going into detail here might actually be rather inefficient, as there are other tips as well. You can find more information in the interview with Fershad.
Should we use generative AI – or is it better to avoid it altogether?
In our project Green Digital Futures – How to Develop our Digital World Sustainably, we have observed just how critically the ongoing AI boom is affecting the social and environmental sustainability of our information technology.
Text-based generative AI is based on fairly simple probability calculations. A modern language model breaks sentences down into their constituent parts and then calculates which sequence of words is most likely to follow the query in question. To do this, the systems require huge datasets and a great deal of computing power. Development and use in their current form require massive investment and are based on exploitation.
The current AI boom is leading, for example, to the disregard of copyright laws in order to utilise as much data as possible. At the same time, millions of people are working on training language models, often under appalling working conditions. But the AI boom is also causing problems at a regional level.
According to forecasts, energy prices are set to rise sharply for local residents in regions where new AI data centres are being built. At the same time, local authorities are running out of water, as AI data centres are cooled using fresh water.
How can generative AI be used more economically?
Can you no longer imagine your daily life or your working life without generative AI?
We’ve put together some tips and tricks on how to use services like ChatGPT, Gemini or AI image generators more efficiently.
The AI-first strategies adopted by many companies, which are incorporating supposedly practical ‘AI features’ into an ever-increasing number of services, are exacerbating all these problems. In our view, companies should take this into account when developing new websites and programmes.
However, these drawbacks mainly apply to generative AI. Specialised or traditional AI applications – such as those that existed long before the AI boom – are not necessarily affected by them. Read more about this in our interview with independent climate and energy analyst Ketan Joshi.
Max Westing from the ‘Green-Coding’ project nevertheless sees generative AI as an opportunity to make websites and software more resource-efficient. In a study, he and other researchers were able to demonstrate that AI agents can indeed optimise written code ‘towards energy efficiency’. This holds potential, as many programmers have little time to optimise code themselves.
It remains questionable, however, whether the efficiency gains are sufficiently high to offset the CO2-intensive training and operation of generative AI..
All sorted? Certificates and green hosting
Software, websites and apps are often the first points of contact that B2B and B2C customers have with other companies. It therefore makes perfect sense to ensure that a sustainable digital strategy is visible precisely in these areas.
There is now, for example, a ‘Blue Angel for Software’ certification scheme, which recognises environmentally friendly software. The Green Web Foundation also offers a free verification service for hosting providers. At the same time, the organisation maintains a list of green service providers.
If you wish to showcase your company’s green website alongside other initiatives promoting sustainable business practices, it is also advisable to draw up a comprehensive commitment.
How can we ensure a green digital future?
Growing e-waste, carbon emissions from AI, data centre water usage—is rampant digitalisation compatible with a healthy planet? Our latest project explores how digital tools and services can be developed with sustainability in mind.
