Understanding the AI Carbon Footprint
Businesses across all sectors are becoming more conscious of their environmental impact. From reviewing supply chains to switching to renewable energy, organisations are taking active steps to measure and reduce their carbon footprints. This shift is being driven not only by regulations and investor pressure but also by growing consumer demand for genuine sustainability.
Yet, while many companies are tracking emissions from things like transport, heating, and packaging, there’s one increasingly significant contributor that’s often overlooked: artificial intelligence.
The Silent Impact of AI on Sustainability Goals
As AI tools become more embedded in everyday business operations, from customer service chatbots to automated marketing and analytics, it’s easy to view them as purely digital and therefore harmless. But the truth is, the algorithms running behind the scenes require massive amounts of data and computing power, and that comes with a carbon cost.
Most businesses don’t immediately associate their use of ChatGPT, automated assistants, or AI-powered recommendations with environmental impact, but it all adds up.
As we increasingly rely on these tools to drive productivity, efficiency, and innovation, it’s important to recognise that AI isn’t energy-neutral.
Why AI Comes with a Big Carbon Footprint
Artificial intelligence is often perceived as a clean, digital solution. But behind every smart assistant, image generator, or large language model lies a significant environmental footprint.
The energy demands of AI are immense, and they begin with the data centres that power it.
Data centres are vast, warehouse-sized facilities filled with thousands of servers that run nonstop. These servers perform the complex calculations needed to train and run AI models, and they generate a lot of heat while doing so.
To stay operational, data centers require constant cooling, which in many cases consumes as much energy as the computing itself. Depending on where these centres are located, that energy is often drawn from grids that still rely heavily on coal, gas, or oil.
One of the most striking examples of AI’s carbon cost comes from a 2019 study which found that training a single large language model can produce over 284 tonnes of carbon dioxide. To put that into perspective, that is equivalent to five times the emissions of an average petrol car across its entire lifespan.
While efficiency improvements have reduced the energy needed to train newer models, the widespread adoption of AI has led to a dramatic increase in overall usage.
Every time a business uses an AI-powered chatbot, recommendation engine, or analytics tool, it taps into this global network of high-energy infrastructure.
Furthermore, the environmental impact of data centres does not stop at emissions. Many centres are built in areas with limited water resources and rely on vast quantities of water for cooling, especially those using evaporative cooling systems.
In some regions, this puts additional pressure on already stressed local water supplies.
Data centres also often require significant land and resource use during construction and operation. The materials needed for servers, including rare earth metals, contribute to global extraction and supply chain pressures.
There are also ethical considerations around the placement and operation of data centres. They are frequently located in regions where electricity is cheapest, which can mean areas with weak environmental regulations or high dependence on fossil fuels. Local communities sometimes face increased energy prices, environmental degradation, or water shortages while receiving few of the economic benefits in return.
This hidden footprint is what makes AI so complex from a sustainability perspective. While it may be helping businesses work more efficiently on the surface, it also comes with a growing energy and ethical cost that must be acknowledged as part of any serious environmental strategy.
Greener AI: Innovation and Industry Shifts
The good news is that as awareness of AI’s environmental footprint increases, there is a growing effort within the tech industry to reduce its impact. While the development and use of AI still requires significant energy, many companies are beginning to explore ways to make AI models more sustainable.
Some are investing in more energy-efficient hardware and improving the way algorithms are trained to reduce both time and electricity consumption. Others are working to decarbonise their data centres by powering them with renewable energy, although the availability and consistency of clean energy vary by region.
There is also a movement toward greater transparency. A few providers have started to report on the emissions linked to their models, and some now offer users the ability to select lower-power or so-called “eco” modes.
In addition, newer models are increasingly being designed to operate with smaller datasets and less computing power, helping to reduce the total energy required.
However, these shifts are still in the early stages. Most AI models remain resource-intensive, and the industry lacks standardised reporting on energy use and emissions. This means that businesses still have a critical role to play in asking questions, making informed choices, and using AI tools in ways that align with their sustainability goals.
How Businesses Can Account for and Reduce their AI Carbon Footprint
If your organisation uses AI, it’s time to start factoring its impact into your sustainability strategy.
1. Audit Your Use of AI
Identify where AI is being used in your business, from automated customer service and sales platforms to research tools and internal analytics. Many tools quietly integrate machine learning, so map them clearly.
2. Ask Your Providers About Sustainability
Reach out to vendors and ask whether their AI tools run on renewable energy, what data centres they use, and whether they offer eco-efficient models.
3. Opt for Lighter AI Models Where Possible
Some tasks don’t require large, energy-intensive models. Lightweight, domain-specific AI can perform efficiently with a much lower carbon cost.
4. Acknowledge and Balance AI-Related Emissions
If your business is using AI regularly, it’s important to include this in your overall carbon accounting. While you may not be able to avoid the emissions entirely, you can take steps to balance them by supporting meaningful climate action. At Leaving a Legacy, we help businesses contribute to long-term ecosystem restoration projects that protect biodiversity, support communities, and build climate resilience. These types of initiatives offer a tangible way to respond to the digital footprint your business is creating, without relying on quick fixes or transactional offsets.
5.Support Green AI Innovation
Back companies and partners that are investing in responsible AI development. The more demand there is for sustainable solutions, the more the market will shift.
AI Is A Sustainability Blind Spot
AI is transforming how we work, offering new opportunities for innovation, efficiency and growth. However, this progress comes at a cost that is often overlooked, especially when we assume digital tools have little or no environmental impact.
As AI becomes more integrated into everyday business operations, it is important to recognise its role in your organisation’s wider sustainability commitments.
Understanding the link between AI use and emissions allows businesses to make informed choices, support responsible innovation and take meaningful steps to ensure their digital transformation aligns with their climate commitments