The environmental implications of technology are vast, complex and embedded across the supply chain. From the extraction of minerals, metals, rare earth elements to build technology, through to the energy consumption when technology is in use, through to the negative impacts of this technology on the environment during its usage and later in end of life – e-waste is now the biggest growing waste sector, increasing 60% per year according to the WTO.

Further,  specifically looking at the widespread roll out of AI models, there is a danger that we will get locked into unchangeable systems that use world-view biased AI models to automate tasks - such as pest control in agriculture, or streamlined fossil fuel extraction - that will make changing highly polluting sectors into more sustainable sectors very challenging. Once automated and embedded there is a fossil fuel lock in, more complex than the systems we are facing today, as the fundamental logic behind such systems would no longer be in the hands of policy change and advocacy, outsourced instead to machines.

The pursuit of technology, and its negative impact on climate and environment are highly integrated and cannot be ignored, in particular where the intersection undermines the protection of humanity and the continued ability to sustain life on this planet. This section in the background document, from a life cycle analysis perspective, breaks down the intersections on the impact of technology on climate, sustainability and environment.

To function, technology requires minerals, metals, and rare earth elements. Whilst a small sum is used from recycled goods, the vast majority comes from extraction (mining). The impacts of natural resource extraction are far, wide and complex. A list of the most common elements to power today’s technology: lithium, cobalt, copper, nickel, manganese, REEs (rare earth elements): neodymium, praseodymium, dysprosium, and terbium. These are used to power everything from our smart phones and data centres to wind turbines, solar panels and drones. These elements hold the juxtaposition, depending on their usage and control. On one side, they are able to aid in the just energy transition and increase connectivity with vast benefits such as poverty alleviation through accessing of new markets, or supporting with early warning messaging; yet on the other side these minerals can and are being used to increase fossil fuel extraction, expand monoculture agriculture and military/ surveillance systems – all of which undermine International (humanitarian) law in a myriad of ways. This list is not exhaustive.

The impacts of extraction on the environmental are vast and well documented, with mining increasing exponentially. The impacts include degradation such as vast deforestation (tree forest cover loss), water pollution through mining, soil contamination in extraction, air pollution (particulate and radioactive), vast fossil fuel usage in extraction process, ecosystem destruction when top soil clearing for any mining, soil acidification, heavy metal pollution which has a negative impact on human and surrounding environmental health; through to challenges regarding land rights, which have led to exacerbating conflicts. In many cases this has led to the violation of indigenous land rights and forcing migration of groups to extract natural resources for technology – as has been seen in the Sampí population in Northern Europe (sic Sweden), as well as the Masai in Eastern Africa, and the Kayapó, Munduruku, and Yanomami groups in Brazilian Amazonia. 

Annually, mining is responsible for 5-7% of global greenhouse gases, with demand increasing – already since the 90s minings has increased 52%, and the current technological age is pushing this further. According to a nature journal article from 2025, from 1970 to 2022, the CO2 emitted from the extraction of 47 main metal ores has quadrupled, from 2.7 gigatonnes (Gt) in 1970 to almost 9.4 Gt in 2022, with the greatest increases in Oceania (+1,222%), South America (+929%) and Asia (+285%). Ore-specific mining activities are generally concentrated, with the top-five producers contributing on average 82.7% of the global supply in 2022. According to UNEPs resource material flow database, it is clear to see there has been a significant jump in metal ore extraction from post covid until now, compared to previous period (1970 – 2020), across Africa in 1970 was 559,987,901, and in 2024 this stood at 1,149,382,779  in 2024.

After the input materials are extracted, there are several stages in manufacturing and processing that must be understood in the context of environmental and climate impacts of technology from a life cycle perspective. This stage is highly carbon intensive, processes such as smelting, refining, purification, chip fabrication, battery production, hardware assembly are included here, and all elements of this process require a huge amount of water, and energy.

The usage of technology is the most complex to understand and calculate from an environmental and climate perspective. The areas this section will focus on are: CO2 emissions from usage, the environmental and sustainability negative impact areas of technology use (focusing on land use change and energy mixes), and how, where, and at what scale the technology is used

Whilst technology’s energy use in general is an area of concern, this section will focus more contemporarily on the most recent jump in energy usage driven by widescale generative AI adoption (or push) which impacts both immediate demand and long term energy usag. This covers the energy usage from the training of models, through to the inference calls to these LLMs (large language models) and the hosting of cloud services for AI – meaning the building and deployment of data centres. Whilst it is not possible to separate data centre energy usage from genAI and BAU (business as usual) cloud usage, it is possible to infer a change when looking at the “pre-genAI” to post energy usage demand numbers. It is also possible to extrapolate the future expected energy usage, when looking at global investment into the building of datacentres and hyperscalers (unemitted, but planned emissons).

From early studies, it has been found that a generativeAI search consumes 4 to 5 times more energy than a previous machine learning powered engine search. Additionally, huge amounts of water are required to cool processors used in the training and delivery of genAI answers, in 2022 a law suit was brought against OpenAI by the local residents of West Des Moines, Iowa, (the home of OpenAIs giatn data centre cluster) in which it was found that this cluster had used about 6% of the district’s water. Similarly, Google and Microsoft’s models (bard and bing) had increases of  20% and 34% of water usage in one year.

One study suggests that, globally, the demand for water for AI could be half that of the United Kingdom by 2027. One study looking at the BLOOM model (176million parameters) used over 50.5 tonnes of CO2, which is the yearly usage of about 6 households, purely for training. For context, GPT 4 mini is estimated to have anything between 400billion and 1.5 trillion parameters, so training alone would be more than 50 000 households annual energy usage. And this is even before people begin to use it.

Whilst there are gains to be made with resilience, climate adaptation, prediction etc, there are also significant rebound effects associated with genAI that are currently indicating that whilst efficiency gains are made, it is currently leading to an overall increase in consumption. The secondary impacts of efficiency gains in AI also have market, economic and socio political impacts, for example where investments are redirected (much moving away from climate adatpion and mitigation for example in favour or AI infrastructure investment, or worryingly into AI systems that allow for better prediction and data analsyis (climate tech) but little invested in the actual infrastructure needed to respond to such a analyses e.g. build flood barriers if there is data suggesting a large flood is coming.

According to the Global Centre on Adaptation, between now and 2035, developing countries will need USD 3.3 trillion. However, at current levels of financing, only USD 840 billion will flow. According to Morgan Stanley, up until 2028 alone ~$2.9 trillion is committed to global data center construction projects. This doesn’t even touch upon investment into mining, chips, training, quantum computing, models, gadgets and beyond.

E-waste is the fastest growing sector of solid waste in the world. In 2022, an estimated 62 million tonnes of e-waste were produced globally. According to WHO only 22.3% was documented as formally collected and recycled. Lead and mercury contamination are closely associated to e-waste recycling, as they are released into the environment during this process – it is dumped, openly burned, and deposited in areas causing massive environmental and human harm.

E-waste, and the waste sector in general power a large part of the informal recycling sector in developing countries. ILO and WHO estimate that millions of women and child labourers working in the informal recycling sector globally may be at risk of hazardous e-waste exposures. More than 18 million children and adolescents, some as young as five, are engaged in the informal industrial sector, of which waste processing is a sub-sector. As is often the case, the intersection of climate justice, gender, environmental harm and economic inequality is seen starkly as e-waste grows.

In addition, workers, aiming to recover such materials as copper and gold, risk exposure to over 1,000 harmful substances, including lead, mercury, nickel, brominated flame retardants, and polycyclic aromatic hydrocarbons.

Only 17.4% of e-waste is recycled responsibly, while the rest is illegally dumped, overwhelmingly in low- or middle-income countries.

The demand for new technology is powering new economic growth in some parts of the world, whilst having astronomically negative impacts on the prospects of millions of at risk and vulnerable groups, whilst increasing oceanic pollution, air pollution, climate change and health related serious conditions as a result of heavy metal contamination. 

Looking at the entire lifecycle of technology raises serious questions about responsibility when it comes to sourcing, usage and disposal. Before adopting and using technology, there are serious questions to first be asked: is the product designed for repair, reuse, and longevity? Is the technology designed for circularity or planned obsolescence? What materials are effectively lost forever? Where are the materials sourced? Who deals with the waste — and under what conditions? Does the usage and use case justify the cost to environment and people?