Sustainability in the Era of AI. Part 1 - The Environment
- Miriam Mukasa - Inclusive Leadership & AI
- Jul 8
- 4 min read
Updated: Jul 10
Organisations that are serious about ESG need to tackle the challenges and damage caused by AI, not just downstream i.e. during or after deployment, but throughout the AI lifecycle, including supply chains. Failure to do so, could lead to backlash including accusations of green washing.
Data centres, which are basically warehouses housing servers and other tech infrastructures, are currently attracting a lot of attention - often not good. These warehouses are not new and have existed pre-GenAI hype. The difference now, however, is that traditional data centres don't impact people nor do they require as much land, water, energy, or infrastructure, as AI data centres do.
Given the vast amounts of resources consumed by AI data centres or “AI factories”, a distinction Jensen Huang, CEO of Nvidia, likes to make, it comes as no surprise then, about the growing concerns over the impact these warehouses may have, or are already having, on both people and planet. In Part 1 today, we will focus on the environment.
Data Centre Impact on the Environment (The list below is not exhaustive)
1. Water Scarcity - AI data centres require a lot of water to keep servers cool. This water needs to be good quality to mitigate rust and bacteria. According to a report by Cornell Chronicle (David Nutt, 10 Nov 2025), the current rate of AI growth would annually drain 731 to 1,125 million cubic meters of water per year, this is equal to the annual household water usage of 6 to 10 million Americans. Naturally, this is cause for concern in many communities especially those in drought-prone regions.
In terms of sustainability, the new generation of closed-loop water cooling is said to be significantly better, as the same water is used over and over again, reducing water usage by 90%, compared to older generation cooling towers. However, these systems are more costly than open systems which means organisations other than hyperscalers, may struggle to afford them.
The question now is, how will local communities react when water rationing is applied to them (including local farmers), but not the data centres located only a few hundred metres from their front doors or farms?
2. High Carbon Emissions - Fossil fuels still provide about 60% of total global electricity generation. AI’s growth and exploding energy use could see increases in green-house emissions due to prolonged use of fossil fuels as well as the deployment of off-grid diesel generators. According to the aforementioned Cornell Chronicle, the current rate of AI growth would annually put 24 to 44 million metric tons of carbon dioxide into the atmosphere, which is the emissions equivalent of adding 5 to 10 million cars to U.S. roadways.
3. Heat Islands – This is where some local areas in cities that house data centres experience higher temperatures than their surrounding rural areas. An example is shared in The Guardian newspaper (26 June 2026), which reports that ‘Emerging research suggests datacentres create a heat island effect, pushing up temperatures in the immediate vicinity by as much as 9C’.
4. Minerals - Whether used for power generation, semiconductors or data storage components, AI requires significant amounts of minerals and metals. This often impacts local communities through deforestation, water usage, soil contamination and more.
5. Marine ecosystem – Some have talked about sea cooling as an alternative to land based (water) cooling systems. While these subsea data centres are said to reduce power consumption compared with land-based data centres, according to a report by the London School of Economics (Global School of Sustainability, 17 September 2025), the heat generated by these undersea data centres (which rely on the surrounding seawater to naturally cool the computer servers), can disturb sediments by raising local water temperature which in turn, reduces oxygen availability, thus threatening marine species’ healthy functioning.
6. Electronic waste (e-waste) - GPUs rapidly become outdated and need replacing far more regularly than CPUs. CPUs (found in traditional data centres and essential for software engineering), are used for general computing and designed for sequential processing while GPUs, (used for AI, gaming or video editing), are built to process massive amounts of data simultaneously. GPUs have a life span of ~ 5-8 years (some say as little as 2-3 years, depending on use), compared to CPUs which have a significantly longer life span of 10+ years. In addition, data centres use a lot of electrical components. What will happen to the huge volumes of transformers, switchgears, batteries etc, when they need replacing and then disposing? Will these be simply shipped to the global south along with other discarded products from the global north?
Organisations that are serious about ESG need to tackle the challenges and damage caused by AI, not just downstream i.e. during or after deployment (where the focus seems to be currently), but throughout the AI lifecycle, including supply chains.
Otherwise, those organisations which have been recognised for, and/or take pride in their green credentials could find it difficult to justify why ESG standards apply only to non-AI related supply chains. This could lead to backlash alongside accusations of green washing and/or, double standards.
******************************
Subscribe below to receive an alert as soon as I publish new editions of my LinkedIn (monthly) newsletters or, to read previous editions:
Responsible AI (this one) - Putting People at the heart of AI Strategy
Inclusive Leadership in the era of AI - Creating a culture of inclusion in the era of AI
Leading with Emotional Intelligence (EQ) - Using EQ to build a culture of empathy and collaboration to drive organisational success
ABOUT ME - Executive Advisor/Consultant ExecutiveGlobalCoaching.com - Advising senior executives on leading strategically, responsibly and inclusively in an AI‑enabled future.
Images from Unsplash: Image 1 Sea turtle (Marcus Lange), Image 2: Tech rack (Massimo Botturi) Image 3: Data centre water cooling fans (Winston Chen) Image 4: Data centre (Geoffrey Moffett) Image 5: Electronic waste (John Cameron)
Comments