The Environmental Impact of AI

Artificial Intelligence (AI) is no longer a concept of the future—it’s a technology that’s actively shaping today’s economy. AI is being widely adopted across the financial sector to boost efficiency, manage risks, and support better decision-making. However, with the growing use of AI, its environmental impact is also increasing—a factor often overlooked in Environmental, Social, and Governance (ESG) discussions.

As banks and financial institutions deepen their commitment to ESG principles, understanding the environmental footprint of AI has become essential. While AI delivers significant advantages, it also comes with notable environmental costs. In an era of growing climate consciousness, factoring in the environmental impacts of digital infrastructure has become more important than ever for sustainable investment strategies.

The technology used to train Artificial Intelligence (AI) requires a significant amount of electricity. Training OpenAI’s GPT-3 consumed roughly 1.287 Gigawatt-hours (GWh) of energy—as much as the annual power use of 120 regular U.S. homes. According to reports, AlphaGo was trained by DeepMind using 1,920 Central Processing Units (CPUs) and 280 Graphics Processing Units (GPUs).

It’s not just how much energy is used, but also where that energy comes from, that counts. There are significant differences in the amount of carbon released from electricity generation across geographical areas. If a model is built in Norway, it will use much less carbon than it would in places that rely heavily on fuels from coal.

High-performance data centers powering AI workloads generally require a considerable amount of water for cooling. As reported by Bloomberg (2023), Microsoft consumed over 1.7 billion gallons of water across its data centers 2022, largely due to heavy generative AI workloads.

Furthermore, Google’s annual environmental review showed that its worldwide water usage rose to 5.6 billion gallons in 2022. This scale of water use has important ethical and environmental implications, particularly due to water scarcity in certain regions across the globe, which is growing worse with rising global temperatures and extended droughts.

Read more: Investing in Water: A Critical Opportunity for Sustainability

What happens to hardware over its lifetime is another important environmental consideration. Using GPUs and Tensor Processing Units (TPUs) helps AI systems run more efficiently; however, this requires hardware to be replaced more often. Rare-earth metals such as cobalt, lithium, and tantalum are used to produce these chips, leading to both environmental harm and health hazards for the workers.

Frequent hardware replacement generates a huge amount of electronic waste. According to the United Nations (UN), global e-waste is projected to increase from 53.6 Million Metric Tonnes (MMT) in 2019 to 74 MMT by 2030. Growth in AI use thus translates to an increasing need for faster, temporary computer hardware.

Companies that score well on ESG metrics, mainly those in the tech sector, tend to receive large financial investments from firms operating ESG-labeled funds. Sustainable portfolios regularly include Microsoft, Alphabet, Meta, and NVIDIA.When the environmental impacts of AI remain overlooked or unquantified, funds may inadvertently contribute to greenwashing by underestimating indirect emissions linked to the activities financed by them.

Climate disclosure rules are expanding to cover various parts of the world. Both the European Union (EU)’s Corporate Sustainability Reporting Directive (CSRD) and the International Sustainability Standards Board (ISSB) are demanding that corporations report on their Scope 3 emissions, for instance, the energy consumption of their digital products.

Even though the environmental impact of AI is usually labeled as Scope 3 for most financial businesses, recording it is now required for compliance with revised standards. Financial institutions need to monitor not only their own CO2 emissions, but also those hidden in their digital environments.

Read more: How AI Is Shaping the Future of Climate Data Collection and Analysis

The growing emphasis on sustainability among Millennials and Gen Z is reshaping investment priorities. These groups are increasingly scrutinizing investment practices, seeking to ensure that their capital contributes meaningfully to positive environmental and social outcomes.

For institutional investors, this shift underscores the importance of careful management and accountability. Overlooking the environmental implications of AI could expose firms to reputational risks and erode trust, particularly among stakeholders demanding greater transparency and sustainability in financial decision-making.

Low-impact AI is being developed by a few startups and academic researchers. Hugging Face, an open-source service, has put out a guide on how to train models efficiently. Organizations such as Anthropic and Cohere are pursuing models that perform well and are less complex.

Nvidia has introduced power-saving chips, and some organizations are exploring training AI entirely on renewable resources. While this experimentation is still at an early stage, these innovations highlight a move toward greener AI growth—although their scalability and feasibility remain largely uncertain.

The largest IT firms aim to make their data centers 100% green. Google hopes to source all its energy from renewable sources around the clock in every location by 2030. Amazon, Microsoft, and Meta are entering Power Purchase Agreements (PPAs) to get clean energy for running their buildings.

RECs (Renewable Energy Certificates) and carbon offsets are included in the mix; however, the evidentiary value is not considered commonly across. A better digital infrastructure is both essential and achievable, with a proper mix of corporate planning, vision and implementation.

Financial institutions are increasingly collaborating with technology providers to advance sustainable approaches to AI. Such collaborations focus on advocating greater transparency around emissions from data centers, supporting lifecycle assessment of AI hardware, and promoting net-zero strategies which collectively impact the digital infrastructure.

To keep pace with the environmental implications of rapid digital growth, ESG frameworks must evolve to consider the sustainability impacts of technology. Key areas of accountability could include tracking computer hours, measuring emissions per AI inference, and assessing the energy sourcing practices of data centers.

Read more: Future of ESG Investing: Key Trends Shaping 2025 and Beyond

AI is supporting innovation in the financial sector, enabling better operations and growth. Risk assessment and portfolio optimization are just a few of the ways in which it is changing how institutions work. However, with the expanding application of AI, its impact on the environment is also increasing, making it important to focus on sustainability when developing AI.

For financial institutions focused on ESG, addressing the environmental impact of AI is becoming increasingly important. Enhancing transparency around AI usage, adopting more precise sustainability metrics, and supporting cleaner technologies are steps that the financial sector can take to contribute to a more sustainable digital future. These actions can help ensure that capital is directed toward solutions that align with long-term sustainability goals.

• Heating up: how much energy does AI use?
• AlphaGo: its creator on the computer that learns by thinking
• 6 billion gallons and counting: Google’s dipping its toes in too much water
• Global E-Waste Surging: Up 21% in 5 Years
• Innovating sustainable ideas. Growing renewable solutions.
• I. tools fueled a 34% spike in Microsoft’s water consumption, and one city with its data centers is concerned about the effect on residential supply