Climate Solution or Hidden Polluter?
Artificial Intelligence (AI) is generally considered a transformative force with the potential to solve some of the most pressing global challenges, including climate change. AI excels at processing vast amounts of data, improving weather forecasting, optimising resource use in agriculture, and identifying pollution sources with great accuracy. However, the environmental benefits of AI come with significant trade-offs, particularly in terms of energy consumption and carbon emissions. This duality - AI's promise as a climate solution versus its environmental costs - raises complex questions about how sustainable finance should approach the technology.
As AI models grow in complexity, so too does the energy required to train and operate them. The MIT Technology Review (May 2024) explains this energy demand places a significant burden on the electrical grid, which is still largely powered by fossil fuels in many parts of the world. For instance, training large language models or running advanced generative AI applications requires immense computing power, leading to higher energy consumption and increased carbon emissions. By 2027, AI's energy consumption is predicted to rival that of entire nations like Argentina or Sweden.
The issue is further complicated by how companies report their greenhouse gas (GHG) emissions. Many companies, some of them tech giants, use "market-based" emissions accounting, which allows them to report emissions figures adjusted for carbon offsets and renewable energy credits (RECs). This method can obscure the true environmental impact of their energy use, particularly when the electricity used to power AI models comes from non-renewable sources. While these companies may appear to be reducing their carbon footprint on paper, the reality is that their operations may still be heavily reliant on fossil fuels, thereby contributing significantly to global emissions.
For example, as reported by the Financial Times (August 2024) Amazon presents itself as a sustainability leader. However, the company remains one of the largest emitters of greenhouse gases due to its extensive electricity use. In the US, where fossil fuels accounted for about60% of electricity generation in 2023, Amazon's environmental impact can be viewed from two perspectives: as a hero for its investments in clean energy oras a heavy emitter due to its reliance on the existing fossil-fuel-heavy grid.This duality highlights the flaws in current emissions reporting methods, which can allow companies to offset their actual pollution through investments in clean energy initiatives that may not directly correlate with their energy consumption patterns.
Meta, similarly, claims to have achieved "net zero" energy emissions, but a closer examination reveals a stark contrast between reported figures and actual CO₂ emissions. According to a recent Financial Times analysis, Meta's real-world CO₂ emissions from power consumption reached 3.9 million tonnes in2023, compared to the 273 net tonnes reported. This discrepancy underscores the challenges of accurately accounting for emissions in a way that reflects the true environmental impact of energy-intensive AI operations.
As companies race to develop more advanced AI technologies, they are likely to become some of the largest electricity consumers, potentially jeopardising their net zero commitments. Anticipating this, some companies have been actively lobbying for changes to the Greenhouse Gas Protocol, the body overseeing carbon accounting, pushing for frameworks that critics argue could allow them to report emissions figures that do not accurately reflect real-world pollution. For instance, a coalition including Amazon and Meta supports a framework that critics say may obscure their actual emissions by allowing for more liberal use of offsets and RECs.
Conversely, Google has proposed a more stringent approach, requiring companies to offset emissions with power sources that more closely match their actual energy consumption patterns. This proposal, however, has been criticised by other companies as being too costly and complex to implement. The debate over these accounting methods reflects broader concerns about how we measure and manage the environmental impact of AI and other energy-intensive technologies.
For sustainability-minded investors, this presents a significant challenge. Those who track Paris-Aligned Benchmarks or Climate Transition Benchmarks in their investment strategies are committed to ensuring an overall emissions reduction in their portfolios year over year. This has increased investors exposure to information technology companies, who have an easier time reducing emissions year over year, However, as large tech companies like Amazon and Meta increase their emissions due to the energy demands of AI, investors may find themselves running out of suitable targets that fulfill the requirements for these strategies. This raises a fundamental question: how should investors weigh the short-term environmental costs of AI development against its potential long-term benefits, such as improved climate modelling and increased agricultural efficiency?
As Emil Stigsgaard Fuglsang, COO and co-founder ofMatter, highlights in PA Future (August 2024), the rapid expansion of AI is forcing a re-evaluation of the trade-offs between sustainability and technological advancement. While AI has the potential to be a climate champion, its current trajectory as an energy-intensive technology presents a significant challenge to global sustainability efforts. Until more robust frameworks for measuring and managing emissions are developed, the full impact of AI on the climate will remain difficult to assess, leaving investors and policymakers with the difficult task of balancing the immediate environmental costs with the promise of future benefits.
In conclusion, AI stands at a crossroads: it can either emerge as a key tool in the fight against climate change or become a major contributor to the very problem it seeks to solve. The outcome will depend on how effectively we can integrate AI into a cleaner energy framework and improve the transparency and accuracy of emissions reporting. Only then can we determine whether AI is truly a hero or a villain from a sustainability perspective.
Publication Details
Author: Sarah Millerton
Date: August 27, 2024
Artificial Intelligence (AI) is generally considered a transformative force with the potential to solve some of the most pressing global challenges, including climate change. AI excels at processing vast amounts of data, improving weather forecasting, optimising resource use in agriculture, and identifying pollution sources with great accuracy. However, the environmental benefits of AI come with significant trade-offs, particularly in terms of energy consumption and carbon emissions. This duality - AI's promise as a climate solution versus its environmental costs - raises complex questions about how sustainable finance should approach the technology.
As AI models grow in complexity, so too does the energy required to train and operate them. The MIT Technology Review (May 2024) explains this energy demand places a significant burden on the electrical grid, which is still largely powered by fossil fuels in many parts of the world. For instance, training large language models or running advanced generative AI applications requires immense computing power, leading to higher energy consumption and increased carbon emissions. By 2027, AI's energy consumption is predicted to rival that of entire nations like Argentina or Sweden.
The issue is further complicated by how companies report their greenhouse gas (GHG) emissions. Many companies, some of them tech giants, use "market-based" emissions accounting, which allows them to report emissions figures adjusted for carbon offsets and renewable energy credits (RECs). This method can obscure the true environmental impact of their energy use, particularly when the electricity used to power AI models comes from non-renewable sources. While these companies may appear to be reducing their carbon footprint on paper, the reality is that their operations may still be heavily reliant on fossil fuels, thereby contributing significantly to global emissions.
For example, as reported by the Financial Times (August 2024) Amazon presents itself as a sustainability leader. However, the company remains one of the largest emitters of greenhouse gases due to its extensive electricity use. In the US, where fossil fuels accounted for about60% of electricity generation in 2023, Amazon's environmental impact can be viewed from two perspectives: as a hero for its investments in clean energy oras a heavy emitter due to its reliance on the existing fossil-fuel-heavy grid.This duality highlights the flaws in current emissions reporting methods, which can allow companies to offset their actual pollution through investments in clean energy initiatives that may not directly correlate with their energy consumption patterns.
Meta, similarly, claims to have achieved "net zero" energy emissions, but a closer examination reveals a stark contrast between reported figures and actual CO₂ emissions. According to a recent Financial Times analysis, Meta's real-world CO₂ emissions from power consumption reached 3.9 million tonnes in2023, compared to the 273 net tonnes reported. This discrepancy underscores the challenges of accurately accounting for emissions in a way that reflects the true environmental impact of energy-intensive AI operations.
As companies race to develop more advanced AI technologies, they are likely to become some of the largest electricity consumers, potentially jeopardising their net zero commitments. Anticipating this, some companies have been actively lobbying for changes to the Greenhouse Gas Protocol, the body overseeing carbon accounting, pushing for frameworks that critics argue could allow them to report emissions figures that do not accurately reflect real-world pollution. For instance, a coalition including Amazon and Meta supports a framework that critics say may obscure their actual emissions by allowing for more liberal use of offsets and RECs.
Conversely, Google has proposed a more stringent approach, requiring companies to offset emissions with power sources that more closely match their actual energy consumption patterns. This proposal, however, has been criticised by other companies as being too costly and complex to implement. The debate over these accounting methods reflects broader concerns about how we measure and manage the environmental impact of AI and other energy-intensive technologies.
For sustainability-minded investors, this presents a significant challenge. Those who track Paris-Aligned Benchmarks or Climate Transition Benchmarks in their investment strategies are committed to ensuring an overall emissions reduction in their portfolios year over year. This has increased investors exposure to information technology companies, who have an easier time reducing emissions year over year, However, as large tech companies like Amazon and Meta increase their emissions due to the energy demands of AI, investors may find themselves running out of suitable targets that fulfill the requirements for these strategies. This raises a fundamental question: how should investors weigh the short-term environmental costs of AI development against its potential long-term benefits, such as improved climate modelling and increased agricultural efficiency?
As Emil Stigsgaard Fuglsang, COO and co-founder ofMatter, highlights in PA Future (August 2024), the rapid expansion of AI is forcing a re-evaluation of the trade-offs between sustainability and technological advancement. While AI has the potential to be a climate champion, its current trajectory as an energy-intensive technology presents a significant challenge to global sustainability efforts. Until more robust frameworks for measuring and managing emissions are developed, the full impact of AI on the climate will remain difficult to assess, leaving investors and policymakers with the difficult task of balancing the immediate environmental costs with the promise of future benefits.
In conclusion, AI stands at a crossroads: it can either emerge as a key tool in the fight against climate change or become a major contributor to the very problem it seeks to solve. The outcome will depend on how effectively we can integrate AI into a cleaner energy framework and improve the transparency and accuracy of emissions reporting. Only then can we determine whether AI is truly a hero or a villain from a sustainability perspective.
Publication Details
Author: Sarah Millerton
Date: August 27, 2024
