Development of AI needs to align with sustainability goals

All Air Pollution Climate Change Energy Environmental Protection Light Pollution Policy Opinion Trail Conservation Plastic lssue Waste lssue Waste Reduction & Recycle Water Resources

Development of AI needs to align with sustainability goals

23 Jul 2025

Category

Keywords

Climate Change Carbon Emission Climate Greenhouse Gases

(23 July 2025, SCMP) Since the launch of ChatGPT in 2022, the global adoption of generative AI has surged, with daily active users reaching an estimated 115 million to 180 million this year.

AI tools are now integral to day-to-day tasks like research and content creation. To support this demand, tech giants like Apple and Google have invested heavily in data centres, with Meta and Microsoft even exploring nuclear power options to meet AI’s massive energy needs.

This raises concerns about the environmental impact of artificial intelligence, particularly its contribution to carbon emissions. In the United States, around 3,000 data centres consumed about 200 terawatt-hours of electricity in 2024 – comparable to Thailand’s annual usage. At least a quarter of that is estimated to have been used by AI-specific servers in those centres.

By 2028, AI-specific electricity consumption is projected to rise to between 165 and 326 terawatt-hours yearly. Ironically, even asking AI how you can reduce your carbon footprint will add to emissions.

While Hong Kong’s Climate Action Plan 2050 acknowledges the issue and the government has proposed expanding the Buildings Energy Efficiency Ordinance to include high-consumption buildings like data centres, progress has been slow.

Data centre energy usage in Hong Kong remains confidential, though Lawrence Iu of Civic Exchange estimates that data centres accounted for 4 per cent to 5 per cent of the commercial sector’s energy use in 2022. This could double by 2028 if trends mirror those in the US.

We urge the Hong Kong government to mandate transparency in data centre energy consumption. Public access to this data would foster accountability and awareness of AI’s environmental footprint.

AI’s impact extends beyond electricity. It includes water for cooling systems, raw materials for graphics card (GPU) production and electronic waste. Each AI prompt contributes to this footprint, underscoring the need for user awareness and responsibility.

Governments should promote sustainable AI usage through education and curb energy and water consumption through regulation.

Mainland China aims to reduce data centres’ power usage effectiveness (PUE) to a ratio of 1.5 by the end of the year, while raising their use of renewable energy by 10 per cent annually. Singapore aims to achieve a PUE ratio below 1.3 by 2034. Hong Kong must follow suit to balance AI growth with climate goals.

While AI offers undeniable convenience, its unchecked expansion threatens net-zero ambitions. Governments and users must act decisively to ensure AI development aligns with sustainability.

Rico Wong, deputy director, The Green Earth

South China Morning Post, Development of AI needs to align with sustainability goals

Back