Hong Kong's data centers are among the world's most carbon-intensive, according to a recent UN study. This is largely due to the city's heavy reliance on fossil fuels for energy, with 67% of its energy derived from fossil fuels. The report, titled "Environmental Cost of AI's Energy Use," highlights the global carbon, land, and water impacts of AI infrastructure. By 2030, data centers could consume 945 terawatt-hours, which is nearly triple the combined annual electricity use of Pakistan, Bangladesh, and Nigeria. Hong Kong's carbon footprint is 43% higher than the global average, making it one of the most carbon-intensive grids in the world. This is despite the fact that Hong Kong's energy mix does not rely on hydropower or renewable energy sources, which require large amounts of water and land. As a trade and logistics hub, Hong Kong remains a prime location for data centers, with over 300 internet service providers and 12 external submarine optical fibre cable systems. The government is building a new 110,000 square meter data facility in Sandy Ridge, 90% of which will be dedicated to data centers. However, the UN report also notes that daily AI use, not training, accounts for around 80-90% of total energy demand. This is particularly concerning, as image generation requires a thousand times more energy than a simple text query. The report also cites the case of ChatGPT, which processes around 2.5 billion prompts per day. Data center operators in Hong Kong are striving to enhance energy efficiency, but the city's heavy reliance on fossil fuels and the carbon-intensive nature of AI infrastructure remain significant challenges. The UN study highlights the need for a more sustainable approach to energy use and the development of renewable energy sources in Hong Kong.
