Mei Ri Jing Ji Xin Wen·2025-08-04 08:13Core Viewpoint - The rapid development of AI technology is leading to significant environmental costs, including increased carbon emissions and resource consumption, which companies like Microsoft are attempting to offset through carbon removal projects [1][4][10]. Group 1: Environmental Impact of AI - AI's infrastructure expansion is associated with substantial electricity, water consumption, and rising carbon emissions, with Microsoft reporting a 23.4% increase in carbon emissions since 2020 due to its cloud computing and AI business growth [1][4]. - The training and operation of AI models, such as Mistral Large 2, have been shown to produce significant carbon footprints, equivalent to the emissions of 5,000 cars over a year, and consume vast amounts of water [7][9]. - Major tech companies, including Amazon and Microsoft, have seen their operational carbon emissions increase dramatically, with Microsoft experiencing a 155% rise from three years ago [6][7]. Group 2: Carbon Removal Initiatives - Microsoft has signed a significant deal with Vaulted Deep to achieve a carbon removal target of 4.9 million tons by 2038, with the deal potentially costing up to $1.7 billion [3][4]. - The carbon removal method employed by Vaulted Deep involves injecting organic waste into deep underground layers, effectively sealing it to prevent environmental contamination [3][4]. Group 3: Criticism of Carbon Offset Strategies - Experts criticize the practice of purchasing carbon credits as a means of offsetting emissions, arguing that it does not equate to actual emission reductions and can mislead the public regarding a company's environmental impact [4][10]. - The current carbon accounting mechanisms used by tech companies are seen as flawed, allowing them to claim zero emissions through market-based accounting without addressing the root causes of emissions [6][10]. Group 4: Future Projections and Recommendations - The AI sector is projected to consume vast amounts of resources, with estimates suggesting it could use up to 66 billion cubic meters of water annually by 2027, and data center energy consumption could double by 2030 [9][10]. - Experts advocate for a shift towards renewable energy and improved transparency in environmental impact reporting, emphasizing the need for government regulation to enforce accountability in carbon neutrality commitments [11].