康奈尔大学研究人员 Fengqi You 和 Cheng Feng 于11月9日在《自然·能源》杂志发表的一项新研究发现,随着人工智能数据中心的快速扩张,其产生的巨额电网升级成本主要由居民承担,尤其是低收入和农村地区家庭 1。研究分析了2020年至2025年美国四大主要电网的数据,显示输电投资从2020年的20亿美元增至2025年的340亿美元 1。
数据显示,2025年与数据中心增长相关的基础设施总账单为82.5亿美元,其中数据中心平均仅支付了约38%(31.2亿美元),剩余62%(51.3亿美元)的费用转嫁给了用户 1。这一成本分摊机制导致部分最低收入家庭的额外能源支出最高可达其收入的28% 1。此外,弗吉尼亚州批发电价上涨15%,加州旧金山湾区南部上涨20%;农村地区承受高能源负担(能源支出超过收入6%)的家庭比例比城市高出76% 1。预计至2030年,数据中心将消耗美国电力的8%至17%,其中70%至80%归因于AI技术 1。尽管每吉瓦(GW)数据中心容量可创造15,000个工作岗位,但直接关联岗位仅约2,700个 1。
研究者建议通过调整成本分摊规则、利用需求响应机制及加强政策协调来平衡经济效益与社会公平 1。具体措施包括每年削减22个高峰时段的电力使用、各州直接按新增需求向数据中心收费,以及建立针对低收入家庭的补贴计划 1。
A study published in Nature Energy on November 9 reveals that the rapid expansion of artificial intelligence data centers is generating substantial grid upgrade costs, which are disproportionately borne by residents, particularly those with lower incomes and in rural areas 1. Conducted by researchers Fengqi You and Cheng Feng from Cornell University, the analysis covered four major U.S. power grids from 2020 to 2025 1. During this period, transmission investment surged from $2 billion to $34 billion 1.
The financial burden falls heavily on consumers because data centers are covering only a fraction of the infrastructure expenses associated with their growth 1. In 2025, the total bill for data center-related infrastructure reached $8.25 billion, yet data centers paid an average of just 38% ($3.12 billion), leaving the remaining 62% ($5.13 billion) to be absorbed by utility users 1. This cost-shifting has significant regional impacts: wholesale electricity prices rose by 15% in Virginia and 20% in the southern San Francisco Bay Area of California 1. For low-income households, these additional energy costs can consume up to 28% of their annual income 1.
Rural communities face even greater challenges, with the proportion of households experiencing high energy burdens—defined as spending more than 6% of income on energy—being 76% higher than in urban areas 1. Looking ahead, data centers are projected to consume 8% to 17% of total U.S. electricity by 2030, with 70% to 80% of that demand attributed to AI operations 1. While each gigawatt (GW) of data center capacity creates approximately 15,000 jobs, including about 2,700 directly linked positions, the researchers argue that economic benefits must be balanced against social equity 1. To address these disparities, the study recommends adjusting cost-allocation rules, utilizing demand response mechanisms such as reducing peak-hour electricity use during 22 periods annually, charging states directly for new demand, and establishing subsidy programs for low-income families 1.
评论
还没有评论,欢迎留下第一条。