Local Climate Actions
Data centers, warehouses that contain ranks of computer servers and internet connection junctions, are the backbone of the internet and an essential part of modern infrastructure. The rush by big tech firms to develop and monetize artificial intelligence (AI) is leading to an outsized demand for new data centers.
AI tools like ChatGPT (GPT stands for generative pre-trained transformer) are also known as large language models (LLMs). They are trained on data in the form of human language and generate answers in human language that can be understood by an ordinary computer user. Generative AI does not really “think” — it produces answers based on statistical information found in datasets of human knowledge and associative leaps from patterns taught to it in the training process.
AI consumes huge amounts of computing power and storage, all housed in data centers which require lots of electricity. It’s estimated that a ChatGPT query uses ten times the energy of a Google search, while AI photo, music and video creation use much more.
Enormous new data center capacity is planned to service the expansion of AI. Data centers were already consuming 3-4% of U.S. electricity demand at the start of 2025 and are projected to use 8-12% by 2030. About half of the electricity used to power data centers in early 2025 was from fossil fuel plants so the climate emissions are significant. Hyperscale data centers specifically designed for AI use are even more energy intensive.
There are signs that the explosive growth of the AI industry is not sustainable. Funding has been pouring into speculative AI startup projects with uncertain outcomes. According to a 2025 MIT report, 95% of businesses that have incorporated early AI tools found that they did not increase profitability. AI financing has become a shell game as companies try to stay in the game but off-load risk.
Stanford AI experts are predicting that data center construction is also a bubble that may deflate in 2026. The medium time-frame needed to build data centers is moving faster than the long time-frame it may take to adopt AI products.
Local opposition to data centers is growing. The impacts of new data center construction are felt physically in communities, with increased construction traffic, inappropriate siting, demand for water for cooling, and potential strain on the electricity grid. New centers without power hookups to the local grid sometimes use diesel or gas generators that pollute the air. Some sit idle.
Community opposition has led to larger-scale protest, with a coalition of 230 environmental groups calling for regulations and a national moratorium on new data centers. In Virginia and New Jersey, data center opposition is thought to have affected state-level elections’ while a group called Data Center Watch claims that local opposition blocked or delayed $64 billion of data center projects in 24 states. Worry about rising electrical bills and water problems are fueling opposition.
From a climate perspective, AI is sucking up the investment dollars and electricity needed for rapid electrification as well as encouraging new use of natural gas and coal. A measure for a national moratorium on data center construction was introduced in the Senate by Bernie Sanders but failed. Cities and counties are considering moratoriums, and 238 data center-related bills were introduced in 2025 over all 50 states. Some regulations being considered for new data centers include energy efficiency requirements, water-use disclosure, grid infrastructure funding and clean-power incentives.
A new Sierra Club report, Data Center State Policies 2026, contains recommendations for state utility regulators and policymakers dealing with data centers.These guidelines are intended to protect customer electricity rates and public health, mandate transparency, and prioritize on-site low-carbon energy use. Check the webpage Demanding Better for more information on this issue.