A recent article from KQED highlights growing alarm over how the surge in artificial-intelligence (AI) demand is putting unprecedented stress on data-centre infrastructure — especially in places like California. It notes that data centres supporting AI workloads now require significantly more electricity and water than in past eras, and that information about their actual consumption remains limited. For example, regulators in the state have introduced bills requiring data-centres to report their energy and water use, but major legislation (such as the one tracking water use) was vetoed.
Experts quoted in the piece point out that while data centres have existed for decades, the “generative AI” wave (models for chatbots, image generation, etc.) has dramatically increased resource demands. One academic cited found that U.S. data-centres consumed about 3 % of national electricity in 2020 — and this could rise to 12 % by 2028 if trends continue. The leap is significant: cooling systems, server farms, and high-performance hardware all draw more power and generate more heat, meaning higher demand for cooling and thus more water in some climates.
In addition to the direct resource use, the article emphasises that the way usage is distributed over time matters. Even if a data-centre consumes the same total amount of water as an office building, concentrating that use during hot months creates different stresses on local water supplies and ecosystems. Thus, the environmental footprint isn’t just about totals, but timing, location and ecosystem sensitivity. Also, the regulatory and transparency gaps stand out: legislation to mandate reporting of water or energy use are proposed, but so far major action remains limited, leaving many decisions opaque.
In sum, the article suggests that the AI-driven data-centre boom is more than a tech-industry phenomenon — it has serious implications for resource planning, climate resilience, and local impact. It raises questions about how to balance innovation with sustainability, how regulators and communities need better data, and how companies must consider the environmental cost of next-generation infrastructure. If you like, I can pull out key numbers and projections from the article and related research.
