Data center energy demand surge driving fossil fuel lock-in and emissions spikes
Key Questions
What are the projected emissions impacts from new data centers in Pennsylvania and Texas?
Pennsylvania could see an additional 68 million tons of CO2e annually, representing a 24% increase from new gas plants. Texas has 63 GW of behind-the-meter gas capacity under development, which would emit 287 million tons of CO2e per year.
How is Google's electricity consumption and emissions changing due to data centers?
Google's electricity use jumped 37% in 2025, while its Scope 3 emissions have doubled since 2019. This highlights the gap between renewable certificate accounting and actual grid emissions.
What material demands are AI data centers driving for copper and uranium?
Copper demand from AI data centers could reach 1.2 million tonnes by 2030, with a 3-4x multiplier when including grid wiring. Uranium demand for new reactors could add 15 million pounds per year by 2030.
Multiple articles quantify the massive energy, emissions, and material impact of AI data centers. Pennsylvania faces 68M tons CO2e (24% increase) from new gas plants; Texas has 63 GW behind-the-meter gas capacity under development (287M tons CO2e annually); global data center electricity could reach 1,050 TWh by 2026 (IEA). Google's electricity use jumped 37% in 2025, with Scope 3 emissions doubling since 2019, exposing the gap between renewable certificate accounting and physical grid emissions—a structural policy failure. A new analysis adds material demand: copper demand from AI data centers could reach 1.2 million tonnes by 2030, with a 3-4x multiplier for grid vs. internal wiring; uranium demand for new reactors could add 15 million pounds per year by 2030. This challenges clean-energy narratives and exposes regulatory gaps in grid planning and climate policy.