[ad_1]
Global emissions traced back to electricity consumption from AI chipmaking rose by 357% in 2024, with the environmental impacts of the technology extending far beyond power-hungry data centers.
According to research released by Greenpeace East Asia on April 10, global electricity consumption for AI chipmaking grew by 350% between 2023 and 2024. Electricity demand from AI chipmaking is also expected to increase 170-fold by 2030.
“The rapidly rising energy costs of AI data centers have captured global headlines, yet the environmental implications of other parts of the hardware lifecycle are often overlooked,” said Alex de Vries, a co-author of the report and the founder of sustainability outlet Digiconomist. “The manufacturing process of AI hardware carries a significant environmental footprint.”
Read More: Data Is Environmentally Dirty. What to Do?
As the report points out, the world’s biggest suppliers of AI chips base their manufacturing facilities in East Asia, where power grids are predominantly powered by fossil fuels such as coal, oil and natural gas. The process to produce those chips is extremely energy-intensive, where a wafer is sliced from a bar of silicon, is cleaned and polished, then is printed with complex circuit designs, and diced into individual chips using a diamond saw, before it’s packaged into a final product to be used in any number of electronic devices.
Citing separate research from McKinsey, the report projects that growth in demand for AI chips will necessitate the production of up to 32.5 million additional wafers by 2030. Emissions over that period are also expected to double, resulting in an estimated 16 million metric tons of CO2. That’s without factoring in the substantial and well-publicized demand on electricity grids created by AI data centers, with Goldman Sachs predicting a 160% growth in data center power demand by 2030. To wit, the International Energy Agency estimates that a single ChatGPT query requires 2.9 watt-hours of electricity, compared to 0.3 watt-hours needed for a standard Google search.
[ad_2]
Source link
