Ais Green Shadow: Genai Expansion Fuels Sudden Sustainability Reassessments Across Industries

The use of generative artificial intelligence (AI) is growing rapidly, but its environmental impact is becoming increasingly clear. A new report from the Capgemini Research Institute found that 47% of organizations implementing GenAI across most or all functions have had to re-evaluate their original sustainability goals.

Google was recently criticized for its significant increase in emissions, which rose by 48% over four years due to the expansion of its data centers to support AI development. The company’s goal to reach net-zero emissions across all operations and value chains by 2030 is now considered “extremely ambitious” and will require navigating significant uncertainty.

The report surveyed executives from 2,000 large organizations worldwide that are already working with GenAI. Almost half of these organizations reported an increase in greenhouse gas emissions over the past year, with a similar proportion attributing the rise to their AI usage.

GenAI has a substantial environmental impact due to the high energy and water required for its operation. Graphics processing units (GPUs) used in AI development require rare earth metals that must be mined, releasing greenhouse gases. The hardware behind GenAI also requires frequent upgrades, which could create up to 5 million tons of electronic waste by 2030.

Training large language models like OpenAI’s GPT-4 consumes significant amounts of energy, equivalent to the annual power usage of 5,000 U.S. households. Additionally, running inferences on these models requires substantial amounts of water, with a single inference using between 10 and 50 queries consuming around 500 milliliters.

The European Union has set a goal to reduce greenhouse gas emissions by at least 11.7% below projected levels by 2030. However, the growing demand for data centers, driven in part by AI usage, could push this goal out of reach, with an estimated increase in energy demand from bit barns of up to 3%.

Despite the significant environmental impact, many businesses are not aware of or prioritizing the emissions associated with their GenAI usage. Only 38% of executives surveyed claimed to be aware of the environmental impact of the GenAI they use, and just 12% said their company measures its footprint.

Executives may prioritize cost competitiveness over environmental concerns when selecting or building GenAI models. This is because organizations quickly become sensitive to inference costs as they scale up AI usage. As a result, they often opt for less energy-intensive models that can reduce carbon impact.

The growing use of GenAI has significant environmental implications that must be taken seriously by organizations and policymakers. By prioritizing sustainability and transparency in AI development, we can work towards reducing the negative impacts of this technology on the environment.