Meta’s AI Push Drives 34% Surge in Data Center Energy Use

Meta’s data center usage surges: Meta’s electricity consumption in its data centers reached nearly 15,000 GWh in 2023, marking a significant 34 percent increase as the company ramps up its AI infrastructure.

• The company’s total data center electricity usage hit 14,975,435 MWh, with a notable 97 percent jump in leased data center facility usage to 2,187,020 MWh.
• Meta’s Prineville, Oregon campus was the largest consumer of electricity among its 16 facilities, using 1,375,321 MWh.
• The company maintains that 100 percent of this electricity comes from renewable energy sources, primarily through Power Purchase Agreements (PPAs) and Renewable Energy Certificates (RECs).

Efficiency metrics and fuel consumption: Despite the increase in overall energy use, Meta has managed to maintain certain efficiency standards while reducing its reliance on traditional fuels.

• The Power Usage Effectiveness (PUE) remained steady at 1.08, indicating relatively efficient energy use in data centers.
• Natural gas consumption decreased by 36 percent to 4,818,116 therms, while diesel and gasoline usage also saw significant reductions.
• Meta has begun testing hydrotreated vegetable oil as a potential diesel replacement at its Clonee, Ireland site, with plans to evaluate its viability for global implementation throughout 2024.

Water usage and sustainability efforts: The company’s water consumption has increased, though efforts are being made to improve efficiency and implement water restoration projects.

• Data center water withdrawal rose by 7 percent to 3,881 megaliters, with a concerning 137 percent increase in groundwater withdrawal to 88 megaliters.
• Water withdrawals from areas with high or extremely high baseline water stress increased by 20 percent to 1,360 megaliters.
• On a positive note, water restoration efforts increased by 150 percent, reaching 5,889 megaliters.
• The Water Usage Effectiveness (WUE) improved slightly from 0.20 to 0.18.

AI-driven infrastructure expansion: Meta is actively expanding its data center capabilities to support its growing AI initiatives, which is driving up costs and resource consumption.

• The company spent $8.5 billion on capital expenditures in the last quarter, with expectations of further increases as it develops its generative AI infrastructure.
• New AI-optimized data centers under construction will feature dry-cooling technology, which uses air as the cooling medium to minimize water usage in suitable geographic locations.

Analyzing the implications: Meta’s substantial increase in data center usage reflects the growing demands of AI technology and raises questions about the sustainability of rapid digital infrastructure expansion.

• While the company is making efforts to use renewable energy and improve water efficiency, the sheer scale of resource consumption poses significant environmental challenges.
• The shift towards AI-optimized data centers with more efficient cooling systems suggests a recognition of the need to balance technological advancement with environmental responsibility.
• As Meta and other tech giants continue to expand their AI capabilities, the industry may face increased scrutiny and pressure to develop more sustainable practices to mitigate the environmental impact of data centers.