Meta has recently signed three agreements to purchase nearly 1 gigawatt of solar power, demonstrating its commitment to powering its AI initiatives with renewable energy.

These deals bring Meta’s total solar power procurement to over 3 gigawatts this year. Solar energy’s cost-effectiveness and rapid deployment make it a favored choice for tech companies expanding their data center operations.

The recent agreements include two projects in Louisiana, totaling 385 megawatts. Meta will purchase the environmental attributes of the electricity generated, with both projects expected to be operational in two years.

These agreements follow a larger deal announced earlier this week, where Meta secured 600 megawatts from a solar farm near Lubbock, Texas. This project is also slated for commercial operations in 2027. While the Texas plant won’t directly power Meta’s data centers, it will feed into the local grid, offsetting the energy used by these facilities.

The Louisiana deals involve environmental attribute certificates (EACs), which allow Meta to offset its carbon emissions. However, some experts have raised concerns that EACs may not accurately reflect the true carbon footprint of tech companies, especially as AI drives up electricity consumption.

Environmental attribute certificates (EACs), also known as renewable energy certificates, have faced scrutiny for potentially masking the actual carbon footprint of tech operations, which have expanded significantly with the rise of AI-driven electricity demand.

Originally, EACs were introduced to incentivize renewable energy development when it was more expensive than fossil fuels. They allowed companies to offset emissions by paying extra to support renewable projects. However, with the decreasing costs of solar and wind power, the effectiveness of EACs in stimulating additional renewable energy development is being questioned. Some argue that companies should focus on directly encouraging the construction of new renewable energy capacity to genuinely offset their energy usage from AI and other power-hungry applications.