Meta is significantly upping its investment in the physical and technology infrastructure needed for its artificial intelligence goals. The company’s recent earnings report revealed plans to more than double spending on AI infrastructure, including data centers and servers. They project capital expenses for 2025 to be between $66 billion and $72 billion, about $30 billion above last year’s midpoint estimate. This growth trajectory will likely extend into 2026, with further substantial increases in spending to enhance their AI capabilities.
Meta’s CFO, Susan Li, emphasized that building superior AI infrastructure is crucial for creating effective AI models and user experiences. Investments are set to surge in 2026, with most AI costs expected to be self-funded, though financial partnerships for data center development are under consideration. The company also detailed two notable AI projects: Prometheus in Ohio, aimed to be one of the first superclusters reaching 1 gigawatt of compute power by 2026, and Hyperion in Louisiana, which could eventually scale to 5 gigawatts. Additional unnamed large-scale clusters are in development, though such projects will demand significant energy and may affect local communities, as seen in Newton County, Georgia, where one venture has already caused water shortages.
Employee compensation is anticipated to be a major factor in driving growth, as Meta plans to spend millions on attracting skilled AI engineers and researchers for its new Superintelligence Labs. After the earnings report, Meta’s stock jumped 10% in after-hours trading, reflecting positive investor sentiment regarding the company’s performance and optimistic outlook for the third quarter. They reported second-quarter revenues of $47.5 billion and expect third-quarter revenue between $47.5 billion and $50.5 billion, largely driven by advertising improved by AI tools. However, the Reality Labs segment recorded a loss of $4.5 billion. This article was updated to include additional funding strategies for Meta’s AI infrastructure projects.
The ainewsarticles.com article you just read is a brief synopsis; the original article can be found here: Read the Full Article…
