Meta has announced that it is building a new gigawatt-sized data center in Texas in order to advance its AI efforts. The new data center, which can scale up to 1GW, will be located in El Paso, Texas, and will mark the 29th such facility for the company.
The company behind social media giants like Instagram and Facebook says it has designed the El Paso data center to have systems that can support both the traditional servers of today and the future generations of AI-enabled hardware.
Meta says it will spend over $1.5 billion to build the new data center, which is expected to be completed by 2028. The facility will support around 100 operational jobs and is expected to employ over 1,800 construction workers on site at its peak.
The company says that it has invested over $10 billion across the three data centers in Texas and has over 2,500 full-time employees across its data centers, offices, and research labs.
The El Paso data center will use 100% renewable energy, and Meta has also committed to restoring 200% of the water consumed by the data center to local watersheds.
Meta says El Paso’s robust infrastructure and skilled workforce played a part in choosing it for the data center.
Meta’s AI Spending:
Meta has been focusing aggressively on AI in the last few years, where it is competing against the likes of OpenAI, Google, and Anthropic in the race towards superintelligence via their AI models. Earlier this year, Meta set up a Superintelligence Labs and hired former Scale AI CEO Alexandr Wang to be its CEO while poaching a lot of the top talent from other companies.
The tech giant has also committed to spending $72 billion in capital expenditures this year. CEO Mark Zuckerberg has previously said that it is better to overspend than underspend on AI, given that the risk of missing the emergence of superintelligence is greater than the risk of financial waste.
Superintelligence or AGI is a hypothetical state where AI systems are said to be able to perform most tasks at the same proficiency levels, if not better. While there is no well-defined definition for this stage, most AI labs are aiming towards it and are spending heavily on building their AI infrastructure.
