Meta has lifted its planned spending on a major Texas data centre to $10 billion, sharply expanding a project in El Paso that the company says is designed to support its accelerating artificial intelligence ambitions and to reach as much as 1 gigawatt of capacity by 2028. The move turns what had been framed as a $1.5 billion first phase into one of the largest single-site digital infrastructure commitments in the state, underlining how the race to secure AI computing power is reshaping corporate investment plans across the United States.
The enlarged commitment places Meta alongside the biggest spenders in the technology sector as companies pour cash into data centres, advanced chips and power contracts to keep pace with demand from AI models and cloud services. Reuters reported that Meta’s El Paso facility is expected to become the group’s 29th data centre worldwide and its third in Texas, with the company pointing to the scale of future computing needs as the rationale for the expansion. The investment also comes amid a broader industry surge that analysts say is pushing cumulative AI-related capital expenditure by the largest technology groups to extraordinary levels this year.
Meta’s decision carries economic and political significance for Texas as well as for the company’s own growth strategy. According to the latest details, the site is expected to support about 300 permanent jobs once operational, while peak construction activity could involve more than 3,000 workers. Earlier reporting on the first phase had put those figures materially lower, showing how quickly the scope has changed as Meta recalibrates the project from a more modest launch into a long-horizon AI campus. That evolution mirrors a pattern across the sector, where companies initially announce one tranche of spending before scaling up once land, permits and grid assumptions are in place.
El Paso’s appeal reflects several forces that are drawing hyperscale developers towards Texas: land availability, energy access, a regulatory climate viewed as comparatively favourable, and a labour market that can support large industrial construction. The state has become a focal point for data-centre expansion well beyond Meta, with other technology groups also committing billions of dollars as AI workloads drive demand for bigger and more power-hungry campuses. That concentration is turning Texas into one of the key battlegrounds in the contest over where the infrastructure behind generative AI will be built and how fast it can be connected.
Power remains central to the story. A one-gigawatt campus sits at the upper end of planned data-centre developments and highlights how AI is changing the physical scale of digital infrastructure. Meta has said it intends to contribute more than 5,000 megawatts of clean energy to the Texas grid and to pair the development with water-related community initiatives. Those pledges are likely to be scrutinised closely, because the sector is facing mounting questions over electricity consumption, transmission bottlenecks, cooling requirements and the environmental cost of ever larger computing clusters. Across the industry, investors and policymakers are increasingly focused not only on how much capacity companies want, but on where the power will come from and who bears the cost of the supporting infrastructure.
For Meta, the Texas expansion fits a wider campaign to lock in compute capacity as it pushes deeper into AI products, advertising tools and large-model development. The company has already disclosed other large-scale data-centre projects, including a $10 billion facility in Indiana and a far larger buildout in Louisiana linked to external financing. Taken together, those projects suggest Meta is moving beyond incremental server expansion towards a network of giant campuses intended to underpin both consumer-facing AI features and the back-end systems needed to train and run more sophisticated models.
That ambition, however, comes with complications. The same forces propelling investment are also creating execution risks, from transformer shortages and grid-connection delays to rising construction costs and concern that spending may outrun near-term returns. Reuters Breakingviews noted this week that the technology sector’s AI splurge could fall short of expectations not because demand is weak, but because physical constraints are becoming harder to ignore. Meta’s enlarged Texas commitment therefore stands as both a statement of confidence and a test of whether the industry can translate financial firepower into functioning, timely infrastructure on the ground.
The enlarged commitment places Meta alongside the biggest spenders in the technology sector as companies pour cash into data centres, advanced chips and power contracts to keep pace with demand from AI models and cloud services. Reuters reported that Meta’s El Paso facility is expected to become the group’s 29th data centre worldwide and its third in Texas, with the company pointing to the scale of future computing needs as the rationale for the expansion. The investment also comes amid a broader industry surge that analysts say is pushing cumulative AI-related capital expenditure by the largest technology groups to extraordinary levels this year.
Meta’s decision carries economic and political significance for Texas as well as for the company’s own growth strategy. According to the latest details, the site is expected to support about 300 permanent jobs once operational, while peak construction activity could involve more than 3,000 workers. Earlier reporting on the first phase had put those figures materially lower, showing how quickly the scope has changed as Meta recalibrates the project from a more modest launch into a long-horizon AI campus. That evolution mirrors a pattern across the sector, where companies initially announce one tranche of spending before scaling up once land, permits and grid assumptions are in place.
El Paso’s appeal reflects several forces that are drawing hyperscale developers towards Texas: land availability, energy access, a regulatory climate viewed as comparatively favourable, and a labour market that can support large industrial construction. The state has become a focal point for data-centre expansion well beyond Meta, with other technology groups also committing billions of dollars as AI workloads drive demand for bigger and more power-hungry campuses. That concentration is turning Texas into one of the key battlegrounds in the contest over where the infrastructure behind generative AI will be built and how fast it can be connected.
Power remains central to the story. A one-gigawatt campus sits at the upper end of planned data-centre developments and highlights how AI is changing the physical scale of digital infrastructure. Meta has said it intends to contribute more than 5,000 megawatts of clean energy to the Texas grid and to pair the development with water-related community initiatives. Those pledges are likely to be scrutinised closely, because the sector is facing mounting questions over electricity consumption, transmission bottlenecks, cooling requirements and the environmental cost of ever larger computing clusters. Across the industry, investors and policymakers are increasingly focused not only on how much capacity companies want, but on where the power will come from and who bears the cost of the supporting infrastructure.
For Meta, the Texas expansion fits a wider campaign to lock in compute capacity as it pushes deeper into AI products, advertising tools and large-model development. The company has already disclosed other large-scale data-centre projects, including a $10 billion facility in Indiana and a far larger buildout in Louisiana linked to external financing. Taken together, those projects suggest Meta is moving beyond incremental server expansion towards a network of giant campuses intended to underpin both consumer-facing AI features and the back-end systems needed to train and run more sophisticated models.
That ambition, however, comes with complications. The same forces propelling investment are also creating execution risks, from transformer shortages and grid-connection delays to rising construction costs and concern that spending may outrun near-term returns. Reuters Breakingviews noted this week that the technology sector’s AI splurge could fall short of expectations not because demand is weak, but because physical constraints are becoming harder to ignore. Meta’s enlarged Texas commitment therefore stands as both a statement of confidence and a test of whether the industry can translate financial firepower into functioning, timely infrastructure on the ground.
Topics
Technology