Meta is pouring $10 billion into a single data center in El Paso, Texas — a sixfold increase from its original $1.5 billion commitment just five months ago. The announcement signals how aggressively hyperscalers are racing to build AI infrastructure capacity.
The Investment
Meta committed $1.5 billion to the El Paso facility in October 2025. On March 26, that number jumped to $10 billion. The data center will bring 1 gigawatt of power capacity online by 2028 — enough to power roughly 750,000 homes.
The project will create 300 permanent jobs and employ more than 4,000 construction workers at peak building. For El Paso, it represents one of the largest private investments in the city’s history.
The Hyperscaler Arms Race
Meta’s El Paso expansion fits into a broader pattern. The five largest US cloud and AI infrastructure providers — Microsoft, Alphabet, Amazon, Meta, and Oracle — have collectively committed between $660 billion and $690 billion in capital expenditure for 2026, nearly double 2025 levels.
The numbers by company:
- Amazon: $200 billion (largest capex commitment)
- Alphabet: $175-185 billion
- Meta: $115-135 billion
- Microsoft: $120 billion+
- Oracle: $50 billion
Roughly 75% of this aggregate spending — around $450 billion — will fund AI-specific infrastructure as cloud demand accelerates.
What’s driving this? All the hyperscalers report their markets are supply-constrained, not demand-constrained. There’s more customer demand for AI computing than available capacity. The race isn’t for customers — it’s for infrastructure.
Who Wins, Who Loses
Winners: Texas continues its aggressive data center recruiting. El Paso gets jobs, tax revenue, and economic diversification away from traditional border economy dynamics. Meta gets power access — West Texas has transmission capacity that Northern Virginia (the world’s largest data center concentration) lacks.
Losers: Power grids face unprecedented strain. US data centers now consume approximately 176 TWh of electricity annually — 4.4% of national power. That percentage is climbing fast. AI training and inference workloads are far more power-intensive than traditional cloud computing.
Energy costs are becoming a competitive factor in AI. Companies with better power access can train larger models more cheaply. Meta’s Texas bet is partly about securing that advantage.
The broader picture: Hyperscalers are increasingly using debt markets to fund these massive capex programs. The era of AI infrastructure being financed entirely from free cash flow is ending. These companies are leveraging up to win the infrastructure race, transforming what were historically cash-funded business models.
The El Paso data center won’t come online until 2028. By then, the AI infrastructure map will look very different — and the companies that moved fastest will have locked in the capacity advantages.