Meta boosts Texas AI datacenter investment to $10B; targets 1 GW by 2028

Meta raised its planned capital investment in an El Paso, Texas AI data center from $1.5 billion to over $10 billion, expanding the facility to one gigawatt of capacity by its 2028 launch. The increase reflects Meta's reassessment of AI compute demands as the company prepares for agent-scale inference workloads and expands its in-house AI model training capacity. The move comes amid Meta's broader infrastructure expansion, with total 2026 capital expenditures projected at $115 billion to $135 billion—nearly double the $72 billion the company spent in 2025.

Meta is consolidating its AI infrastructure strategy across custom silicon and heterogeneous hardware. The Texas facility represents a cornerstone of Meta's sovereign compute capacity, reducing reliance on third-party cloud providers and enabling the company to match workload optimization directly to its hardware stack. The 1 GW target positions Meta alongside other hyperscalers building similar-scale facilities: Amazon is targeting multiple gigawatts across Trainium deployments, Google has scaled TPU capacity across multiple regions, and Microsoft is adding gigawatts of capacity across Azure data centers.

The capex intensity reflects an industry-wide shift: frontier model inference is becoming increasingly capital-intensive as models scale to trillion-parameter scales and agents require sustained compute rather than sporadic API calls. Meta's internal restructuring, including 700 initial job cuts announced alongside the capex guidance, signals that infrastructure capex is now prioritized over headcount growth. The Texas facility's timeline (2028 completion) aligns with industry expectations that on-device inference will plateau by 2027-2028, forcing inference workloads to consolidate in hyperscale facilities optimized for agentic compute patterns.