Meta boosts West Texas AI data center investment to $10 billion amid legal and business pressures

Meta Platforms has expanded its commitment to a West Texas AI data center by more than sixfold, bringing total planned spending on the project to 10 billion dollars. The announcement landed in the same week that two separate juries found Meta liable in child harm cases, and a New Mexico court ordered the company to pay 375 million dollars in damages. The timing is awkward, but the spending decision itself was not made this week. Data center investments at this scale take years to plan, and Meta is not going to pause a 10 billion dollar infrastructure project because of a bad news cycle.

What Meta is actually building in West Texas

The West Texas facility is one of several major data center projects Meta has underway in the United States. The company has not disclosed the exact location within the state, but West Texas offers a combination of cheap land, access to renewable energy from wind farms in the Permian Basin region, and proximity to existing power grid infrastructure. Data centers at this scale require enormous amounts of electricity. A 10 billion dollar facility will likely consume multiple gigawatts of power at peak capacity, making the local energy supply as important a factor as construction costs.

Meta is building this facility to run its AI training and inference workloads. The company's AI ambitions span multiple product lines, including Meta AI, the assistant built into WhatsApp, Instagram, and Facebook, as well as the Llama family of open-weight models that Meta releases publicly. Training the next generation of Llama models requires substantially more compute than previous versions, and renting capacity from cloud providers at that scale becomes more expensive than owning the infrastructure outright.

Meta's broader AI infrastructure spending plan for 2025

Meta updated its full-year 2025 capital expenditure guidance in January 2025, raising it to a range of 60 billion to 65 billion dollars. That figure is up from approximately 37 billion dollars in capital expenditure during 2024. The West Texas expansion is one piece of a much larger infrastructure build. Meta is also constructing a data center in Louisiana that it describes as the largest in the world, covering approximately 4 square kilometers when complete.

The scale of this spending reflects a strategic bet that proprietary AI infrastructure will be a competitive advantage over the next five to ten years. Meta's chief financial officer Susan Li told analysts in January 2025 that the company expects AI to drive meaningful improvements in advertising efficiency and user engagement across its apps. The revenue logic is that better AI recommendations keep users on the platform longer, which increases the number of ads served, which improves Meta's revenue per user without requiring a proportional increase in the user base.

The legal verdicts that landed in the same week

Two jury verdicts found Meta liable in child harm cases this week, and a New Mexico state court awarded 375 million dollars in damages related to the exploitation of minors on Meta's platforms. These are not the first legal actions of this kind against the company. Meta faces litigation from dozens of state attorneys general over allegations that Instagram and Facebook caused harm to children through addictive design features and inadequate protections against predatory behavior.

The 375 million dollar New Mexico award is significant as a single verdict, but it is unlikely to move Meta's financials in a material way on its own. Meta reported net income of 62.4 billion dollars in fiscal year 2024. The more consequential risk is if these verdicts establish precedents that lead to much larger consolidated damages across the full body of state and federal litigation, or if they push Congress to pass legislation requiring structural changes to how Meta operates its platforms for younger users.

Why Meta's stock fell 8% despite the data center news

Meta's stock dropped 8% in a single session this week, and the data center announcement did not offset the selling pressure. The legal verdicts were part of the reason, but the broader context matters. The Nasdaq 100 was already under pressure from the memory chip sell-off triggered by Google's AI memory research. Meta carries a large weighting in the Nasdaq 100, so when the index is falling, Meta tends to amplify the move in both directions. A day when sentiment is already negative is a bad day to absorb litigation news.

Meta's stock had been trading near all-time highs before this week's decline, up roughly 25% over the prior twelve months. At that valuation, the market had priced in continued strong advertising growth and the AI investment paying off over time. Any news that introduces doubt about either of those assumptions, whether through legal liability or questions about AI spending returns, gets punished quickly in a stock that is already priced for good outcomes.

What the Texas investment signals about Meta's AI strategy

Meta releasing the Llama model series as open-weight software means the company does not generate direct revenue from those models the way OpenAI does through its API. The return on Meta's AI infrastructure investment comes through its own products. Every dollar spent on data centers needs to justify itself through better ad targeting on Facebook and Instagram, more engaging AI features on WhatsApp, and the ability to attract developers who build on top of Llama, which in turn strengthens Meta's position in the AI developer community.

The West Texas facility is expected to be operational in phases starting in 2026. Meta has not confirmed a specific timeline for when the full 10 billion dollar investment will be deployed, but data center construction at this scale typically runs three to five years from groundbreaking to full capacity. The next concrete milestone investors will watch for is Meta's Q1 2025 earnings report, scheduled for late April 2025, where management is likely to provide an updated view on construction timelines and the AI products this infrastructure is expected to power.