Meta and BlackRock Form $14B Joint Venture for 1-Gigawatt Texas AI Data Center

Meta Platforms and BlackRock announced Monday a joint venture to build and operate a 1-gigawatt AI data center campus in El Paso, Texas, at a total development cost of approximately $14 billion. The deal, structured as a sale-leaseback with institutional financing, represents a new template for how hyperscalers can fund the extraordinary capital requirements of AI infrastructure without completely draining their balance sheets — and it arrives just hours before Meta reports what analysts expect to be a landmark quarter for AI-driven advertising.

The Financial Architecture

The deal’s structure is more intricate than a simple partnership announcement. Meta will contribute the land and its in-progress construction assets, valued at approximately $2.3 billion, while BlackRock funds managed through its Global Infrastructure Partners arm and HPS Investment Partners will make a cash contribution of roughly $4.9 billion. An additional $12.5 billion in debt will be raised against the project, bringing total capitalization to the headline $14 billion figure.

BlackRock takes an 80% ownership stake in the joint venture; Meta retains 20%. But Meta will lease back the entire campus under a long-term agreement: a four-year initial term with four extension options, creating a potential 20-year total arrangement. The lease includes aggregate residual value guarantees of approximately $13 billion, declining over the lease term — a structure that effectively gives the lender confidence that the asset will retain value even in a scenario where Meta’s compute needs shift.

Meta CEO Mark Zuckerberg framed the deal in the company’s preferred idiom of AI ambition: “Building infrastructure for superintelligence ensures technology benefits reach everyone. Our BlackRock partnership enables faster, larger-scale operations combining our data center expertise with leading infrastructure investment.” BlackRock CEO Larry Fink, in language more familiar to infrastructure investors than AI enthusiasts, emphasized the local economic impact: “This El Paso campus will create skilled jobs and drive local economic growth.”

The El Paso Campus

The campus is already under construction, with more than 2,300 workers currently on site. At peak construction, the project is expected to employ more than 4,000 workers. Once operational — targeted for 2028 — the facility will support approximately 300 permanent jobs and will provide 1 gigawatt of computing capacity, making it one of the largest single-site AI compute facilities in the world when it comes online.

Meta has committed more than $10 billion to the El Paso project in total, including its land contribution and ongoing construction costs. The company has also announced $500,000 in community grants to El Paso public schools, a standard but modest gesture for a project of this scale.

Why Sell-Leaseback?

The strategic logic of the deal becomes clear against the backdrop of investor scrutiny over Big Tech’s AI capital expenditures. Meta raised its full-year 2026 capital expenditure guidance to a range of $125 billion to $145 billion — nearly double the prior year — and markets have responded with acute concern about the impact on free cash flow. Alphabet’s own recent earnings triggered a selloff in large-cap tech after its capex guidance spooked investors worried about cash generation.

The sale-leaseback structure allows Meta to accelerate its build-out of 1-gigawatt-class facilities while simultaneously moving the real estate ownership off its balance sheet. Under accounting rules, the lease obligations do appear as liabilities, but the structure nonetheless reduces the total equity capital that Meta needs to commit per gigawatt of capacity, effectively increasing the leverage of its infrastructure spending.

For BlackRock, the deal is an attractive infrastructure investment: a long-term lease from a creditworthy counterparty (Meta’s 2025 revenue exceeded $160 billion), secured against physical assets in a market where data center demand is projected to grow for at least a decade. Infrastructure funds have been seeking exactly this kind of risk-adjusted yield in a market where traditional real assets are priced richly.

A New Funding Template for AI

What makes this deal significant beyond its size is its potential as a replicable model. The hyperscalers — Meta, Google, Microsoft, Amazon, and a small set of AI-native operators — face a collective capital requirement for AI infrastructure that runs into the trillions of dollars over the next decade. Their balance sheets are large but not infinite, and institutional investors sitting on sovereign wealth funds, pension capital, and infrastructure mandates are desperate for exactly the kind of stable, long-duration, investment-grade cash flows that a leased data center provides.

Meta itself pioneered a version of this logic with its earlier Meta Compute neocloud initiative, where it began selling GPU compute capacity to third parties to defray infrastructure costs. The BlackRock deal extends the same logic to the real estate layer: rather than owning the building, Meta becomes a high-credit tenant, preserving capital for the hardware and software that it treats as core competitive assets.

The announcement is likely to be studied carefully by Microsoft, Amazon Web Services, and Google Cloud, all of which are managing similar tensions between massive infrastructure commitments and shareholder pressure on free cash flow. If the structure proves manageable from an accounting and operational standpoint, expect more hyperscaler-infrastructure fund pairings before the end of 2026.

The Earnings Backdrop

The timing of the announcement — the day before Meta’s Q2 2026 earnings report — was almost certainly deliberate. Ahead of a report where investors will scrutinize capex guidance more than nearly any other line item, Meta wanted to demonstrate that it has a credible mechanism for managing the capital burden of its AI ambitions without sacrificing operating leverage. Whether that framing succeeds will be visible in the stock price by Wednesday morning.

Analysts entering the print expect Meta to report approximately $60.2 billion in Q2 revenue, up about 27% year-over-year, with an EPS of approximately $7.18. The advertising engine continues to perform — Meta’s Family of Apps ad business generated $55 billion in Q1 revenue, up 33% year-over-year. The question, as always, is whether that machine can keep running faster than the cost structure expands.