• VIX
    Loading…
  • BIST 100
    Loading…
  • UST Yield 10y
    Loading…
  • S&P 500
    Loading…
  • Brent Petrol
    Loading…
  • XAU/TRY
    Loading…
  • EUR/TRY
    Loading…
  • USD/TRY
    Loading…
  • XAU/USD
    Loading…
  • EUR/USD
    Loading…

/

Kategori

/

Meta just sold 80% of a $14 billion Texas data center to pay for AI

Meta just sold 80% of a $14 billion Texas data center to pay for AI

Photo: iam luisao

Meta just sold most of a $14 billion data center it is already building, and the buyer is not another tech company. It is BlackRock, the world's largest asset manager, putting up nearly $5 billion in cash and financing another $12.5 billion in debt to take an 80% stake in a campus Meta had already started constructing in El Paso, Texas.

That is a striking move for a company worth over a trillion dollars. Meta is not doing this because it cannot afford the bill. It is doing this because the AI bill is so large that even the richest tech companies are looking for ways to spread the cost around.

What the deal actually looks like

Meta contributes land and construction work already underway, valued at roughly $2.3 billion. BlackRock brings $4.9 billion in fresh cash. BlackRock-managed funds own 80% of the resulting venture; Meta keeps 20%. Meta also receives a $1 billion distribution as part of the ownership alignment. The campus, once complete in 2028, is designed to run at 1 gigawatt of computing capacity, enough to power the AI systems behind Meta's social apps, its AI assistant, its ad tools, and its smart glasses.

One gigawatt is a lot. For reference, Meta is also building a separate campus in rural Louisiana that it expects to eventually reach 5 gigawatts, with investment climbing past $50 billion on its own.

Why this tells you something larger

Meta has said it plans to spend $600 billion building data centers by 2028. That figure is almost incomprehensible, and it explains why the company is turning to outside capital rather than funding everything from its own balance sheet. Debt markets and asset managers like BlackRock are being recruited to carry a portion of the load in exchange for ownership stakes and steady long-term returns on infrastructure.

This is a relatively new financing model for tech, borrowed from the playbook used to fund airports, toll roads, and pipelines. Infrastructure investors like BlackRock are comfortable holding assets for decades at predictable yields. AI data centers, it turns out, look a lot like infrastructure: expensive to build, long-lived, and essential once they exist.

The arrangement benefits both sides. Meta gets to build faster than it could otherwise and frees up capital for other bets. BlackRock gets a long-term asset tied to one of the most consequential technology shifts in a generation.

For El Paso, the announcement means construction jobs in the near term and, once the campus opens in 2028, a large permanent facility on the Texas-New Mexico border that will consume enormous amounts of power and water. Those local resource questions tend to follow gigawatt-scale data centers wherever they land.

For Meta's investors, the picture is more complicated. Shares have fallen about 10% this year as Wall Street grows impatient with the scale of AI spending. The El Paso deal offloads some of that spending, but the core concern remains: Meta is making enormous bets on AI products, including a personal AI assistant and image-to-video advertising tools, that have not yet generated the revenue needed to justify the infrastructure behind them. Second-quarter results were due the day after this deal was announced, which means investors got a very quick read on whether the spending is paying off.

The broader pattern is worth watching. If Meta can finance AI infrastructure the way utilities finance power plants, the constraint on AI buildout shifts from "how much can the tech giants afford" to "how much appetite does global capital have for these assets." Given how much institutional money is chasing long-duration returns right now, that appetite appears to be very large.