BlackRock Raises $12 Billion for Meta's Texas Data Center — and Owns It Itself
BlackRock is raising at least $12 billion through bonds to build a data center for Meta in El Paso, Texas. JPMorgan and Morgan Stanley are running the offering. The structure itself is unusual: BlackRock units receive 80% of the facility, so BlackRock will simultaneously own the data center, finance it, and lease it to Meta. *Meta is designated an extremist organization and banned in Russia.
AI-processed from TNW; edited by Hamidun News
On July 20, 2026, BlackRock began raising at least $12 billion through bond sales to build a data center for Meta in El Paso, Texas. JPMorgan and Morgan Stanley are arranging the offering, and BlackRock's infrastructure and private-credit units will receive 80% of the entity that will own the facility, finance it, and lease it to Meta itself.
How the deal is structured
BlackRock's infrastructure and private-credit units control 80% of the entity that will finance, build, and own the data center, with Meta as its tenant. According to The Wall Street Journal, JPMorgan and Morgan Stanley began pitching investors on July 20, 2026, offering at least $12 billion in bonds for the El Paso project.
The arrangement loops back to a single asset manager. BlackRock ends up as the facility's owner, the deal's lender, and its landlord at once. That is a rare configuration for the market: the entire chain from capital to lease runs through one player, rather than being split among a developer, a bank, and an operator.
- Offering size: at least $12 billion in bonds
- Arrangers: JPMorgan and Morgan Stanley
- Investor pitch began on July 20, 2026
- Location: El Paso, Texas
- Stake held by BlackRock and its units: 80%
- Tenant of the facility: Meta
Why Meta isn't building it itself
Leasing instead of buying moves billions in capital expenditure off Meta's balance sheet: BlackRock takes on the construction and ownership of a data center costing more than $12 billion, while Meta pays for access to the capacity. That lets Meta scale up AI compute without inflating its own capex in the middle of a race where hyperscalers are pouring tens of billions of dollars into infrastructure.
Similar arrangements, in which an outside investor owns a data center and leases it to a tech giant, increasingly accompany the largest AI buildouts: demand for compute is growing faster than companies are willing to carry such assets on their own balance sheets.
For BlackRock, the deal is a way to turn AI infrastructure into a financial product investors can easily understand: bonds backed by a specific facility with an anchor tenant on the scale of Meta. As The Next Web notes, the intrigue lies in the structure of the deal, not in the fact that yet another data center is being built.
"The whole point is the structure of the deal,"
The Next Web writes about the $12 billion offering.
What this means
Financing for AI infrastructure is moving away from the "hyperscaler builds it itself" model toward complex structures in which the owner, lender, and landlord are the same asset manager. Data centers for artificial intelligence are becoming a distinct class of debt assets, and $12 billion for a single facility in El Paso shows how much capital the race for compute is drawing in. For investors, it is a new way to profit from the AI boom: not by buying shares of tech giants directly, but by putting money into debt backed by physical infrastructure.
Frequently asked questions
Who will own Meta's data center in Texas?
BlackRock will own the data center: its infrastructure and private-credit units hold 80% of the entity, while Meta is the tenant of the El Paso facility.
How much is BlackRock raising for Meta's data center?
At least $12 billion through bond sales. JPMorgan and Morgan Stanley are arranging the offering, and they began pitching the securities to investors on July 20, 2026.
*Meta is designated an extremist organization and banned in Russia.
Access Frontier LLMs & Multimodal AI in One Workspace
Discover NomadNet AI (nomadnet.ai) — unified access to frontier language models, research synthesis, and multimodal workflows.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.