Tech builds the data center. Wall Street becomes the landlord.

Meta is spending more on data centers this year than most countries spend on their national infrastructure budgets — and a growing share of that spending won’t show up as owned real estate on its own balance sheet. Under the joint venture structures Meta has now used twice, in Louisiana and in Texas, the company designs, builds, and operates the campus. An institutional capital partner owns it. Meta pays rent. The technology company is the builder and the tenant; Wall Street is the landlord.

Project 1: Louisiana — Blue Owl Becomes the Landlord

Hyperion, Meta’s Richland Parish campus, illustrated the template. Meta contributed the site and construction-in-progress to a joint venture with funds managed by Blue Owl Capital for a 20% equity stake; Blue Owl contributed roughly $7 billion in cash for the remaining 80%, and Meta received a $3 billion true-up distribution to align the split. The bulk of the financing came from the debt markets: a special-purpose vehicle, Beignet Investor, issued approximately $27 billion in project bonds, purchased by institutional investors led by PIMCO (~$18B) and BlackRock (~$3B). Meta leases the finished campus back as anchor tenant while handling construction and property management — builder, operator, and tenant of a campus it doesn’t own. Hyperion has since more than doubled in scope, to a 5-gigawatt facility now costing over $50 billion.

Project 2: Texas — BlackRock Becomes the Landlord

El Paso runs the same play, with BlackRock in the landlord’s seat. In July 2026, Meta and BlackRock-managed funds formed a $14 billion joint venture for a one-gigawatt campus — Meta’s third Texas data center. BlackRock’s funds hold 80% for roughly $4.9 billion in cash; Meta holds 20% via land and construction-in-progress (~$2.3B) and receives a $1 billion true-up distribution. The remaining ~$12.5 billion is financed through project-level debt, following the Beignet Investor pattern. Meta is sole initial tenant under a long-term lease, with the campus targeted to open in 2028.

The parallel is close enough that it reads as one financing structure run twice — with BlackRock appearing in both deals: bond buyer in Louisiana, majority owner in Texas.

Why This Structure Exists

1. Balance Sheet Discipline. Meta’s 2026 capex guidance runs $125–145 billion, with AI infrastructure as the largest driver. Contributing land and infrastructure rather than cash, and holding a minority JV stake, keeps the venture and its debt off Meta’s consolidated balance sheet — converting a lumpy capital expenditure into a predictable lease obligation.

2. Infrastructure-Grade Yield. To an allocator like Blue Owl or BlackRock, a data center with Meta as anchor tenant looks less like speculative real estate and more like infrastructure — a long-lived asset generating contracted income from an extraordinarily creditworthy tenant, over a term measured in decades.

3. Renewal Leverage Runs the Other Way. Today Meta is the anchor tenant every capital partner wants, and sets terms accordingly. At the end of a roughly 15-year lease, that leverage inverts: the landlord holds a purpose-built asset that’s costly for the tenant to leave and hard to re-let elsewhere.

4. Capital Concentration Among a Few Names. Blue Owl, BlackRock, and PIMCO have each shown up as equity, debt, or both across the sector’s largest transactions — a market-wide vantage point no individual technology company has.

What This Means for Developers and Capital Partners If you’re structuring a data center joint venture, expect capital partners to negotiate renewal, re-tenanting, and residual value protections as aggressively as the initial lease terms. If you’re advising an institutional client entering this space, treat SPE separateness, true-lease characterization, and cross-collateralization across future phases as first-order issues — not afterthoughts to be papered once the deal is signed.

Does This Give Wall Street More Power Over AI?

Over the substance of AI — models, compute allocation, product roadmap — probably not much. Meta keeps operational control, IP, and exclusive use of the capacity; Blue Owl and BlackRock own the building, not what runs on top of it. Over the pace and terms of the buildout, the picture looks different:

• Capital, not compute or talent, is becoming the binding constraint. Whoever controls the capital increasingly controls how fast any developer can build at scale — a form of gatekeeping power that requires owning no AI technology at all.
• Renewal leverage inverts over time. The landlord’s position strengthens materially once the initial long-term lease approaches expiration.
• A small set of institutional names keeps appearing across deals. If that concentration holds, a handful of asset managers end up as counterparty to most major hyperscalers at once.
• A more pointed critique exists outside the financial press, framing these deals as private capital enclosing the physical layer of AI, with durable returns accruing to infrastructure owners rather than tenants.

The more grounded read: this is less Wall Street chasing relevance in AI than AI’s capital needs outgrowing what any single balance sheet can absorb. Whether that dependency hardens into durable institutional power will likely turn on how the first wave of these leases actually gets renegotiated — for Hyperion, still roughly a decade away. If that trend continues, some of the most valuable AI assets may end up owned not by technology companies, but by the infrastructure investors financing them.

Implications for Practice

For Real Estate and Structured Finance Counsel: This is the same body of law we already know — ground leases, sale-leaseback structuring, SPE and bankruptcy-remoteness covenants, project finance intercreditor arrangements — applied to a tenant and asset class the market hasn’t fully priced or litigated yet.

For Institutional Capital Advisors: Underwrite these leases the way you’d underwrite any single-tenant infrastructure asset — residual value, re-tenanting risk, and technological obsolescence deserve the same weight as counterparty credit quality.

For Corporate and Developer Counsel: Structure JV governance, capital call obligations, and phase-expansion mechanics before the relationship is tested by a market downturn or a shift in either party’s strategy.

As institutional capital keeps flowing into data center joint ventures at this scale, expect the legal work to look increasingly like structured finance and real estate private equity work, with the added complexity of a technology tenant whose product cycle moves far faster than the assets it runs on.

Disclaimer:

This publication is for general informational purposes only and does not constitute legal advice. The information herein may not reflect the most current legal developments and should not be relied upon as a substitute for consultation with qualified legal counsel. Readers should seek professional advice tailored to their specific circumstances.