Meta
2026-07-29
Meta and BlackRock Form $14B JV for 1GW AI Data Center, Meta Leases Compute to Anthropic
Summary
Key Takeaways
Meta and BlackRock announced a $14B joint venture to build a 1GW AI data center campus in El Paso, Texas. BlackRock-managed funds take 80% ownership via $4.9B cash and $12.5B bonds issued July 27. Meta retains 20% and receives $1B distribution. The campus goes online in 2028, creating 4,000+ construction jobs and 300 operational jobs. Reuters reports Meta is in talks to lease compute to Anthropic for up to $10B over two years, positioning Meta as an AI cloud provider. The $14B is development cost only; a 1GW data center typically costs $35-50B all-in with chips. This mirrors Meta's Hyperion campus structure where Blue Owl Capital took 80% via $27B in project bonds. Meta Q2 earnings reporting July 29 with CapEx forecast update expected. The joint venture model allows Meta to expand capacity without full balance sheet impact, while the lease to Anthropic signals a strategic shift from AI model developer to infrastructure provider, potentially reshaping the AI cloud market.
Why It Matters
Meta's move is a defensive play against AWS, Azure, and GCP's dominance in AI cloud, while encircling OpenAI by supplying compute to Anthropic. The $14B development cost hides the true $35-50B total cost including chips, implying massive future capital needs. The 2028 timeline risks technological obsolescence of current-gen GPUs like H100. The joint venture is a form of off-balance-sheet financing, masking Meta's leverage. Anthropic's reliance on Meta's infrastructure could lead to ecosystem lock-in via PyTorch and Llama stacks. Meta's lack of enterprise cloud experience raises concerns about multi-tenancy and reliability.
PRO Decision
【Vendors】Competitors (Microsoft, Google, Amazon, NVIDIA) should accelerate their own AI data center builds and compute leasing programs to counter Meta. NVIDIA must diversify GPU supply to prevent Meta from locking exclusive capacity. Highlight Meta's lack of enterprise cloud experience.
【Enterprises】CIOs and architects should conduct zero-trust audits: demand cross-cloud portability tools to avoid PyTorch/Llama lock-in, require independent benchmarks for compute performance, and maintain multi-cloud strategies to reduce vendor concentration risk.
【Investors】Look beyond the PR: Meta's off-balance-sheet financing masks true leverage. Lease revenue depends on Anthropic, which may eventually build its own capacity. The 2028 timeline risks GPU obsolescence. Monitor Meta's ability to transition from model developer to infrastructure operator.
Get 3-5 key AI infrastructure signals weekly →
💬 Comments (0)