Meta signed a $27 billion five-year contract with Nebius for AI data center capacity, while Alphabet announced a $40 billion investment in Texas AI data centers.1 The deals represent the largest single commitments to AI-specific infrastructure to date.
The capital flow differs from traditional cloud infrastructure investments. AI workloads require higher power density, specialized cooling systems, and proximity to energy sources. IREN acquired a 1.6-gigawatt site in Oklahoma to meet this demand.1
New Era Energy & Digital closed multiple financings and signed a non-binding letter of intent with Stream Data Centers for Texas Critical Data Centers LLC.1 E. Will Gray II noted the facility is "strategically positioned for near-term development and power delivery, presenting a compelling opportunity for data center operators."1
Power capacity has become the primary constraint. Traditional data centers typically operate at 5-10 kilowatts per rack. AI facilities require 30-50 kilowatts or more. This gap is driving development of greenfield sites with dedicated power infrastructure rather than retrofitting existing facilities.
The infrastructure shift is creating opportunities for specialized developers. NN, Inc. stated "Electric Grid and Data Center is on a plan to become NN's #1 end market," indicating suppliers are realigning priorities.1
Multi-year contracts exceeding $20 billion were previously rare in the data center sector. Cloud providers typically signed shorter-term leases with flexibility to scale. The new deal structures suggest hyperscalers view AI compute as a long-term competitive requirement rather than a cyclical investment.
Traditional data center REITs have not announced comparable AI-specific deals. The divergence suggests the market is bifurcating between general-purpose facilities and purpose-built AI infrastructure. Developers with access to large power allocations and experience managing high-density deployments are positioned to capture capital flows from hyperscalers racing to build AI capacity.
