What is So Hard About Behind-The-Meter Power For Datacenters? Part 1
Key Points:
- Behind-the-meter (BTM) onsite gas generation has rapidly become the primary power solution for AI labs and hyperscalers facing grid constraints, with 75GW of firm orders tracked for BTM AI compute, including 20GW ordered in Q2 2026 alone.
- Major tech companies like Microsoft, Google, Anthropic, Meta, and OpenAI are deploying multi-gigawatt off-grid power plants using a mix of gas turbines, reciprocating engines, and fuel cells to ensure reliable, high-margin AI compute capacity, often paying a premium for speed and reliability.
- The BTM buildout faces significant challenges including complex permitting processes, securing firm natural gas supply and pipelines, equipment supply constraints, skilled labor shortages, and operational complexities of islanded power systems without grid backup.
- Permitting hurdles are particularly acute due to strict air quality regulations, with some projects pivoting to fuel cells to reduce emissions and accelerate approvals, while others face delays or redesigns, exemplified by Oracle’s Project Jupiter.
- Despite the challenges, BTM power solutions are increasingly competitive versus grid connections due to grid capacity limits and slow utility interconnections, driving innovation in equipment types, financing models like Energy-as-a-Service, and hybrid solutions combining gas generation with batteries and renewables.