US Grid Constraints: Towards 40GW+ of Behind-The-Meter Datacenter by 2028?

Jeremie Eliahou Ontiveros · SemiAnalysis · June 25, 2026 at 19:48 · ⏱ 33 min read  | Read on Substack ↗
Summary
US grid capacity additions are structurally insufficient to meet surging AI datacenter demand, forcing a shift to behind-the-meter (BTM) solutions. By 2028, BTM is expected to power over half of new datacenter capacity, creating a massive market for BTM gas equipment providers like Bloom Energy.
  • •US datacenter gross power demand is forecast to rise from +21GW in 2026 to +84GW by 2030.
  • •Net-new ELCC capacity added to the US grid is only about 15GW/year, rising toward 20GW+ by decade-end, far below datacenter demand.
  • •Grid headroom (spare accredited capacity) is approaching zero and turns negative by 2027 across many ISOs.
  • •BTM will power well over half of new US datacenters by 2028+, with the TAM for DC BTM equipment crossing 50GW/year by 2029.
  • •Gas turbine and transformer lead times have extended to 3–4 years, pushing gas plant development to 4–6 years versus a historical ~24-month baseline.
  • •Renewables and storage add over 20GW nameplate each per year, but their ELCC contribution is minimal and declining, so they cannot fill the firm capacity gap.
  • •AI labs and hyperscalers are increasingly accepting lower uptime (e.g., Meta targets two nines, skipping backup generators), reducing the historical cost disadvantage of BTM.
  • •Hybrid BTM-grid structures are emerging in ERCOT under Batch Zero, including Withdrawal-Limited Private Use Networks (WLPUN) and Provisional Controllable Load Resources (PCLR).
Read time 33 min
Length 33,592 chars
Category finance
Ideas
Jeremie Eliahou Ontiveros Substack author, SemiAnalysis
The article explicitly calls Bloom Energy the 'biggest beneficiary' of the BTM trend (noting it was first called out in Dec 2024), and the analysis quantifies a massive TAM for BTM gas equipment (fuel
The article explicitly calls Bloom Energy the 'biggest beneficiary' of the BTM trend (noting it was first called out in Dec 2024), and the analysis quantifies a massive TAM for BTM gas equipment (fuel cells and RICE) driven by grid constraints and AI demand. Risk: Execution risk on manufacturing ramp; customer concentration; potential regulatory changes favoring grid over BTM.
BE
More from SemiAnalysis
China’s CXMT Is Set to Challenge DRAM Incumbents
Jun 23, 14:51
Stop Saying Half of 2026 US Datacenter Capacity Is Canceled
Jun 18, 15:54
RL Systems Mind the Gap: Matching Trainer and Generator Throughput
Jun 16, 17:32
Is SMIC N+3’s Metal Pitch Smaller than Intel 18A’s?
Jun 14, 19:14
Intel Should Raise Capital
Jun 11, 14:50

This newsletter, published June 25, 2026, features Jeremie Eliahou Ontiveros discussing BE. 1 trade idea extracted by AI with direction and confidence scoring.

Speakers: Jeremie Eliahou Ontiveros  · Tickers: BE