Summary
Enki AI documents the 2025–2026 midstream-gas pivot: pipeline operators (Williams, Energy Transfer, Pembina, Kinder Morgan, Exxon) moving from commodity transport to dedicated on-site power providers for AI data centers. The article’s central frame is “speed-to-power” — behind-the-meter gas plants deploying in ~18 months versus multi-year grid interconnection queues. Williams Companies leads with $5.1B committed to a “power innovation” portfolio; Project Socrates ($1.6B) is the headline buildout. AI-driven natural gas consumption projected at up to 6 Bcf/d by 2030.
Key points
- 18-month behind-the-meter gas plant deployment vs. multi-year interconnection queue
- Williams Companies: $5.1B committed; $3.1B in two October 2025 projects; $1.6B “Project Socrates” (H2 2026 completion)
- Williams + Meta Platforms: dedicated Ohio facility partnership
- Energy Transfer: February 2025 deal supplying gas to Cloud Burst’s Texas data center
- Pembina Pipeline: near-agreement with Meta in Alberta
- Kinder Morgan: 7 landfill gas-to-energy facilities ($135M, October 2025)
- Exxon Mobil: 2.7+ GW data center power pipeline
- AI-driven gas consumption projection: up to 6 Bcf/d by 2030
Newsletter angles
- “Speed-to-power” is the cleanest framing of the off-grid pivot. The 18-month-vs-multi-year delta makes the operator decision obvious — hyperscalers exit the grid where they can. Newsletter angle: name speed-to-power as the operating constraint that has flipped the hyperscaler power strategy from “negotiate with utility” to “build a private grid.”
- Carbon lock-in tension. The article surfaces 15–20-year asset lifetimes as the ESG counter-weight. Newsletter angle: the AI buildout is locking in fossil-fuel infrastructure at a moment when most climate trajectories assume those assets are being retired. The contradiction is on the page in the SWOT.
- Midstream financialization of AI capex. Williams’ $5.1B “power innovation” portfolio is the pipeline industry pricing AI revenue. Newsletter angle: when pipeline companies start running “power innovation” portfolios, midstream gas has been re-priced as an AI-infrastructure asset class.
Entities mentioned
- Williams Companies (new)
- Energy Transfer (new)
- Pembina Pipeline (new)
- Kinder Morgan (new)
- Exxon Mobil
- Meta
- Cloud Burst — Texas data center operator (new)
- Erhan Eren — Enki AI CEO (article author, new)
Concepts mentioned
- Speed to Power — behind-the-meter deployment timeline as competitive advantage (new)
- Behind the Meter Generation (new)
- AI Buildout Grid Constraint
- Interconnection Queue
Notes
This article does NOT mention Oracle Project Jupiter, Bloom Energy, or Crusoe. The Oracle/Bloom microgrid story comes from a separate Data Center Knowledge piece (Oracle Project Jupiter Bloom Fuel Cells — DCK - 2026-04-29). The Day 15 plan’s first draft mis-attributed the Project Jupiter line to this article via a merged search snippet; the attribution was corrected during ingest.
Author bias note: Erhan Eren is the CEO of Enki AI, which sells software to data-center operators. The framing is sympathetic to the operator/buildout perspective; community-impact, carbon, and ratepayer-incidence angles are present but not centered.