Energy Transfer Just Delayed a New Mexico Pipeline. Here's Why I'm Not Worried Yet.

Energy Transfer postponed its 18-mile Green Chile natural-gas pipeline in New Mexico from August 2026 to February 2027 following regulatory setbacks, land-rights disputes, and historical-site concerns. The project would supply up to 400 million cubic feet per day to Bloom Energy fuel cells supporting Oracle and OpenAI's Project Jupiter AI data-center campus, but represents only about 1.25% of Energy Transfer's more than 32 billion cubic feet per day of gas throughput. The article views the delay as immaterial to Energy Transfer's broader 140,000-mile pipeline network and long-term income-investment case.

Analysis

The relevant exposure is not ET's existing throughput but the incremental economics of a dedicated, high-utilization lateral tied to a concentrated AI-load customer. A six-month slip likely has immaterial consolidated EBITDA impact, but it defers a potentially premium-return contract and raises the probability that future data-center gas infrastructure faces longer permitting timelines, higher legal costs, and more onerous cultural/environmental mitigation. That is more consequential for ET's cost of growth capital and backlog valuation than for near-term distributable cash flow.

BE is the more operationally exposed listed name: delayed gas availability can shift Project Jupiter commissioning, module acceptance, and service-revenue timing, while leaving fixed project-development costs in place. ORCL's financial exposure is likely negligible relative to its broader data-center buildout, but the episode highlights a bottleneck the market may be underpricing: grid interconnection and gas-delivery infrastructure can constrain AI capacity monetization even when chips and servers are available. Beneficiaries of prolonged constraints include incumbent utilities and midstream systems already connected to load centers; developers dependent on greenfield laterals face a higher hurdle rate.

Near term, this is not a standalone ET trading catalyst; income-oriented holders should focus on whether management quantifies capital spend, contracted returns, and reimbursement protections at the next earnings call. Over 6-18 months, repeated permitting setbacks across AI-related gas projects could widen the valuation gap between diversified, existing-asset midstream operators and growth stories whose projected EBITDA depends on new-build execution. The thesis is falsified if ET discloses material customer termination rights, unrecovered development spend, or a broader revision to its growth-capex/EBITDA outlook.

Key Decisions for Investors

• Maintain, do not add aggressively, to ET on this item alone; treat any 3-5% event-driven weakness as an accumulation opportunity only after confirmation that project capex is protected by contract and 2027 in-service timing does not reduce consolidated EBITDA guidance. Downside risk is a permitting-driven rerating of the growth backlog rather than lost base earnings.
• Avoid using BE as a direct AI-data-center proxy until management quantifies Project Jupiter revenue, acceptance, and working-capital exposure. Set an alert for a cut to 2027 deployment or service-revenue guidance; that would support a tactical short or put spread, whereas confirmation of alternative gas supply would remove the thesis.
• Watch ORCL's next capex and cloud-capacity commentary for evidence that physical-power constraints, rather than GPU availability, are delaying customer deployment. A material increase in capex without corresponding cloud RPO/revenue conversion would favor a relative short ORCL versus MSFT rather than an outright position.
• For a 6-18 month infrastructure allocation, prefer diversified, contracted midstream exposure such as ET or KMI over single-project power-to-data-center developers; require evidence of existing interconnects and take-or-pay protections before assigning premium multiples to AI-power backlog.