Natural Gas Is Plentiful. Getting It Into AI Power Is the Hard Part

The article highlights growing demand for gas-fired and onsite power to support AI and data centers, alongside rising electricity prices: the EIA forecasts 2026 wholesale prices up 11% across tracked hubs and 41% in PJM year over year. Baker Hughes announced turbine orders supporting about 1.1 GW for advanced computing, while Kinder Morgan reported record third-quarter net income of $764 million and a $14.4 billion project backlog. Bloom Energy reported third-quarter revenue up 138.1% to $1.236 billion and raised 2026 revenue guidance to $4.4 billion–$4.6 billion; EQT reported $1.083 billion in operating cash flow and highlighted its announced acquisition of Range Resources.

Analysis

The attractive exposure is to the bottleneck, not the fuel: ample gas and a softer price outlook may limit upstream pricing power even as constrained power access supports spending on pipes, turbines and onsite generation. That creates a potential relative-value split between contracted infrastructure and gas-price-sensitive producers. Kinder Morgan’s contracted project economics appear more insulated from Henry Hub than EQT’s, though project returns, construction costs and customer concentration matter more than headline backlog. EQT’s planned combination with Range Resources could improve scale and supply optionality, but could also add integration and execution risk; the article provides no deal terms to assess those trade-offs.

Baker Hughes’ awards are evidence of demand, not yet proof of durable earnings: delivery schedules, conversion to recognized revenue and customer concentration are key. Bloom Energy may capture demand where grid connections lag, but its growth claims need validation against cash generation, capacity and unit economics; high growth alone does not establish attractive valuation. NERC’s standards could raise interconnection and compliance costs or slow some projects, reinforcing the case for onsite solutions while potentially delaying load additions. FERC’s draft is only a proposal, not a permit or approval shortcut yet.

Over the next 1–3 months, watch project awards, backlog conversion, deal terms and regulatory filings. Over 6–18 months, the thesis depends on power-load growth becoming funded, connected projects rather than announced demand. The contrarian risk is that market enthusiasm prices turbine and onsite-power orders as immediate earnings while overlooking lead times, grid rules and capital intensity.

Key Decisions for Investors

• Consider a measured long KMI / short EQT relative-value position after checking valuation and post-earnings price action: the hypothesis is that contracted transport demand is less exposed to falling gas prices than producer realizations. Falsify if KMI project returns or schedules deteriorate, or EQT’s deal terms and realized pricing materially improve its relative outlook.
• Treat BKR’s data-center orders as a backlog-conversion watch item, not an earnings upgrade by themselves. Verify delivery timing, order cancellation terms and segment margins in subsequent filings; reassess if orders fail to translate into revenue or guidance.
• Keep BE on a watchlist rather than chase reported growth without valuation and cash-flow evidence. A more constructive entry requires confirmation of cash generation, production capacity and repeat customer demand; downside risk rises if guidance slips or growth consumes materially more capital than expected.
• Monitor the November 16 FERC comment deadline and subsequent policy action, alongside NERC/FERC implementation details. Delays or stricter requirements would weaken near-term pipeline and data-center buildout assumptions; faster permitting could strengthen KMI’s project pipeline, subject to project-level approvals.