US Data Centers Set to Burn More Natural Gas Than Most Nations

BloombergNEF forecasts that US data-center electricity generation will require an additional 15 billion cubic feet per day of natural gas by 2035, more than double its prior December estimate of 6.9 Bcf/d. The projected incremental demand exceeds current gas consumption in every country other than China, Russia, Iran and the US, underscoring a potentially material long-term demand catalyst for US natural gas and related power infrastructure.

Analysis

The investable implication is not simply higher Henry Hub pricing: incremental data-center load will be concentrated in constrained power regions, increasing gas-fired utilization, pipeline throughput and local basis volatility. WMB, KMI and ET should monetize through contracted takeaway and lateral expansions with lower commodity sensitivity than EQT, AR or RRC; the producers need sustained price realization and drilling discipline to capture the upside. The less appreciated bottleneck is compression and balancing infrastructure, making NGS a high-beta beneficiary if fleet utilization and day rates tighten, albeit with microcap liquidity risk.

Over the next 1-3 months, the cleaner catalyst is utility load-forecast revisions, signed generation or pipeline capacity agreements, and PJM/ERCOT capacity-price evidence rather than another long-term demand forecast. Merchant generators CEG, VST and TLN have the greatest near-term earnings torque where they can contract firm power to hyperscalers, but their valuations already embed meaningful load-growth optionality; avoid treating all power names as equally exposed. A regional gas shortage can also create a negative second-order effect for data-center economics, delaying construction or shifting projects toward nuclear-backed, renewable-plus-storage, or behind-the-meter generation solutions.

The consensus likely overweights gas producers and underweights midstream reliability assets. A higher national demand outlook does not guarantee materially higher Henry Hub prices if associated gas growth, LNG-project slippage, or weak industrial demand absorbs the volume; meanwhile, contracted pipeline and compression assets can benefit from physical flows even in a subdued-price environment. The thesis is falsified by repeated hyperscaler capex deferrals, utility load forecasts rolling over, or capacity additions that keep PJM/ERCOT reserve margins and gas basis spreads contained through 2027.

Key Decisions for Investors

• Build a 6-12 month long WMB / short EQT pair: favor contracted volume growth and lower balance-sheet sensitivity over unhedged Appalachian gas exposure. Reassess if Appalachian basis does not tighten or WMB fails to disclose incremental data-center-linked capacity contracts by the next two earnings cycles.
• Place NGS on an accumulation watchlist rather than taking a full-size position immediately; initiate only after evidence of improving compression utilization, pricing, or backlog in the next 1-2 quarterly reports. Size at microcap/liquidity limits; thesis fails if utilization remains flat despite customer demand commentary.
• Maintain selective, not broad, exposure to CEG or VST over 3-6 months only where new long-dated hyperscaler power contracts are disclosed. Take profits if contract pricing is below market expectations or if forward power curves and capacity auction outcomes fail to validate incremental scarcity.
• Use PJM capacity-auction results, ERCOT load updates, and regional gas-basis spreads as confirmation triggers before adding energy beta. A broad long-gas-producer position is not warranted without evidence that physical demand is converting into higher realized pricing rather than merely higher midstream volumes.