Natural Gas and the LNG Export

The question

Is US natural gas priced as a shoulder-season commodity with a structural glut from Appalachian productivity, or as the feedstock for the largest coincident LNG export capacity build in the country's history meeting a domestic demand step-change from gas-fired generation serving AI data-center load?

Base rates

We start by naming the setup. This cohort has underperformed SPY over the trailing twelve months -- the index compounded near 28.6% while the gas-weighted basket was mixed, with several E&Ps in flat-to-negative territory. The screener dropped the one-year performance hurdle for this idea deliberately. The underperformance is the imputed-expectations gap: the market is pricing a continuation of the 2023-2024 Henry Hub malaise into a forward period in which the demand curve inflects through three independent channels.

The reference class is prior coincident supply-demand inflections in North American gas: the 1999-2001 power-generation build that pulled Henry Hub from $2 to $10, and the 2016-2019 first wave of US LNG exports that absorbed roughly 9 Bcf/d of domestic supply. Tolling operators with contracted capacity compounded revenue 12-20% annually for four-to-six years at utility-like multiples, while the upstream cohort traded in a wider band tied to strip.

The second-wave US LNG capacity coming online between 2025 and 2027 is roughly 10-12 Bcf/d of incremental feedgas demand -- Plaquemines Phases 1 and 2, Corpus Christi Stage 3, Rio Grande Phase 1, Port Arthur Phase 1, and Golden Pass. That is a 10-12% increase in total US dry gas demand on a single demand channel. The imputed price embedded in forward consensus is roughly $3.25-3.75/MMBtu Henry Hub through 2027. The incentive price to bring marginal Haynesville and Appalachian rigs back to sustaining activity is closer to $3.75-4.25. The forward curve does not clear the incentive curve at projected demand.

Why consensus is wrong

Consensus treats LNG as a feedgas tax on domestic price. In a contracted-tolling model -- which is what Cheniere runs -- the tolling fee and the liquefaction spread are decoupled from Henry Hub; LNG's cash flows compound regardless of where the front month prints. The market continues to price LNG Inc. closer to an upstream beta than a contracted-midstream quality name, and the multiple gap has widened rather than closed.

The second miss is data-center gas generation. The AI load-growth narrative has attached to nuclear and renewables in the sell-side; the near-term incremental generation that actually serves hyperscaler load is a gas-fired combined-cycle or peaker, because the interconnect queue for firm baseload is three-to-five years longer than the commissioning timeline of a gas unit. The utility names tied to gas-weighted generation -- Ameren in Missouri-Illinois and OGE in Oklahoma -- are pricing a flat load-growth scenario while their integrated resource plans already reflect double-digit incremental MW tied to data-center commitments.

Third, the Appalachian productivity narrative has already monetized. Takeaway capacity is constrained, the rig count is disciplined, and the marginal producer is no longer willing to hedge below the forward strip. When the binding constraint shifts from takeaway to well-level deliverability, the economics of the E&P improve because basis differentials tighten, not loosen.

Position construction

The book has one 20% anchor and four sub-books.

LNG tolling anchor (~20%). LNG at 20% is Cheniere -- the contracted tolling book at Sabine Pass and Corpus Christi, with Stage 3 trains commissioning through 2026 and Corpus Christi Mid-Scale trains behind them. Of every contracted name in the energy complex, this is the one whose cash flows most closely resemble a regulated midstream and whose market pricing still embeds an upstream multiple.

Appalachian and Haynesville E&P (~46.9%). EQT at ~18.8% is the largest US dry-gas producer with the deepest Appalachian inventory and the cleanest LNG-linked export exposure through a Southeast takeaway book. EXE at ~13.4% is the combined Chesapeake-Southwestern entity with the Haynesville footprint closest to Gulf Coast liquefaction. CTRA at ~9.7% is the Permian-Marcellus balance with a gas-weighted free-cash posture. AR at ~5.3% and RRC at ~4.8% are the Appalachian liquids-rich complement; AR's integrated midstream and C3+ exposure adds a price-flex lever that pure-dry-gas names lack.

Utility and data-center-linked generation (~21.1%). AEE at ~15.9% is the Missouri-Illinois integrated utility with the clearest data-center tariff filings and a gas-weighted generation stack. OGE at ~5.2% is the Oklahoma utility with growing data-center load and a rate base that compounds with gas-peaker additions.

LNG shipping and compression (~6.9%). GLNG at ~2.2% and FLNG at ~0.8% are the LNG shipping and FSRU specialists -- sized as optionality on fleet day rates. AROC at ~2.4% and KGS at ~1.7% are the gas-compression sub-tier; their revenue scales with throughput rather than price and they compound with every incremental Bcf/d of takeaway.

Asymmetric payoff

If the 2025-2027 LNG capacity wave commissions within six months of guided dates, data-center gas-peaker load adds 3-5 GW of incremental combined-cycle dispatch by late 2027, and Appalachian takeaway stays constrained, the weighted book returns roughly 18-28% annualized over three years. If Henry Hub prints below $2.75 for two consecutive summers, the book returns roughly -8% to -15%. If a second wave of data-center commitments moves a major utility's integrated resource plan to a gas-weighted stack, the right tail is 35-50% with multiple re-rating on the utility and LNG sleeves.

At 50% base, 30% bear, and 20% bull, expected value is roughly +10 to +16% annualized against an SPY base rate near +8%. The payoff is asymmetric because the tolling anchor and the utility sleeve truncate the bear case while the E&P cohort captures the strip upside if the demand math clears.

Three things that would change our mind

1. Two or more of the 2025-2027 LNG projects slipping commissioning by more than nine months, with supplier-side causes suggesting the wave extends rather than lands, which would push the incremental feedgas call out past the window in which the E&P cohort can sustain current free-cash posture.
2. FERC or DOE imposing a second pause on non-FTA LNG export authorizations with language indicating a durable policy shift rather than a temporary study, which would remove the forward contracting pipeline behind Port Arthur Phase 2 and CP2.
3. Data-center load forecasts from PJM, MISO, and SPP revised downward by 30% or more in aggregate, indicating the hyperscaler demand curve is softer than utility integrated resource plans currently embed.

What we are explicitly NOT betting on

We are not betting on a specific Henry Hub price print. We are not betting on any single E&P maintaining rig cadence at current levels -- the 18.8%/13.4%/9.7% sizing of EQT/EXE/CTRA is deliberate to avoid single-operator execution risk. We are not betting on oil prices; the liquids-rich names are sized such that a $60 crude print does not re-rate the basket. We are not betting on renewables displacement being slower or faster than current trajectories. We are not betting on a specific LNG project sponsor winning FID on an expansion train. The thesis requires only that the contracted LNG tolling backlog compounds, that data-center load grows into a gas-weighted stack, and that takeaway stays a binding constraint on Appalachian basis. All three are strictly weaker claims than picking a Henry Hub target.