The Looming Natural Gas Crisis
The United States is not running out of gas in the ground. It is running out of spare molecules above ground — just as LNG contracts and AI data centers start pulling on the same system. By 2028–29, storage could print below any historical precedent.
00:30
The coming deficitAbundant today. Historically tight by 2028–29.
Despite currently ample supplies, U.S. natural gas storage is projected to drop below any historical precedent in 2028 and 2029. That is the ignition point: fierce competition for physical gas, violent price volatility, and a bill that shows up in household power rates because gas sets the marginal cost of so much American electricity.
- Natural gas now fuels more than 40% of U.S. electricity generation.
- Exports grew from essentially nothing to 15 BCF/day, and are scheduled to hit 35 BCF/day by 2030 — demand that is already contracted via long-term LNG projects.
- AI data centers add a new, always-on claimant. Credible cases put incremental gas demand near 5 BCF/day by 2030; extreme cases can triple that.
Stacked against current U.S. production of ~110–112 BCF/day. Toggle the AI case.
The acreage can theoretically add ~20 BCF/day. LNG alone wants that entire increment. AI is additional. Processing plants, gathering lines, and interstate pipes have to exist in time — they do not, on the current clock.
03:00
How we got hereThe shale boom built an export machine that now competes with the grid
Since 2010, shale turned the United States from a net gas importer into a major global exporter. Marcellus, Utica, Haynesville, and the Permian made the molecules. Export terminals made the destination.
That success locked in a high base of export capacity. Those cargoes do not politely step aside when a Virginia data hall or a New England winter needs the same gas. Domestic power and AI compute are now in a queue with Europe and Asia.
Basins that built the surplus
Marcellus · Utica · Haynesville · Permian
What the surplus became
LNG terminals, long-term offtake, a 15 → 35 BCF/day export ramp
The new claimant
Always-on AI load that cannot be curtailed like a factory shift
05:45
Supply-side realityThe resource is underground. The crisis is above ground.
Is the U.S. running out of natural gas? No. There is still a substantial underground resource in Appalachia, Haynesville, and the Permian. The binding constraints are processing, gathering, and interstate pipe.
Industry well-level mapping can estimate producible gas from operator portfolios with unusual precision. Maximum deliverability from known acreage and existing infrastructure is large — but operators already face declining well productivity and the depletion of prime locations. They must drill more just to hold output flat, let alone grow it by 20 BCF/day.
The three above-ground chokepoints
Gas processing
Raw gas is not pipeline gas. Plants that strip impurities will not cover a 20 BCF/day growth target. New trains take two to three years.
Gathering systems
The unglamorous pipes from pad to plant. Without a burst of gathering investment, wellhead potential never becomes deliverable supply.
Interstate network
The last major project, Mountain Valley, crawled through politics and courts. Current rules make the next one just as slow — and winter balancing depends on it.
23:00
If nothing changesStorage breaks, power prices follow, exports get politically ugly
If no mitigation arrives in time:
- Storage prints below historical lows by 2029–30. The system holds about 4 TCF, swinging with the seasons. Once that cushion is gone, every demand surprise is a price event.
- Electricity prices rise sharply because gas is the marginal burner for so much of the fleet. Short disruptions have already printed $6–$10+ / MMBtu. A structural deficit can push prints toward $20 or more. Exact paths are uncertain; the convexity is not.
- Export cuts become a live option — and a diplomatic problem. Long-term LNG contracts and allied dependence on U.S. cargoes make a sudden halt anything but simple.
Near-term forwards around $3.50–$3.60 / MMBtu in 2026–27 sit next to an $8–$20+ tail after storage tightens. The curve is complacent. The physical system may not be.
25:30
Winners and losersA tight gas market redistributes a lot of money
Gas producers
Encap Energy — high-quality Haynesville inventory, the crown-jewel dry-gas basin, still marked as if surplus is permanent.
Range Resources — Appalachian growth torque if basin differentials heal with pipe.
Solar and utility-scale renewables
Assets with merchant or market-linked PPAs collect higher power prices without new capex. Residential solar-plus-storage is a household hedge against the 10 a.m.–6 p.m. spike window.
Nuclear
Large light-water plants, especially AP-1000s on the Vogtle template, are the only scalable firm substitute for the mid-2030s. Watch Westinghouse, BWXT, and Cameco / related nuclear-fuel names as the U.S. tries to fast-track steel in the ground.
U.S. consumers
Rising power bills first, especially households with no solar, no battery, and no time-of-use flexibility.
Distributed gas generation
Fuel cells and small turbines — Bloom Energy, Caterpillar’s Solar Turbines and peers — get squeezed as fuel soars and contracts tighten.
New gas-plant developers
A 2029–30 fuel-risk window can stall FID on combined-cycle builds that assumed cheap, infinite molecules.
Data centers and hyperscalers
Energy can climb from ~10% to 20–30% of operating cost by 2030. That forces slower campus growth, harder efficiency targets, and less casual reliance on behind-the-meter gas.
39:30
Could this be a false alarm?Five things that would defuse the spike
Softer AI watts
Slower compute buildout, or efficiency gains that crush power per token, takes the 5–15 BCF/day wedge off the table.
Pipes and plants on time
New processing and interstate capacity. Regulatory clocks make a late-2020s rescue unlikely unless permitting changes now.
Better storage tech
Post-lithium batteries that decouple evening peaks from gas peakers. Commercially uncertain at the scale required.
Permian associated gas
Oil-driven gas can fill gaps — but only if oil prices stay high enough to keep the rigs running.
Slowing LNG exports is the remaining lever. Contracts and allied politics make it a last resort, not a plan.
47:30
What the U.S. should doFour moves that still fit the clock
1. Accelerate nuclear
Fund and expedite 2–4 AP-1000-scale plants on existing designs and supply chains. Target steel-in-the-ground usefulness by 2033–34. That is the firm, large block the mid-2030s actually needs.
2. Flood solar + batteries
Residential arrays with storage are a consumer hedge. Utility-scale plus storage shaves the gas peak. This is the fastest electron that does not need a new interstate gas line.
3. Build the unsexy midstream now
Processing and gathering unlock near-term production. Streamlined interstate permitting is the difference between 2028 tightness and a manageable market.
4. Open a Canadian buffer
Canada has the resource and not enough southbound pipe. A dedicated 1–2 BCF/day line into U.S. power markets is a winter-balancing valve, not a miracle — and still worth building.
50:00
Beyond the U.S.Allies are on the other end of those cargoes
America is increasingly the world’s swing LNG supplier to Europe and Asia. If U.S. availability tightens, allies shop the residual market — including politically toxic alternatives. Energy security abroad is now downstream of Appalachian gathering lines and Louisiana processing trains.
That is why “just cut exports” is not a free domestic policy tool. It is a foreign-policy event.
53:00 — 55:50
TakeawaysThe clock, the curve, and who should move first
Energy-intensive businesses, hyperscalers, and utilities need physical supply reviews now — not in 2028. Companies tied to gas-fired generation need a plan for a multiple of today’s fuel price. Upstream valuations that assume perpetual surplus may be the wrong base case. Watch forwards from 2028 out. The front of the curve can stay sleepy until it cannot.
Key metrics
| Aspect | Current / forecast | Notes |
|---|---|---|
| U.S. gas production | ~110–112 BCF/day | Existing acreage can add ~20 BCF/day if midstream exists |
| LNG export capacity | 15 → 35 BCF/day by 2030 | Locked by contracts and projects already underway |
| Incremental AI demand | +5 BCF/day base, up to +15 extreme | Wide uncertainty; compute efficiency is the swing factor |
| Working gas storage | ~4 TCF | At risk of breaking historical lows by 2029–30 |
| Price path | $3.50–$3.60 (2026–27) → $8–$20+ tail | Convex upside, not a single number |
| Nuclear relief | 2–4 AP-1000s by 2033–34 | Only large firm option for the mid-2030s |
Practical moves
Policymakers
License nuclear faster. Permit midstream. Incentivize solar-plus-storage as a consumer shield, not a slogan.
Industry
Lock physical supply early. Build a cost-spike playbook before the 2028 print.
Households
Rooftop solar and a battery are an insurance policy against a gas-set power price.
Investors
Re-rate quality dry-gas inventory and nuclear supply chains. Fade the idea that $3 gas is a law of nature.
By mid-2028 the surplus story gets a live stress test. Consumers, exporters, and compute growth will be bidding for the same limited deliverability. The U.S. has navigated energy transitions before. This one rewards urgency on pipes, plants, and atoms — not another year of assuming the shale miracle is infinite above ground.