Matthew Smith — Natural Gas: The Next Bottleneck - [Invest Like the Best, EP.483]

Summary

Matthew Smith lays out a detailed thesis that the U.S. is moving toward an historic natural gas deficit, with the market underestimating how quickly AI data centers and LNG exports will tighten supply. He argues the problem is not that the U.S. is “running out” of gas in the ground, but that processing, gathering, pipelines, and export commitments limit how fast gas can reach market. In his model, supply is adequate through 2026-2027, but storage begins to get drawn down in 2028 and price risk becomes sharply convex into 2029-2030. He says the obvious policy fixes are constrained by contracts, finance, and geopolitics, so the durable solution is large-scale nuclear, while residential solar and batteries become attractive hedges for consumers.

Notes

• Smith says the U.S. is heading toward an historic natural-gas deficit, with natural gas becoming the country’s most important fuel for power.
• He frames the thesis as a timing problem, not a geological shortage.
• His base case says gas is adequately supplied through 2026-2027.
• He expects tightness to begin in 2028 as AI compute demand and LNG exports keep ramping.
• He says price risk becomes “unbounded and convex” as storage is drawn down.
• He thinks electricity prices will feel the stress most acutely in 2028-2030.
• He says the U.S. exports about 15 Bcf/day of LNG today on 110-112 Bcf/day of domestic production.
• He says exports already consume roughly 12-15% of daily U.S. gas supply.
• He says LNG export capacity is scheduled to reach 35 Bcf/day by end-2030.
• He estimates credible incremental AI-compute gas demand at about 5 Bcf/day in his P50 case.
• He notes that more aggressive AI demand scenarios could consume much more gas than the base case.
• He says export shutdown is not a simple fix because of contract law and project finance.
• He argues the U.S. cannot just turn off exports because allies and free-trade partners depend on American gas.
• He says the market is complacent because the forward curve remains flat despite AI buildout.
• He says the U.S. has gas in the ground, but deliverability is constrained by infrastructure.
• Processing is his first major bottleneck.
• He says the U.S. does not yet have enough processing to reach his assumed 20 Bcf/day production target.
• He says new processing plants take about 2-3 years to build.
• He says the U.S. has built only one major interstate gas pipeline in the last 10-12 years.
• He says his base case already includes more than 7 Bcf/day of pipelines under construction or development.
• He argues any truly new gas pipeline before 2030 would already be visible in regulation and permitting.
• He says Permian gas cannot solve the shortage without much higher oil prices.
• He says that would require higher oil for longer, which would worsen the consumer energy problem.
• He believes storage could fall below all historical experience by 2029.
• He says by 2030 storage could look very low if nothing changes.
• He compares shortage pricing to prior international crises where gas spiked to $8-$10/MMBtu.
• He says gas is currently around $3.50-$3.60 and the curve implies complacency.
• He sees gas producers with remaining core inventory as likely winners.
• He specifically highlights Expand Energy as the biggest winner in his framework.
• He thinks utility-scale solar benefits from higher power prices.
• He argues residential solar becomes economically attractive as a hedge against high peak electricity prices.
• He says batteries make the residential solar case stronger because they extend availability around the clock.
• He says large-scale nuclear is the only long-term solution that makes sense.
• He wants large-scale nuclear online by 2033 or 2034.
• He says SMRs are still mostly science experiments and may not scale fast enough.
• He says SMR vendors are not yet built to manufacture tens of gigawatts of capacity.
• He prefers AP1000-class reactors as the practical path.
• He says the U.S. government should build two to four AP1000s end-to-end to de-risk the supply chain.
• He argues first-of-a-kind nuclear projects would create a template for follow-on builds.
• He says the U.S. loan program could help finance more nuclear deployment.
• He says Canada can help seasonally, but not as a full year-round fix.
• He says the U.S. is not really configured to take Canadian gas year-round.
• He says U.S. exports have gone from nearly zero to about 15 Bcf/day and may reach 35 Bcf/day.
• He warns that countries around the world are now dependent on U.S. gas deliveries.
• He says utilities will eventually hedge or buy gas in 2028, triggering a rush for physical supply.
• He says gas buyers should lock up physical supply and counterparty protections early.
• He says the biggest loser is the U.S. consumer through higher electricity bills.