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

Episode Summary

Executive Summary: Matthew Smith argues the U.S. is heading toward a severe natural gas shortage by 2028-2030 as LNG exports and AI data center demand collide with limited growth in supply, infrastructure, and storage. He says prices could become “unbounded and convex,” with electricity costs, industrial investment, and U.S. consumers bearing the brunt unless gas infrastructure, large-scale nuclear, and distributed solar are accelerated.

Main Topics: Historic natural gas supply deficit (Priority: 5/5): Smith’s central thesis is that U.S. gas demand will outstrip deliverable supply within the next few years, creating a structural shortage rather than a temporary price spike. LNG exports as the original demand driver (Priority: 5/5): The discussion emphasizes that LNG export growth was the first major structural demand shock and is already largely locked in through long-term project finance, contracts, and permitting. AI/data centers as the new incremental load (Priority: 5/5): AI compute is described as the newest and potentially decisive demand source, adding several BCF/day of gas demand through power generation for data centers. Infrastructure bottlenecks: processing, gathering, pipelines (Priority: 4/5): Even where gas exists in the ground, Smith argues processing plants, gathering systems, and interstate pipelines will constrain how much can reach market. Market consequences: storage depletion and price convexity (Priority: 5/5): He predicts that by 2028-2030 the system will draw down working gas storage in an unprecedented way, causing electricity and gas prices to rise sharply and unpredictably. Winners and losers across energy sectors (Priority: 4/5): Likely beneficiaries include top gas producers, utility-scale solar, residential solar, and large-scale nuclear; losers include consumers, some gas turbine manufacturers, and gas-dependent generators. Policy and investment response (Priority: 4/5): Smith argues the best long-term fix is to accelerate large-scale nuclear construction and expand solar adoption, while also improving gas transport and acknowledging counterparty risk in power contracting.

Key Arguments: LNG exports are already a large and durable draw on U.S. gas supply, with growth from about 15 BCF/day to 35 BCF/day by 2030, so they cannot easily be ignored or reversed. AI data center power demand is not the only issue, but it pushes an already tightening system over the edge in the late 2020s. The U.S. can likely produce more gas, but deliverability is limited by processing, gathering, and pipeline constraints rather than geology alone. By Smith’s base case, credible AI-related gas demand is about 5 BCF/day, but extreme cases could reach 12-15 BCF/day by the early 2030s. Working gas storage is the key buffer in the system, and he believes that buffer will be materially depleted starting in 2028. Price discovery has been muted because the forward curve remains complacent and current supply looks adequate through 2026-2027. If gas prices rise enough, LNG cargoes and even contracted exports may be pressured, but cutting exports is legally, politically, and commercially difficult. Large-scale nuclear, especially AP1000-style reactors, is presented as the only durable long-term solution that can arrive in time to matter in the 2030s. Residential solar plus batteries is framed as the most practical near-term hedge for households against peak power prices. Companies and hyperscalers need to secure physical gas supply and counterparties now, because future availability may become a competitive bottleneck.

Data Points: Current U.S. natural gas production: 110-112 BCF/day - Baseline daily U.S. supply level discussed as the denominator for export growth and future constraints. Current LNG export capacity: about 15 BCF/day - Approximate current nameplate U.S. LNG export capacity. Planned LNG export capacity by 2030: up to 35 BCF/day - Scheduled nameplate export capacity the U.S. may reach by the end of 2030. Share of current production exported: 12-15% - Current exports as a share of daily U.S. gas supply. Incremental gas production capacity: about 20 BCF/day - Smith’s estimate of how much additional U.S. gas production could be added from existing acreage and known basins. Base-case AI-related gas demand: about 5 BCF/day - Credible incremental demand from AI compute in Smith’s P50 case. Extreme AI-related gas demand: 12-15 BCF/day - Potential demand by early 2030s if lower-probability projects are realized. Natural gas used in power generation: over 40% of U.S. power generation - Smith says gas is already the dominant power fuel and is becoming even more important. U.S. working gas storage: about 4 TCF - Total storage buffer that Smith expects to be drawn down sharply in the late 2020s. Historical market price reference: $8-$10 per MCF - Referenced as a comparison from prior shortage episodes like Russia-Ukraine and weather-driven spikes. Current gas price level cited: about $3.50-$3.60 - Forward curve / market pricing discussed for 2026-2027. Fuel cell gas consumption: 150 million cubic feet/day per gigawatt - Bloom Energy fuel-cell consumption cited as a benchmark for distributed generation demand. Nuclear build timeline: 2033-2034 - Earliest feasible timing Smith gives for large-scale nuclear to solve the looming shortage. Canada export capability: 11-12 BCF/day seasonally - Canadian gas imports to the U.S., especially in winter, discussed as a partial relief valve. U.S. government nuclear funding window: $260 billion - Mentioned as available through the loan program office / EDF to seed nuclear buildout. Recent U.S. nuclear builds: 2 reactor units in 30 years - Vogtle 3 and 4 cited as the most recent U.S. large-scale nuclear experience. China nuclear buildout: 39+ reactors - Used to argue that large-scale nuclear is still buildable and scalable in practice. Solar adoption in new interconnects: 90% of interconnects with batteries - Smith notes solar-plus-battery has dominated new interconnection activity over the last decade.

Pivotal Quotes: "We are headed into a place where we see an historic deficit in natural gas supply available in the United States." — Matthew Smith: Opening thesis on the scale of the impending gas imbalance. "I think we will come to the conclusion that the upside risk, the price of natural gas, is both unbounded and convex." — Matthew Smith: Smith’s core view on pricing if shortages emerge in 2028-2030. "There’s no bridge fuel other than solar." — Matthew Smith: His argument that solar is the only near-term substitute while large-scale nuclear is built.

Implications: If Smith is right, energy prices, data center economics, and utility planning will be reshaped by a gas crunch in the late 2020s. Investors and operators should secure physical supply, accelerate firm power solutions, and prepare for a bigger role for nuclear and solar.