By playing, you agree to uListen's Terms and Privacy Policy and YouTube's Terms of Service.
Invest Like the BestWhy Natural Gas Will Be AI’s Next Great Shortage
Summaries, chapters, transcripts, and FAQs on this page are made by uListen, not YouTube.
CHAPTERS
Jump to publisher timestampsChapter summaries
Forecast: US gas storage breaks historical lows by 2028–2030
Matt Smith opens with his central finding: the US is adequately supplied through 2026–2027, but storage begins falling sharply in 2028. He argues the tightening becomes historically extreme by 2029–2030, creating a high-convexity price outcome and pushing the biggest pain onto electricity bills.
- •Natural gas is described as well supplied in 2026–2027, then materially undersupplied starting mid-2028
- •Working gas storage is presented as the key nexus of supply and demand, and the projected draw becomes unprecedented
- •Natural gas prices are framed as having “unbounded and convex” upside risk in a shortage
- •Electricity prices are positioned as the main place households feel the shock
- •A “knife fight” for physical gas is forecast for 2028
What’s driving demand: LNG exports first, AI compute second
Asked whether data centers are the whole story, Matt says the system was set up for tightness before AI arrived. He points to the long build-out of US LNG exports as the primary incremental demand driver, with AI compute now adding to an already-committed path.
- •Shale-driven abundance enabled the US to move from importing gas to exporting it
- •US LNG export capacity is described as ~15 Bcf/day today, with plans up to ~35 Bcf/day by end of 2030
- •LNG project timelines, approvals, and financing create inertia that’s hard to reverse quickly
- •Electricity demand had been stagnant for years; AI compute is a newer incremental load
- •The combined effect of LNG growth plus compute demand tightens the system
Quantifying AI gas demand: base case vs. “extreme” scenarios
Matt explains how his team assigns probabilities to power projects intended to serve compute, then translates that into incremental gas demand. He estimates a credible base case of additional gas demand and notes it could be much higher if more proposals actually get built.
- •A probability-weighted approach (e.g., a “P50” set) is used to decide which projects are credible
- •The base case estimate is about 5 Bcf/day of incremental gas demand tied to AI compute
- •Multiple behind-the-meter and distributed generation options are discussed (combined cycle, turbines, fuel cells)
- •He highlights that proposals beyond the base case could more than double incremental demand
- •An “unmitigated” high-end case is described as roughly 12–15 Bcf/day by the early 2030s
Why “just stop exports” isn’t simple: contracts, financing, and geopolitics
Patrick proposes curtailing exports as an obvious fix; Matt argues it’s constrained by contract law, project financing, and international reliance. They frame LNG as both legally and strategically difficult to unwind, even if prices spike domestically.
- •Shutting off LNG exports is described as constrained by contract law and rules governing projects
- •Tens of billions in project financing and long-term obligations are said to sit behind LNG terminals
- •The US is framed as becoming a major share of global gas supply, increasing foreign dependence
- •Free trade agreement (FTA) context is clarified in conversation
- •Matt notes LNG curtailment is a lever, but not a frictionless one
Is there enough gas in the ground? Resource vs. deliverability
Patrick separates the ‘resource’ question from the ‘logistics’ question; Matt says the US has gas, but production is a flow problem constrained by decline rates and development pace. He describes modeling basin-by-basin potential and argues inventory depletion and economics complicate the outlook.
- •Major basins discussed include Appalachia (Marcellus/Utica), Haynesville, Permian, and Eagle Ford
- •Oil-directed basins are described as producing gas as a byproduct, tying gas volumes to oil activity
- •Matt explains mapping acreage and wells to estimate what’s left and what can be developed
- •He stresses decline rates and the need for continuous drilling to sustain flows
- •A modeled maximum deliverability range of roughly 128–132 Bcf/day is described under optimistic development assumptions
Why supply can’t catch up: processing, gathering, and interstate pipelines
Matt details the midstream bottlenecks that prevent fast supply growth, even if the resource exists. He emphasizes the time required to build processing and gathering systems, and especially the difficulty of permitting and constructing new interstate pipelines.
- •Processing constraints: gas must be conditioned to pipeline specs and stripped of liquids/impurities
- •Processing capacity is said to require multi-year build times and needs to be expanded quickly
- •Gathering (small-diameter pipes) must expand materially to connect new wells to systems
- •Interstate pipeline build-out is described as extremely slow, with Mountain Valley Pipeline cited as a rare recent example
- •Permitting and regulatory barriers are portrayed as central to the inability to respond quickly
The 2030 crisis pathway: complacent curves, then historic draws on storage
Matt describes a market that’s complacent because near-term gas looks abundant, discouraging investment. He argues that structural demand arrives in 2027–2028, forcing unprecedented storage draws and creating extreme price outcomes that spot LNG flexibility can’t fully fix.
- •Natural gas is described as adequately supplied today and through 2026–2027, reducing investment incentives
- •The forward curve is characterized as flat into 2028–2030 despite AI announcements
- •Storage dynamics are central: the model projects storage falling below historical evidence by 2029
- •Matt contrasts transitory spikes (weather events) with a structural shortage driven by demand vs. deliverability
- •He argues that reducing only spot LNG liftings won’t solve the deficit; contracted cargo curtailment would be required to matter
Winners: gas producers and “free fuel” power assets (utility solar)
Matt names specific upstream producers he believes would benefit from higher gas prices, then explains why solar assets can see windfall economics when gas sets the marginal power price. He also highlights residential solar as a potential beneficiary if household electricity costs rise.
- •Natural gas producers are framed as clear winners if prices spike; Expand Energy and Range are mentioned as examples
- •He argues solar can benefit because gas raises the marginal price of power while solar’s fuel cost is effectively unchanged
- •He points to utility-scale solar/battery interconnection trends as supportive for solar’s role
- •Specific examples of beneficiaries discussed include XPLR (XIFR) and Clearway Energy
- •Residential solar (especially paired with batteries) is described as becoming more economic as peak power prices rise
Nuclear and solar: why large-scale nuclear is framed as the only durable fix
Matt argues there is no single silver bullet, but positions large-scale nuclear as the long-term solution that can displace gas demand in the 2030s. He contrasts large reactors with SMRs, emphasizing timelines, scaling, and the importance of rebuilding industrial “muscle memory.”
- •Natural gas is framed as the only flex fuel available at scale in the near term, making the bridge difficult
- •Large-scale nuclear is presented as the only solution that “makes sense” for the 2030s in this framework
- •He discusses US experience with Vogtle 3 and 4 and the difficulty/cost overruns as recent memory
- •SMRs are described as not yet ready to scale to the required “tens of gigawatts” need
- •He expects nuclear needs to come online around 2033–2034 to address worsening deficits beyond 2030
Losers and second-order effects: consumers, turbine makers, and hyperscalers’ cost structure
Matt identifies US consumers as the primary losers through rising electricity bills and painful trade-offs between exports, compute, and affordability. He also warns that some distributed generation and turbine build-outs could be mistimed if gas becomes too expensive, and he explains how hyperscalers’ energy costs could rise as a share of total compute cost.
- •US consumers are described as the biggest losers due to higher electricity prices
- •A trade-off is posed between exporting LNG, powering AI compute, and keeping bills down
- •Manufacturers of gas turbines and distributed gen sets are discussed as potential losers if new gas generation becomes uneconomic later in the decade
- •Hyperscalers are discussed as potentially seeing energy move from ~10% of cost toward 20–30% if gas doubles or triples
- •Patrick and Matt compare the dynamic to shortages that move “slowly at first, then all at once,” using a memory-market analogy
Devil’s advocate: what could make the shortage thesis wrong?
Patrick asks what would need to happen for gas to stay cheap; Matt outlines the main pushbacks he hears and explains why he believes they’re already incorporated or still constrained. He highlights Permian deliverability limits and notes that only a step-change in storage technology (batteries) would materially alter the picture.
- •A common pushback is that the Permian can supply more gas; Matt says pipeline/regulatory timelines limit surprises
- •He says his base case already includes announced Permian pipeline expansions coming online through 2030
- •He argues more Permian gas often implies higher sustained oil-driven drilling, which creates its own consumer pressures
- •Behind-the-meter Permian generation is discussed as shifting gas usage rather than creating new deliverability
- •A disruptive improvement in battery technology is presented as a major potential offset, though not assumed
Solutions and global stakes: build nukes, expand residential solar, and tap Canada
Asked what he would do with authority to act, Matt prioritizes jumpstarting large-scale nuclear builds and reinvigorating residential solar adoption. He also discusses Canada’s resource potential and pipeline needs, then widens the lens to global reliance on US LNG exports and the geopolitical consequences of curtailment.
- •He would have the US government build 2–4 large nuclear reactors to de-risk the supply chain and catalyze more builds
- •He supports accelerating residential solar adoption and revisiting incentives, emphasizing exposure to daytime peak prices
- •Canada is discussed as seasonally helpful today; longer-term, Matt argues Canada has deep gas resources trapped behind limited pipe
- •He proposes building a 1–2 Bcf/day pipeline into the US Midwest and moving gas across major power markets
- •He warns global buyers depend on rising US LNG exports and that curtailments could force painful rebalancing for allies
The coming “gas knife fight”: why contracting and risk management will define 2028
Matt says he’s surprised how little the market has focused on contracting and supply certainty given what he sees coming in 2028. He closes by urging business leaders to understand their gas sourcing, lock up physical supply, and prepare for counterparty and regulatory risk if prices and scarcity spike.
- •He’s surprised investors and project sponsors haven’t questioned gas availability more directly
- •He notes limited liquidity in 2028 markets as a reason the forward curve hasn’t moved much yet
- •He expects urgency to arrive when utilities and others begin hedging or buying for 2028
- •He emphasizes counterparty risk in a volatile commodity environment and the need for supply certainty
- •He frames the core planning question as: what if gas isn’t ~$3.50 but $10 or more, and what if physical supply is uncertain?
