#32. Correct and unfunded: American natural gas
August 24, 2026
What if you knew something was going to happen and were powerless to stop it? Not because nobody believed you. Everybody could come around. But you’d be powerless even if they did. Convince every investor, buyer, and skeptic today, and it still wouldn’t matter
In #30. Time v. Human Nature, I argued that delegated decisions break when holding a position is too painful, even if correct. In #31. Waiting in Public, I argued that only systems engineered to survive that regret pressure can hold a correct idea long enough to be paid for it.
I recently came across a live example of a bet testing both of these theses: American natural gas.
Natural gas
Natural gas is the most important source of energy in the United States. It supplied 41% of utility-scale electricity generation in 2025. Number two was *nuclear* at less than half that: 18%. And it’s not just great for us at home. Because we produce natural gas so cheaply, we are the world’s largest exporter of liquified natural gas (LNG; its transportable form), supplying roughly 30% of the global market. In terms of total energy, natural gas is set to overtake petroleum on an energy-equivalent consumption basis and, together with petroleum, services 3/4 of American energy consumption.
Natural gas is also the “marginal” source of power, meaning the cost of electricity is very closely tied to how cheap or expensive it is to produce electricity from natural gas. In wholesale markets, power prices are set by the marginal cost of the last required unit of power; the unit that perfectly balances supply and demand. Whatever that unit costs to produce is the price per unit for all power dispatched in that moment. This is called merit-order dispatch. Natural gas is called upon when cheaper sources can’t fully supply demand, which they typically can’t. So natural gas ends up being the “marginal unit” that perfectly balances supply and demand.
There is one path to produce natural gas and two paths to consume it. All three are set by construction schedules rather than price. Natural gas is produced from wells, some drilled for gas itself in Appalachia and Louisiana, some producing it as a byproduct of oil drilling in West Texas. Then there are the systems to collect gas from each wellhead, processing plants, and pipelines to carry it wherever it needs to go. Years of permitting and construction before any of it reaches a market. The gas produced is spoken for well before it’s consumed. The consumption paths are gas-fired power plants or LNG exports. All those facilities that are built or under construction are financed by customers. Foreign buyers signed long-term offtake agreements that funded the export plants. Utilities and hyperscalers underwrote the generation that feeds their grids and data centers. Those contracts are legal claims on a molecule that has been paid for, not mere forecasts.
Common knowledge dictates that we have so much natural gas under our feet in the US that power prices can remain relatively stable in the intermediate term, as natural gas will continue to be that marginal unit.
One investor says otherwise. One investor says US natural gas capacity is woefully under-built for that already-contracted demand, let alone new demand from more data centers. So much so that we will have an energy price crisis in the US this decade. And importantly, the global financial market - including the US market, the deepest and most sophisticated financial market - is powerless to stop it.
Nat gas, meet Matt gas
Matthew Smith is founder and CIO of Chronometer Partners, a long/short hedge fund. Matt and his team built a bottom-up model over fifteen months, modeling essentially every relevant asset in the natural gas system: active wells, undeveloped acreage, gathering and processing facilities, pipelines, storage, LNG trains, contracted exports, gas generation, data center load. His conclusion is that natural gas production tops out lower and sooner than most assume, primarily because the wells to reach more gas are deeper than most assume and the infrastructure to move any of it doesn’t exist. Flatly, the fuel that everyone thinks is abundant is just not. He recently published a letter and then spoke publicly on Invest Like the Best arguing that the United States begins depleting natural gas storage in an unprecedented way in 2028, and approaches exhaustion of working storage by 2030.
So when those legally binding claims come due on a much lower supply ceiling, somebody is going to feel pain. More production? Sure, but production is beholden to construction schedules that take nearly a decade. Money can’t solve that. So in the meantime, the US, he says, is hurtling toward the decision point of whether to break commitments to allied buyers who built their energy security around American LNG exports, significantly slow down the compute buildout that priced its power off a fuel that no longer costs what it did, or force American households who are subject to the marginal source of power to pay electricity bills an order of magnitude larger.
Much of the energy industry thinks Matthew is wrong. Not early. Wrong. And this isn’t generalist consensus waving off a specialist. Among detractors are private equity investors and commodities traders who themselves spent their first careers as drilling engineers. The exact people who make you feel uncomfortable in a stance. Specialists vehemently waving off a specialist, in volume, with money behind their view.
Chronometer is waiting in public
I wrote in #31. Waiting in Public that delegated decision-makers face consequences long before their investments resolve, so they optimize for looking right before they can be fired rather than just being right. I titled it Waiting in Public because that’s what you have to do if you’re going to make a long-term unpopular bet. You just have to make the bet, wait, and take whatever the market throws at you in the interim.
Matthew and Chronometer are quite literally waiting in public. He made a call that puts his firm visibly outside consensus for at least two years on a thesis nobody can adjudicate until the back half of the decade. That is the trade that almost nobody can hold. And, to his credit, he’s not hiding from it. He’s open about his call. He went on the podcast. He updates his letter to respond directly to criticism. It’s honestly amazing.
So private capital *can* underwrite ahead of the signal. He lists several companies he’s now bullish on and presumably holds as a result. But then why not just build production capacity and own that? Matthew names midstream as a winner. It’s because even he sees that kind of direct expression of his thesis as not being worth the wait. Public equity is liquid and reversible. Construction costs of putting steel in the ground to build capacity are not. The most convicted, best-resourced, most explicitly early actor in this market solved his own waiting-in-public problem by routing *around* the asset that actually closes the gap because, even to him, holding five years of construction risk against a flat curve is not survivable. This gap exists because the market doesn’t pay participants to close it. And so Matthew has structured his investment for better survivability.
And while this is a gas problem, it rhymes with a lot of potential gaps in infrastructure funding. Grid transmission has the same structure. So do nuclear enrichment and conversion capacity, and parts of critical-minerals processing. Those gaps intensify as our world becomes more defined by shocks rather than cycles (see: #27. Collective Illusion and Resilience). You can build ahead of cycles, not shocks. And the general condition in which this type of air pocket can open is when a physical build takes longer than the market can price against, when we experience a shock nobody saw coming or, as with gas, due to a disputed outcome nobody’s willing to pay to insure against until we’re past the window to build that insurance.
So what if he’s wrong? Honestly no big deal. He loses out on his Soros-esque call. Either because we have plenty of produceable natural gas or because resources showed up: prices induced development, midstream got built on the ordinary schedule, demand moderated. Chronometer bears the cost, Matthew’s letter ages poorly.
But what if he’s right? If Matthew is right, the market watches in disbelief as a deficit assembles out of already-financed LNG trains and already-contracted generation. A deficit nobody could build against because no actor had both the balance sheet and the horizon to put steel in the ground in time. And somebody, likely the federal government, has to choose between slowing AI compute (now tied heavily to geopolitical strength), sending allies into their own energy crisis, and making residential electricity prices largely unaffordable. Three commitments, one molecule, major domestic and international repercussions.
From a national interest perspective, the country’s downside needs to be bounded. But how? And by whom?
We have already run this experiment
Matthew’s analogy is the post-COVID chip shortage. It was slow, and then it happened all at once.
The dependency was visible for years. The just-in-time system was intentional; it requires lower overhead and working capital, and it delivers orders when they’re needed. As demand grew, so did greenfield fab developments. But they take 3-5 years to build, and when supply suddenly starts falling short of demand, you get shortages and price spikes (see: #14. Microchips: India’s entrance). That risk was always inherent in the model, but nobody built ahead of the chip shortage because there simply hadn’t been a chip shortage. It was a risk that was not priced into the forward market and so did not pay to get ahead of.
And then we were behind it. Lead times for chips grew and they quickly became unattainable. I remember calling anybody and everybody who claimed to have a handle or oversupply in India, China, Hong Kong, and Taiwan to get any chips for my company. It felt like sand falling through our fingers. In a matter of months, the entire economy discovered it was short a component that costs a dollar and gates a vehicle worth forty thousand and meters the speed of companies that have taken on millions in venture funding.
Remember this?
How did that resolve? Public capital. Tens of billions of dollars of it. On time? No. Well-after the shortage, with significant bullwhip effect across the supply chain. We are *still* building those fabs. The federal government was forced into a position to underwrite the asset nobody could build early at a moment when it had no leverage over cost, timing, or terms, and years after its money could have prevented anything. Being late cost more.
That is what this natural gas kind of risk looks like to me. Gradually, then suddenly, then incredibly costly.
What to do?
One obvious answer is grants and subsidies to make economic the production and storage that are today uneconomic. If something important isn’t getting built, make it cheaper to build. Tax credits, loan guarantees, grants, price floors. Congress has all of these tools and uses them constantly. Often, they work exactly as advertised. Investment and production tax credits built American solar and wind. Advance purchase orders delivered a vaccine in nine months. Price guarantees financed Hinkley Point and most of European offshore wind.
But there’s the part we just watched happen with chips. A subsidy is legislation, and legislation needs a coalition, and a coalition needs the problem to be obvious. Nobody sponsors a bill to build ahead of a shortage that most of the industry says will never happen. Congress as a body is simply not good at taking on that calculated risk. There’s no constituency for it and real humiliation if you’re wrong. Congress is very good at ratifying a consensus. It cannot be early. That’s not really even a Congress thing. Being early in public is something no delegated body can do, which is the argument I’ve been making for four essays about investment firms and is apparently just as true one level up.
On the commercial side, nobody will commit to buying a pipeline’s capacity if they don’t think it’s needed. Remember, they have signed contracts for the capacity they need.
So money exists. Instruments exist. But still inaction persists because the cushion for a non-consensus view requires committing capital on a further time horizon than private markets are willing to hold.
Enter Scott Bessent
Bessent was charged early on in the administration with designing a sovereign wealth fund. The usual case for a sovereign wealth fund is that private markets won’t fund strategically important sectors, so the government must. That case is mostly wrong. Private capital funds quantum, biotech, defense tech, critical minerals, fission and fusion. It financed the largest private infrastructure buildout in modern history in data centers. Although fun to play in, there’s no meaningful list of sectors that capital ignores when there’s a strong enough buyer signal.
The case for sovereign capital here is narrower and stranger. Capital exists, is informed, and is arguably correct. There is not a strong buyer signal for more storage or supply because customers all hold contracts that say they’re fine. And private capital has very little incentive to provide this insurance because, if they’re right, they make no money. The value of allies who don’t get squeezed, compute that stays affordable, electricity bills that don’t become crippling accrues to the nation in the form of stable, reliable electricity. The NPV of suppliable power that may not be needed 5 years out is not good. The country is the party who eats the consequence of the shortage.
So what would sovereign capital commit to? Purchasing agreements. Those agreements would be for
Pipeline capacity, including the Canada-to-U.S. line Matthew proposes.
Storage capacity, which is the cheapest insurance in the whole system and the most chronically underbuilt.
Alternative generation and storage - Matthew makes the case for solar and nuclear (fission, not fusion). The former in combination with battery storage. The latter being his preference for immediate action given its longer timeline and significant, consistent generation capability.
None of this involves building and owning anything. Helpfully, that means a sovereign fund doesn’t need to be as big. The capital is to fund obligations, not investments. It’s signing purchase agreements early and holding the obligation. When demand arrives, sell that capacity to the private buyers who now want it and realize modest proceeds. If demand never materializes, there’s no one to sell the capacity to, and the government pays out the contract. That’s the bounded loss of an insurance premium. Or, in the frame of one sovereign fund advocate, Sadek Wahba, the cost of maintaining national infrastructure as if it were part of a national balance sheet.
What Wahba says
Dr. Sadek Wahba has been making the institutional aspect of this argument via NYU’s Wahba Initiative for Strategic Competition and most recently in his American Affairs piece (see: The Missing Institution). His framing is that infrastructure spending is a balance sheet issue, not a budget issue. Instead of funding once and walking away until it breaks and you’re forced to revisit, you maintain infrastructure against a system that keeps growing around it.
In our natural gas context, it’s the gas system that needs its buffer maintained in proportion to everything hanging off it. And nobody has been doing that.
Working natural gas storage capacity grew 7% between 2010 and 2025. Over the same period demand grew *54%*, and Matthew has it growing *another* 39% by the end of the decade. Our days of cover went from 72 days in 2010 to 50 last year, heading toward 37 by 2030. That is a maintenance failure. By letting demand outpace storage, we are creating the same just-in-time system for natural gas that we had for chips. The system we all found was too fragile and are now feverishly borrowing from our future selves to fix now. What could go wrong?
Proper guardrails
If a sovereign fund were to exist, it would be at an incredibly high risk of turning our government, with a congress that loves to trade stocks based on “public” information, into a corrupt institution. Guardrails are the most important aspect of standing one up. In this case, a sovereign fund should not pick winners, and it should not be taking a view on the price of gas. A government buying producers because it expects a shortage is a government trading its own market. We will be terrible at it and eventually be corrupt about it.
Wahba’s piece lays out concrete steps that you’d want to have at many investment institutions. Independent board, staggered terms, delegated authority to a threshold. The design is meant to prevent partisan creep into a vehicle whose decisions should outlast any administration.
What this would end up looking like in the immediate term is not like many of the sovereign funds we know. It wouldn’t be Temasek anchoring venture funds, GIC investing in hedge funds, or Norges owning1.53% of every public company. The US doesn’t need to make returns abroad to bring them back domestically. The returns are domestic and they are taxed here. What it would look like is narrow: committing capital where the time to build something lies beyond what the market can see or has priced. Outside of that, the market should do what the market does.
With natural gas, every actor in this story is rational. Producers, hedge funds, utilities, hyperscalers, customers. All of them are behaving correctly inside their own constraints. Even Matthew. He is very clearly betting that he’s right and openly debating those who disagree. But he is also very vocal that he hopes he’s wrong or that his argument somehow mobilizes somebody to do something; although it’s not clear to him what that could possibly be outside of government intervention. None of these rational actors get paid to be early or to buy more than they need. The country is the only singular party that will feel the pain of him being right and has an impetus to act. In this story, somebody has to be willing to look wrong for a decade. Chronometer is willing to look wrong for a decade in order to get paid if it happens. Somebody has to be willing to look wrong for a decade to make sure it doesn’t.
