This interview warns that US natural gas faces a severe shortage starting in 2028, as AI data centers and LNG exports will outpace supply and pipeline capacity, potentially draining storage by 2030 and causing prices to soar. Expert Matthew Smith is bullish on Expand Energy (controls ~70% of core Haynesville wells, stock cheap after CEO departure) and Range Resources (has room to grow output). He flags risks for Bloom Energy (its generators may lack gas fuel) and Caterpillar's solar turbine unit (bad timing for expansion).
Matthew Smith, founder and CIO of Chronometer Partners, warned on the Invest Like the Best podcast that the U.S. natural gas market will face severe bottlenecks after 2028. His team spent 18 months modeling nearly all U.S. natural gas wells, pipelines, and processing facilities, concluding that dema
Matthew Smith, founder and CIO of Chronometer Partners, brings 20 years of experience in energy markets. Over 18 months, his team modeled nearly all U.S. natural gas wells, pipelines, and processing facilities, reaching a conclusion sharply at odds with mainstream market views: Starting in 2028, AI data centers and LNG exports will demand more natural gas than the U.S. can produce and transport; by his estimates, the U.S. could exhaust working gas inventories by 2030, making the upside price risk "unbounded and convex."
Matthew Smith believes that the seeds of the U.S. natural gas crisis were planted long before the rise of AI.
U.S. shale gas has experienced explosive growth since 2010, transforming the country from a natural gas importer into an exporter. Currently, U.S. LNG export capacity stands at approximately 15 BCF/day (accounting for 12-15% of the daily production of 110-112 BCF). By the end of 2030, approved projects under construction will boost export capacity to 35 BCF/day. Smith points out that these LNG projects require years of advance approval, financing, and construction, and "most of the projects moving from 15-16 BCF to 35 BCF are already underway"—this has been the primary driver of incremental U.S. natural gas demand over the past 10 years and the next five.
On this foundation, the additional electricity demand from AI data centers further intensifies the pressure. Smith's team categorizes AI-related natural gas demand into "P50" (projects with a ≥50% probability of being signed, having power purchase agreements, or interconnection agreements) and more aggressive scenarios. Under the P50 scenario, incremental natural gas demand from AI is approximately 5 BCF/day; if all proposed projects are considered (P30 or even P0), this figure could more than double, reaching 12-15 BCF/day.
> "After evaluating every gas well and the entire pipeline, processing, and gathering system, our conclusion is that the U.S. has the capacity to add approximately 20 BCF/day of natural gas production. Even without AI computing, LNG exports and new supply capacity were only barely matched. Now, with AI added, a gap emerges."
Key Data Chain:
Smith emphasizes that the issue is not a lack of sufficient natural gas underground in the U.S., but rather that the extraction rate of existing known resources and the infrastructure cannot match demand.
Smith's team calculated the remaining recoverable reserves in the major U.S. gas-producing basins (Appalachia's Marcellus+Utica, Haynesville, Permian) by digitizing each company's mineral rights plots (polygon shapes) and production data for each well. The conclusion is: the maximum deliverable capacity in the U.S. is approximately 128-132 BCF/day — this assumes all companies drill out their remaining mineral rights in an optimal manner.
But the key point is: "We will deplete most of the existing captured inventory of well locations over the next four to five years." New wells have high initial decline rates, and the overall decline curve of existing companies is maturing. Smith notes that some companies claim to have far more drillable well inventory than they actually do, and "when people meet with companies, they should ask to see precise engineered location maps."
Even after natural gas is extracted from the ground, it must pass through three stages to reach users:
1. Processing: Natural gas contains impurities such as NGLs, sulfur, and nitrogen, and must be processed to pipeline specifications (1010-1030 BTU). Current processing capacity is near saturation, and building a new processing plant takes 2-3 years. To achieve the target of increasing production by 20 BCF/day, construction of more processing facilities must begin immediately.
2. Gathering Pipelines: These are "small-diameter pipelines" from the wellhead to processing plants or trunk lines. Smith's team has mapped all disclosed expansion projects, but "even to achieve a small portion of the production increase target, massive investment in gathering systems is needed immediately."
3. Interstate Natural Gas Pipelines: This is the most severe bottleneck. Over the past 10-12 years, only one interstate natural gas pipeline has been built in the U.S. (the Mountain Valley Pipeline, connecting Appalachia to the Mid-Atlantic region). Environmental permitting and regulatory changes have made constructing interstate pipelines extremely difficult. The current administration is working to lower these barriers, but "the urgency lies in the need to build more connections to transport natural gas across the country to serve AI computing loads."
Smith's model shows that if everything continues along its current trajectory, the U.S. natural gas system will face unprecedented stress between 2028 and 2030.
Smith declines to give a specific price target but notes that historically, natural gas reached $8-10/MMBtu during the Russia-Ukraine crisis and was pushed to $6-10 by weather events like polar vortices—but these were "temporary." The current situation, by contrast, is driven by structural demand that does not align with known producible supply.
> "The upside risk to natural gas prices becomes unbounded and convex."
In extreme cases, spot LNG cargoes may no longer leave the U.S. (as domestic prices become higher), but this would only alleviate part of the pressure. To truly resolve the issue, contracted cargoes would need to be cut—which involves the power supply of foreign buyers such as Japanese utility companies, making it extremely difficult to execute.
Smith points out that the current natural gas forward curve for 2028-2030 remains flat (around $3.50), because "natural gas has put everyone to sleep." The market has been numbed by 15 years of ample supply, and no one is truly investing in new supply. EQT is actively shutting in wells (believing the future will be more valuable), and rig counts are not increasing—"this complacency will take us straight to the point where it is too late."
| Category | Specific Name | Rationale |
|---|---|---|
| Natural Gas Producers | Expand Energy (Most Benefited) | Controls approximately 70% of remaining core well locations in the Haynesville Basin, with the highest quality rock parameters. Currently trading at a steep discount due to a vacant CEO position, at 4x EBITDA (based on a forward curve that no one believes Smith's view). "The assets haven't changed, but the stock price has fallen." |
| Natural Gas Producers | Range Resources (Best in Appalachia) | Has significant room for production growth and the ability to return capital to investors. |
| Utility-Scale Solar | XPLR (formerly NextEra Yieldco), Clearway Energy | Natural gas is the marginal fuel for power generation; rising gas prices lift overall electricity prices, while solar fuel costs are zero. These companies "can achieve profit expansion without additional capital expenditure"—PPAs are revalued at market prices. |
| Residential Solar | The entire industry | Even if tax incentives are removed, the economics of installing solar + batteries become extremely compelling in a high electricity price environment. "This is one of the only ways to protect yourself from peak electricity prices between 10 a.m. and 6 p.m." |
| Category | Specific Name | Rationale |
|---|---|---|
| U.S. Consumers | All | Rising electricity prices will directly impact household bills. "Are we exporting natural gas to foreign buyers, using it for AI computing, or keeping consumer electricity prices low? This is a terrible trade-off." |
| Distributed Generation Equipment Manufacturers | Bloom Energy, Caterpillar (Solar Turbines business) | These companies are ramping up production at scale, but Smith believes that by 2029-2030, natural gas scarcity will prevent these assets from securing fuel. "Bloom's 2GW or more of assets will not be able to compete with all other gas-consuming assets." Caterpillar plans to double Solar Turbines capacity by the end of 2029—"the timing is completely wrong." |
| Hyperscale Cloud Providers | Unspecified companies | Energy currently accounts for approximately 10% of their costs. If natural gas prices double or triple, energy costs could rise to 20-30%. "Performance per watt" will become a more important computing metric. |
Smith argues that no single "silver bullet" can solve this problem, but large-scale nuclear power plants are the only viable long-term solution.
"Many SMRs are still scientific experiments." The NRC and the government are working to lower barriers, but "these SMR companies lack the manufacturing and scaling capabilities needed to truly address the problem—we need tens of gigawatts of installed capacity."
Smith points to the Westinghouse AP1000 design. Although the most recent U.S. construction (Vogtle Units 3 and 4) saw costs overrun by 3 times and took 15 years, China is building 39+ nuclear reactors, with about one-third based on the AP1000 design. "Our understanding of building large-scale nuclear plants today is far greater than when those mistakes were made."
Timeline: The earliest possible commissioning is 2033-2034—which is already "as fast as it gets."
| Company | Rationale |
|---|---|
| Cameco (49% ownership, Brookfield holds 51%) | Westinghouse is undervalued within its structure; value may be unlocked at IPO |
| BWXT | Primary nuclear supplier to the U.S. Navy, with significant value content in the AP1000 |
Smith would do two things:
1. Government-funded construction of 2-4 AP1000 units (utilizing approximately $260 billion in loan program office capacity) to de-risk the supply chain and attract subsequent investors. "Currently, 10-20 nuclear plants are envisioned in the U.S., but no one wants to be the first."
2. Aggressively incentivize residential solar—"get everyone to install solar systems on their homes."
The United States has become one of the world's largest natural gas suppliers, with exports rising from zero to 15 BCF/day and plans to reach 35 BCF/day by 2030. Many countries are building power generation assets that rely on U.S. natural gas supplies.
If the U.S. fails to fulfill its export commitments:
Smith identifies the factors most likely to invalidate his forecast:
1. Permian Basin production increase: However, higher oil prices are needed to incentivize more drilling (with natural gas as a byproduct), and high oil prices themselves would exacerbate the consumer crisis.
2. Breakthrough in battery technology: Commercialization of sodium-ion or other non-lithium-ion batteries significantly altering the economics of electricity storage — "This would be a watershed moment, and I would welcome it."
3. AI computing demand falling far short of expectations: Either due to performance breakthroughs or demand contraction.
Questions he suggests CEOs ask themselves:
| Position | Analyst View | Key Data |
|---|---|---|
| Expand Energy | Bullish (biggest winner) | Controls ~70% of core well locations in Haynesville; currently trading at 4x EBITDA |
| Range Resources | Bullish | Has significant room for production growth |
| XPLR (formerly NextEra Yieldco) | Bullish | PPAs will be revalued as electricity prices rise, requiring no additional capital expenditure |
| Clearway Energy | Bullish | Similar logic |
| Cameco | Bullish | Holds 49% of Westinghouse; Westinghouse IPO may unlock value |
| BWXT | Bullish | U.S. Navy nuclear supplier, benefiting from the nuclear construction cycle |
| Bloom Energy | Risk Warning | 2GW+ capacity may not secure natural gas fuel supply |
| Caterpillar (Solar Turbines business) | Risk Warning | Plans to double capacity by end of 2029, timing is wrong |
| Westinghouse (unlisted) | Bullish | AP1000 design is the only viable large-scale nuclear solution |
1. "The upside risk to natural gas prices has become unbounded and convex." — Matthew Smith, describing the structural shortage characteristics of the natural gas market after 2028.
2. "Over the past 10-12 years, only one interstate natural gas pipeline has been built in the U.S. Environmental permitting and regulatory changes have made construction extremely difficult." — Matthew Smith, pointing out that infrastructure bottlenecks are one of the core causes of the crisis.