The Nat Gas trade is where the memory trade was one year back, and nobody seems to be noticing. Here’s the hidden truth nobody is telling you, and what I’m buying (save this) The US is heading into a historic natural gas supply deficit in 2028 and nobody is ready for what’s coming. Current production sits around ~110-112 Bcf/d of natural gas per day, and the market can add up a further ~20 Bcf/d of new supply through additional drilling across the key basins of Appalachia, Haynesville, and the Permian which brings the total max deliverability of natural gas to ~132 Bcf/d. But that max deliverability figure is the ceiling, not a promise. It assumes producers can optimally run their drilling operations, and more important, the infrastructure to deliver that gas gets built. This is where the argument breaks. It took the United States 10+ years to build exactly one interstate gas pipeline, the Mountain Valley Pipeline between Appalachia, and the Mid-Atlantic capable of transmitting ~2 Bcf/d. The other bottleneck even before a millimetre of gas is available for transmission sits at the processing stage. At this leg of the natural gas value chain, raw gas extracted upstream off the gas-bearing rock is processed to get rid of contaminants to extract pure methane. The processing equipment deployed today is operating at max capacity, and any new buildout will take a minimum of ~2-3 years before any real capacity addition impacts the gas supply available for sale. Gas already accounts for >40% of the US power generation requirements. Stack the total current and projected demand addition in comparison to the ~20 Bcf/d potential supply headroom, and it's clear why the nat gas trade is both immediate and convex. The primary demand forces come from two sources, LNG exports, and power demand from AI data centers. The US produces ~112 Bcf/d and plans to scale it further by ~20 Bcf/d (a base growth of ~18%). However, take into account LNG exports which today are ~15 Bcf/d and are projected to hit the mid-30s Bcf/d by 2030 so exports as a % of total production go from roughly 13% to a number that will eat away the entire excess production even before a singular unit is available for the energy demand from AI compute workloads which is projected to add anywhere between ~5-15 Bcf/d of net new demand with no source to fulfill it. This is what makes this a demand side convex opportunity much like the memory trade of 2025. Critics argue the world isn’t heading for a natural gas supply crunch. The lazy argument is that gas is abundant, and quote the flat forward price curve at the Henry Hub, the official delivery point for the NYMEX natural gas futures contract as evidence. Data shows the current spot price for gas is ~$2.63, and average across calendar years 2027 (~$3.35), 2028 (~$3.6-3.8), 2029 (~$3.7-3.9), and 2030 (~$3.7-3.9) shows that the market is pricing gas essentially at the same level in 2030 as it did in 2028. Most physical gas today trades at the Henry Hub spot/monthly index price, not at a fixed future price. So when the deficit does indeed hit, the spot prices run to ~$8-10, and the bulk of gas eventually re-prices to spot. This is what the market is missing, and is exactly the whole trade. So to close, the bottleneck is not gas. The US has plenty of it, and current prices reflect this excess. The problem is the bottlenecks around processing and transmission which makes the whole setup so convex. As an investor I’m always on the lookout for such opportunities, and looking at the natural gas value chain closely enough reveals the actual names primed to capture the maximum upside. The first name is Expand Energy (EXE), the largest US gas producer at 7.48 Bcf/d, operating at a breakeven cost below $3, sitting at 0.5x leverage, and generating ~$343m of FCF and ~$522m of net income last quarter. Trading at 5x trailing, and 4x forward EBITDA it is the cheapest large-cap gas producer in the US. The reason it's the best positioned play moving into the 2028 convexity is because its 2028 book remains open. What this means is it sells into new demand at rising spot prices rather than a locked forward price of ~$3.50 today. The current multiples don’t capture this, and both its cash flows and multiples will re-rate aggressively when that curve eventually wakes up. Its close twin Range Resources (RRC) comes in second place. It sits on more than 30 years of premium, low cost Marcellus inventory with only 0.5x leverage. It earned $195m last quarter and the reason it belongs here is exactly the same as EXE. 95% of its 2028 production is unhedged meaning just like EXE, it too will capture higher cash flows, and see its multiples re-rate aggressively selling into demand in that rising price curve. The next one on my radar is Kinder Morgan (KMI) which runs the largest gas network in the US. It moves ~40% of the natural gas consumed in the US and sits as the toll booth operator in the exact bottleneck the thesis posits. The most recent quarter saw the company print a record $2.2B of quarterly EBITDA on a $9.6B project backlog that is 92% gas and about 60% tied to power generation, so the backlog itself is the AI buildout. At ~12x forward EBITDA, 3.6x leverage and a 3.7% yield, it captures the rising volume of gas that has to move through a constrained system. The catch is that a toll road earns on throughput, not on the gas price, so the upside is real but still capped. The last name on the record is directly tied to the rising demand for gas from AI data centers. Williams (WMB) owns Transco, the largest volume gas pipeline in the US, and is the backbone carrying Gulf Coast gas into the highest demand markets including Virginia, the biggest data center cluster on the planet. It is also the only midstream name building power directly for AI: Socrates, 400 MW of dedicated off-grid gas generation for a Meta data center, live by year end, and a 750 MMcf/d Transco expansion to feed the Virginia load. Management guides FY26 EBITDA at ~$8B+ with ~$7B of power-focused capex. It owns the physical link between the gas and the data center, which is the cleanest demand pull in the chain. Just like KMI, Williams is a beneficiary of this demand, but doesn’t directly benefit from gas price increase like the producers, and as such offers a steady, but not asymmetric upside. To close, the whole trade is to long the cheap gas producers into a rising demand curve, and flat forward price curve about to get ripped, EXE and RRC are the whole trade. One honest caveat, because this is not the memory trade in one important way. Gas supply is short cycle meaning that where memory had three makers and multi-year fabs capping supply, a gas producer can drill and complete a well in months, and Permian associated gas grows regardless of the price, since it comes up as a byproduct of oil drilling, adding several Bcf/d of price-insensitive new supply right into the deficit window. So the flat curve is not only complacency, it is also a bet that any 2028 tightness gets drilled away. This trade works only if three things hold: producer capital discipline stays intact, the processing and pipeline bottlenecks stay tight, and LNG demand shows up on schedule. The day a large producer breaks ranks and floods supply at $6 gas, the thesis weakens into rising supply. I am betting the curve is asleep, not efficient, and that is the other side of the trade you need to know and factor before dropping a penny into any of the names above. I’m @degenrsc, an investor and deep tech researcher looking to make money putting in an honest day of work trying to find asymmetry in financial markets. Follow me for investment insights across emerging frontier tech segments like AI, quantum, bio/acc, and crypto.