For three decades the United States ran a quiet arbitrage: build almost no new refineries, but keep squeezing more product out of the ones already standing. Debottlenecking, catalyst upgrades, and crude-slate optimization let operable capacity drift upward even as the plant count fell. That arbitrage has now run out of room. In a span of roughly fifteen months, four large refineries have closed or announced closure, and the inventory math that used to absorb shocks has gone from comfortable to brittle. This is not a price spike. It is a structural change in how much middle distillate — diesel and jet fuel — the country can make.

The headline number is blunt. Seven major closures and conversions have removed roughly 1.235 million barrels per day of crude processing capacity, more than 6.5% of U.S. operable atmospheric distillation capacity, which stood at about 18.4 million barrels per calendar day in January 2025. Absent further debottlenecking, U.S. capacity in 2026 falls below the 2023 level of 18.06 million bpd. The plant is shrinking while consumption is not.

The Closures Are Concrete, Not Forecast

This is worth stating precisely because the energy debate is full of speculative timelines. These closures are not projections; they are filings and shuttered units. LyondellBasell permanently closed its 263,776 bpd Houston refinery in early 2025. Phillips 66 is shutting its 138,700 bpd Los Angeles (Wilmington) complex by the end of this year. Valero filed in April to end refining at its 145,000 bpd Benicia refinery in the San Francisco Bay Area by the end of April 2026. The Wilmington and Benicia shutdowns alone strip roughly 284,000 bpd of capacity out of California, a market that is already an island — pipeline-isolated from Gulf Coast supply and dependent on a specialized CARB fuel slate that few outside refiners produce.

The reasons these plants are closing are not cyclical. They are old, complex coastal facilities carrying high maintenance burdens, sitting on valuable real estate, facing tightening state policy and falling long-run gasoline demand. Once a refinery this size is decommissioned, it does not come back. The capital, the permits, the skilled operators, and the social license to rebuild on that footprint are effectively gone. Each closure is a one-way door.

Why Distillate Is the Pressure Point

Not all barrels are equal, and this is where the story gets interesting for anyone who actually runs a process unit. A refinery is a yield machine: a barrel of crude is split into gasoline, distillate (diesel and jet), and heavier and lighter cuts according to the configuration of the plant and the crude it runs. Gasoline demand in the U.S. is in slow structural decline as the light-duty fleet electrifies. Distillate demand is not. Diesel moves freight, powers agriculture and construction, and runs the backup generators behind data centers. Jet fuel demand is growing. These are the molecules that are hard to electrify and hard to substitute.

So the closures hit the wrong barrel. The capacity coming offline was producing distillate the economy still needs, and the plants are not being replaced by units optimized for the cuts that are tightening. The U.S. Energy Information Administration forecasts that distillate inventories will fall to their lowest levels since 2000 in 2026, driven by the inventory draw in 2025, strong export demand, and reduced domestic production as refineries close. Days of supply — the metric that actually matters for resilience — sits well below historical averages even before the next unplanned outage.

One partial cushion deserves an honest mention. Biofuels are filling some of the gap: biodiesel and renewable diesel are forecast to make up about 9% of U.S. distillate fuel oil consumption in 2026, up from 5% in 2021. Counting biofuel stocks adds roughly 10% more days of distillate supply than the petroleum-only figure. That helps. But it does not change the direction of travel, and renewable diesel feedstock economics carry their own volatility. The structural petroleum distillate base is still thinning.

The Resilience Problem Hiding in the Average

Aggregate national capacity numbers conceal the real risk, which is geographic and operational concentration. When the refining base was larger and more distributed, a single unplanned outage — a fire, a hurricane, an unplanned turnaround — was absorbed by spare capacity elsewhere. As the fleet contracts and consolidates onto fewer, larger Gulf Coast plants, the system loses that shock absorber. A 2026 inventory level not seen since the early 2000s means there is less buffer between a regional supply disruption and a genuine shortage at the rack.

This is a process-safety and reliability story as much as an economics story. The remaining plants will be asked to run harder, at higher utilization, for longer stretches, to cover the lost volume. Higher utilization with thinner margins for unplanned downtime is precisely the operating regime where mechanical-integrity discipline and management-of-change rigor stop being compliance line items and start being the difference between supply and shortage. A fleet running at the edge of its operating envelope has no slack for the kind of latent equipment risk that a well-run inspection program is supposed to catch early.

The Demand Side Is Getting Less Elastic, Not More

The usual comfort in a tightening fuel market is that demand will bend — high prices ration consumption and the system rebalances. For distillate, that elasticity is weakening, and the AI build-out is part of the reason. Diesel sits underneath the load-bearing parts of the economy that cannot pause: freight that has to move, harvests that have to be brought in, and increasingly the backup generation behind critical digital infrastructure. The data centers being built to train and serve AI models are provisioning enormous standby diesel generator fleets to bridge grid interruptions, because an unplanned outage at a hyperscale facility carries an extraordinary hourly cost. That is distillate demand that does not care about price and does not substitute — it is insurance, and operators buy it regardless of the crack spread.

Layer that onto freight, agriculture, construction, marine, and a jet-fuel demand curve that is still climbing, and the picture is a distillate demand base that is structurally inelastic at exactly the moment the supply base is contracting. Gasoline can give ground gracefully as electrification advances; diesel and jet cannot, because the things they power have no near-term electric substitute at scale. A market where supply falls in irreversible steps and demand refuses to bend is the textbook setup for sustained, volatile premiums — not a passing spike.

What This Means for the Next Decade

The conventional response to a supply gap is imports, and the U.S. will lean on the global product market to balance. But importing diesel and jet into pipeline-isolated regions like the West Coast is expensive, slow, and exposed to the same global tightness everyone else is bidding into. Imports balance the books on paper; they do not rebuild domestic resilience, and they hand pricing power to refiners abroad.

The more durable answer is to rethink what new domestic refining capacity looks like. The era of the 250,000 bpd grassroots coastal mega-refinery is over in the United States — the permitting, capital, and demand assumptions that justified those plants no longer hold. What the closure data argues for instead is the opposite shape: smaller, modular, distillate-first units built close to demand, configured deliberately to maximize the middle-of-the-barrel cuts that are structurally short rather than the gasoline that is in structural decline. Capacity you can site near a freight corridor, an agricultural basin, or a data-center load center, and build in phases rather than betting a decade of capital on a single site.

This is the thesis Porritt Inc. has been building toward: modular, distillate-first refining that targets the barrels the market actually needs, paired with the AI-driven process-safety and engineering-compliance tooling that lets a small operation run a complex unit safely. The macro data released through the first half of 2026 does not just permit that thesis — it demands it. When 1.2 million barrels a day of capacity leaves the system and the cut that tightens fastest is the one that is hardest to replace, the market is telling builders exactly where to point.

The Economics That Matter

Strip away the policy noise and the picture is simple. Supply of domestic distillate capacity is falling in discrete, irreversible steps. Demand for the specific molecules that capacity produced — diesel and jet — is flat to growing and stubbornly hard to electrify. Inventories, the buffer that smooths the gap, are draining to levels not seen in a generation. That combination does not resolve through a single good quarter of refining margins. It resolves through structurally higher and more volatile distillate cracks, more frequent regional supply scares, and a sustained premium for whoever can put new, well-configured distillate capacity on the ground.

The refineries closing in 2025 and 2026 are not coming back. The question is whether the next generation of capacity is a smaller, smarter, distillate-weighted version of what was lost — or whether the country simply imports its way into permanent dependence. The molecules will decide, and right now the molecules are pointing at diesel.

---

Timothy Porritt is founder of Porritt Inc., building AI-powered tools for process safety, engineering compliance, and industrial operations. Based in Salt Lake City, Utah.