PJM Near-Record Demand Meets Shrinking U.S. Refinery Capacity

U.S. refinery capacity just hit its lowest level since before the pandemic - 18.2 million barrels per calendar day as of January 1, 2026, per the EIA. The same week, PJM Interconnection⁴ - the grid serving 67 million Americans - pushed within striking distance of an all-time power demand record, driven by a heat wave colliding with data center load that never sleeps. The mainstream story says this is an electricity crisis, full stop. That story is missing half the evidence. Refineries are running at 96.1% utilization with less capacity than they had a year ago, jet fuel production is at record highs post-Hormuz, and the largest data center project ever proposed in the United States just died in a Virginia courtroom. One of these facts is supposed to mean demand is manageable. It is not. We are going to show you why.

The Grid Is Screaming - and Nobody Is Counting the Diesel

PJM Interconnection manages the largest wholesale electricity market in the world. This week it nearly broke its all-time demand record. The cause was not a mystery: a heat wave layered on top of a data center load that has been climbing for three consecutive years without pause. According to reporting from OilPrice.com this week, PJM operators were managing demand levels that put the grid within reach of records set during the worst heat events in the region's history.

Here is the number that matters: 67 million people. That is how many Americans live inside PJM's service territory - from New Jersey to Illinois, from Maryland to Michigan. When that grid strains, it is not an abstraction. It is hospitals, water treatment plants, and - critically - the diesel backup generators that every major data center in Northern Virginia keeps spinning on standby.

The mainstream energy press covered this as a power story. Megawatts, transmission lines, grid reliability. What they did not cover is the fuel sitting in tanks behind every one of those data center campuses. Data centers do not run on hope when the grid goes down. They run on diesel. And that diesel demand is structurally absent from every demand forecast that calls this an electricity-only story. That is not an oversight. That is a category error with real consequences for petroleum markets.

The Blackstone/QTS Collapse Changes the Math Entirely

On July 2, 2026, Blackstone and QTS withdrew their appeal to the Virginia Supreme Court, officially killing what would have been the largest data center project ever proposed in the United States. The project was not canceled because demand for data center capacity disappeared. It was canceled because the power infrastructure to support it does not exist on the timeline the project required.

This is the fact that breaks the mainstream narrative wide open. The bull case for grid-connected, utility-powered data centers just lost its flagship project. What replaces it? Distributed generation³. Smaller facilities. Facilities that cannot wait for new transmission lines to be permitted, approved, and built over a decade-long timeline. Facilities that run on natural gas generators and diesel backup systems that log far more operating hours than the industry's public-facing sustainability reports suggest.

Think of it this way: the Blackstone/QTS project dying is not a demand story going cold. It is a demand story getting rerouted. The compute capacity the market needs does not vanish because one project failed. It gets built in smaller increments, in more locations, with more distributed fuel consumption that never shows up cleanly in a single EIA category. The grid crunch that killed the mega-campus is the same grid crunch that guarantees diesel-dependent distributed generation grows faster than anyone is modeling.

18.2 Million Barrels - and the Floor Is Still Falling

The EIA published a number on January 1, 2026, that should have stopped every energy analyst in their tracks: U.S. operable refinery capacity stood at 18.2 million barrels per calendar day. That is a decline of more than 250,000 barrels per calendar day from the prior year - roughly a 1.4% drop in the nation's ability to turn crude oil into the products that actually run the economy.

250,000 barrels per calendar day. To put that in physical terms: that is the equivalent of a mid-sized refinery simply ceasing to exist. Not cutting runs. Not doing maintenance. Gone. Permanently removed from the nation's processing infrastructure. And it happened in a single year.

That is not a maintenance story. That is not a turnaround story. That is a structural retirement of American refining muscle at exactly the moment product demand is hitting records. Imagine a hospital system closing one of its largest facilities while the population it serves keeps growing. You do not solve that by running the remaining hospitals harder. You create a system with no margin for error - which is precisely what 96.1% utilization means.

96.1% - The Number That Proves There Is No Cushion

For the week ending June 19, 2026, the EIA reported U.S. refinery utilization at 96.1%. That is not a sign of health. That is a system operating at the edge of what is physically sustainable over time.

Refineries are industrial facilities that require planned maintenance, unplanned repairs, and operational buffer to handle feedstock variability. Industry practitioners generally consider 90-92% utilization to be a healthy, high-utilization operating rate. At 96.1%, there is almost no room left. Any unplanned outage - a fire, a mechanical failure, a hurricane making landfall on the Gulf Coast - does not get absorbed by spare capacity. It goes directly into product markets as a supply shock.

Now combine that with the capacity number. The refineries running at 96.1% are doing so against a smaller base than existed twelve months ago. The absolute throughput ceiling is lower. The margin for error is lower. And the demand side - jet fuel at record production, diesel for distributed generation, gasoline for a driving season that has not softened - is not lower. The bears who say refinery utilization is high because demand is being met are reading the instrument correctly and drawing the wrong conclusion. High utilization on a shrinking capacity base is not a sign of balance. It is a sign of a system with no shock absorbers left.

Jet Fuel at Records - the Hormuz Effect Nobody Priced In

When the Strait of Hormuz⁵ effectively closed to normal commercial traffic in March 2026, jet fuel prices doubled. That is not a figure of speech. The EIA documented the price move, and the market response was immediate: U.S. refiners pushed jet fuel production to record highs to fill the gap left by Middle Eastern supply disruption. As of the most recent EIA data, U.S. jet fuel production remains at those record levels.

Record jet fuel production from a shrinking refinery base running at 96.1% utilization means something specific: every barrel of crude being processed is being allocated with almost no slack. Refiners are making choices about yield - more jet fuel means less of something else. The refinery is a zero-sum system. You cannot maximize everything simultaneously. When jet fuel gets prioritized because prices doubled and demand is inelastic, other product yields get squeezed. That squeeze does not show up in a single headline. It shows up in basis differentials, in regional product shortages, and eventually in price.

The Hormuz disruption was supposed to be temporary. The production records it created are not temporary. They are baked into a refinery system that has no room to add throughput and no new capacity coming online to relieve the pressure.

The Mainstream Case - and Where It Breaks

The bears are not stupid, and their argument deserves a fair hearing. Here is their best version of it: OPEC production is rising, global demand growth is moderating, and the AI data center buildout is primarily a problem for grid operators and utility companies - not for oil markets. They point to crude inventory builds earlier this year as evidence that supply is adequate. They note that EV adoption is accelerating in key markets. They are winning the tape on crude prices in the near term, and it would be dishonest to pretend otherwise.

But here is where the fine print breaks their case. The crude inventory builds they cite happened before refinery utilization hit 96.1% on a capacity base that is now 250,000 barrels per day smaller than it was. Crude in a tank is not the same as refined product on a shelf. The bottleneck has shifted from crude supply to refining throughput - and that bottleneck is tightening, not loosening.

On data centers: the bears are correct that electricity is the primary input. They are wrong that diesel is irrelevant. Every hyperscale⁶ facility in Northern Virginia - the largest data center market on earth - maintains diesel backup generation sized to run the entire facility for 48 to 72 hours. When PJM strains toward record demand, those generators do not sit idle as a precaution. They run. And that fuel consumption does not appear in any demand model that categorizes data centers as an electricity story only. The Blackstone/QTS collapse makes this worse, not better, because distributed generation burns more diesel per unit of compute than centralized utility power does.

According to Wood Mackenzie's mid-2026 power and renewables research, the failure of large-scale grid-connected data center projects in constrained markets like Northern Virginia is accelerating the adoption of behind-the-meter and distributed generation solutions - a trend that carries meaningful implications for diesel and natural gas liquids demand that traditional electricity-focused demand models have not fully captured.

Where Kingdom Exploration Stands

The data in this article - shrinking refinery capacity running at 96.1% while demand hits records from PJM to the jet fuel market - is exactly the supply-demand tension we screen for when evaluating American upstream production projects. Kingdom Exploration focuses on direct participation in U.S. oil and gas development, with projects structured to generate production economics that hold up in the $40s per barrel. Costs associated with these programs may be deductible up to one hundred percent in the year of investment - talk to your tax advisor about your specific situation before making any decisions.

Oil and gas investments involve significant risk, including potential loss of principal. Past performance does not guarantee future results. Tax benefits depend on individual circumstances. Consult your financial advisor and tax professional before investing.

U.S. refinery capacity fell 250,000 barrels per day in 2026 while utilization hit 96.1% and PJM neared an all-time demand record - the AI data center boom is a petroleum story, and the market has not priced it yet.