The Economics of Owning an Oil Refinery

ELI5 / TLDR

A refinery doesn’t make money on oil. It makes money on the gap between what it pays for crude going in and what it sells the gasoline and diesel for coming out. That gap is called the crack spread, and it swings wildly for reasons no operator controls. So the same plant can mint a billion dollars one year and barely break even the next. Add a permitting system so brutal that nobody had built a new refinery in the US for 50 years (until March 2026), plus the risk that one bad startup can kill workers and cost $2 billion, and you have one of the most paradoxical businesses on Earth: monopoly-like scarcity wrapped around razor-thin, uncontrollable margins.

The Full Story

The business is the gap, not the oil

The instinct is that a refinery profits when oil is expensive. Wrong. What it actually wants is a wide gap between the price of crude and the price of the finished products.

A refinery doesn’t actually want high oil prices. It wants a wide gap between oil prices and product prices.

The industry name for that gap is the crack spread — “crack” because the refinery cracks one barrel of crude into several lighter products. Wall Street’s standard yardstick is the 3-2-1 crack spread: three barrels of crude in, two barrels of gasoline and one of diesel out, with the dollar difference per barrel being the gross margin. Historically a healthy spread runs $10–20 a barrel; during wars, hurricanes, or outages it can blow past $30–40.

Here’s the mechanism that confuses people. When crude spikes — say a Middle East conflict pushes Brent toward $112 — pump prices usually chase it fast, but the refinery’s cost of buying that crude just spiked too. If gasoline catches up quickly, the refiner’s margin barely moves. The money is made in the lag: when product prices trail crude for a few weeks before catching up, the refiner pockets an enormous, almost accidental profit. Earlier in 2026, Gulf Coast refiners buying cheaper domestic WTI crude but selling at global product prices saw margins widen overnight — nothing to do with how well they ran their plants.

Five businesses in one building

A refinery doesn’t make one product. From a single barrel it simultaneously pulls gasoline, diesel, jet fuel, naphtha, and liquefied petroleum gas — all priced differently, all moving independently in their own markets.

These markets have their own calendars. Gasoline peaks every summer in what the industry literally calls “driving season” (June–September), which can widen gasoline cracks by $3–8 a barrel over winter. Diesel is steadier but spikes in planting season and cold winters, when heating oil draws from the same barrel. So the operator is juggling five overlapping bets at once.

It’s less like running a factory and more like running a hedge fund that happens to own some very large metal towers.

Which is why serious players hedge — selling crack-spread futures months ahead to lock in a margin. One real example: a refiner locked a 3-2-1 spread of $28 a barrel for Q3 2026, guaranteed regardless of where crude or gasoline went.

Why nobody builds them

The building was never the hard part — permitting was. Environmental review alone can take a decade: air quality permits, water discharge permits, community impact studies, overlapping federal and state rules. The industry’s rule of thumb for decades:

It’s easier to expand an existing refinery by 50,000 barrels a day rather than build a new one anywhere in the country.

So that’s what everyone did. The last greenfield US refinery was finished by Marathon Oil in Garyville, Louisiana, in 1976 — when Gerald Ford was president and gas was 59 cents a gallon. That plant still runs, expanded in 2009 for $3.9 billion to nearly 600,000 barrels a day. Every barrel of new US capacity for half a century came from expansions, not new builds.

That freeze broke in March 2026, when America First Refining broke ground at the Port of Brownsville, Texas — a 20-year, $300 billion commercial deal anchored by Mukesh Ambani’s Reliance Industries. The headline figure is the commercial contract, not the build: phase one is ~$1.2 billion for about 50,000 barrels a day, scaling to ~164,000 across three phases. Notably, it’s engineered to run on American shale oil rather than the imported heavy crude the rest of the Gulf Coast was built around.

Geography is destiny

Your gas price has almost nothing to do with the station down the street and everything to do with which refinery your state sits near. The Gulf Coast (Texas and Louisiana) is a tight cluster feeding pipelines nationwide — if one plant goes down for maintenance, six neighbors absorb the slack. One shared safety net.

California has none of that. It’s geographically isolated from the national refining network and mandates its own special gasoline blend almost nobody else makes — so it can’t truck in spare capacity. When a California refinery shuts, there’s no backup. Right now two are closing at once: Phillips 66’s LA refinery (end of 2025) and Valero’s Benicia plant (2026), together nearly a fifth of in-state capacity, leaving just seven refineries for ~40 million people. UC Davis economists model a possible $1+/gallon rise; others project more. The lesson: a refinery is regional infrastructure, and when it disappears, the cost lands entirely on whoever stood closest.

The two squeezes — and the body count

Despite the scarcity, refining is one of the worst-performing parts of the oil business right now. Two forces are at work.

Margin collapse. Per the EIA, crack spreads have declined steadily since 2022. Downstream (refining) earnings at big oil companies dropped roughly 50% in 2024 versus the prior year, running about 60% below 2022. Capacity is shrinking — LyondellBasell’s Houston plant and Phillips 66’s LA facility together pulled 400,000+ barrels/day off the market. Analysts see global refining capacity shrinking 10–30% over the next decade as EVs erode gasoline demand (jet fuel stays resilient).

Catastrophic liability. On March 23, 2005, an explosion at BP’s Texas City refinery — then the third largest in the US — killed 15 workers and injured 170+. A blowdown stack was overfilled during a unit startup and a spark from an idling truck lit the vapor cloud. The plant had earned over a billion dollars the year before, roughly $100 million a month right up to the disaster. The aftermath: a felony Clean Air Act guilty plea, a record $50 million criminal fine, a record $21 million OSHA fine, then an $87 million fine when inspectors found the hazards still unfixed. All told, BP paid more than $2 billion tied to one afternoon, plus over a billion more on safety upgrades. That liability sits under every refinery, every day it runs.

Key Takeaways

A refinery’s profit is the crack spread — the gap between crude cost and product sale price — not the absolute price of oil.
The Wall Street benchmark is the 3-2-1 crack spread: 3 barrels crude → 2 gasoline + 1 diesel; healthy is $10–20/barrel, blowing out to $30–40 during shocks.
High oil prices don’t help refiners; what helps is gasoline prices lagging crude on the way up, opening the gap temporarily.
One barrel yields ~five products (gasoline, diesel, jet fuel, naphtha, LPG), each priced in its own market with its own seasonality — so refining is really five simultaneous bets.
Gasoline cracks widen $3–8/barrel in summer “driving season”; diesel/heating oil spike in winter and planting season.
Refiners hedge with crack-spread futures, locking margins months ahead — operationally closer to a hedge fund than a factory.
The last greenfield US refinery before 2026 was built in 1976 (Garyville, LA); permitting, not construction, is the barrier (environmental review can take a decade).
The 50-year freeze broke in March 2026 with the Reliance-anchored America First Refining project at Brownsville, TX — ~$1.2B for phase one (50k bpd), built for shale crude.
Refinery economics are regional infrastructure: the Gulf Coast cluster shares backup capacity; isolated California (with its unique mandated fuel blend) cannot, so closures there could add $1+/gallon.
Refining is currently one of the least profitable links in the oil chain: downstream earnings fell ~50% in 2024, and capacity may shrink 10–30% over the next decade as EVs cut gasoline demand.
The 2005 BP Texas City explosion (15 dead) cost BP $2B+ in fines and settlements from a single startup failure — the standing safety liability under every refinery.
Construction-cost ladder: tiny topping refinery = low millions; real mid-size = $5–15 billion depending on conversion units (hydrocracking, catalytic cracking, reforming) added.

Claude’s Take

This is a tight, well-structured explainer that does the one thing a good economics video should: it replaces a wrong mental model (refineries profit from expensive oil) with a correct and more interesting one (they profit from a spread they don’t control). The crack-spread framing is the genuine payload, and it’s explained cleanly.

The numbers mostly check out against the public record — the 1976 Garyville fact, the 2005 BP Texas City disaster and its ~$2B aftermath, California’s isolation and unique fuel blend, the EIA margin decline. Treat the freshest 2026 figures with mild caution: the $300 billion Reliance/Brownsville deal and the “$112 Brent earlier this year” are recent enough that the framing may be simplified for narrative punch. The “casino with government-mandated scarcity” line is a nice hook but slightly oversells the moat — scarcity here is a regulatory accident, not a deliberate barrier, and the video itself undercuts the monopoly vibe by showing margins are terrible. Minor tension, not a flaw.

What’s missing is depth on why margins are structurally compressing beyond “EVs” — refining is global, and Asian and Middle Eastern mega-complexes (the Brownsville design copying that playbook) are a big part of the Western squeeze. But for a ~15-minute piece, leaving that out is a fair edit. Solid 7: accurate, genuinely clarifying, no obvious BS, just not deep enough on the global supply picture to rate higher.