The Global Refinery Bottleneck Behind $4.48 Gasoline
$100 oil is a familiar tax. Record diesel is not.
September 21, 2026
The national average for regular gasoline is about $4.48 a gallon — roughly $1.30 above a year ago, and about $1.45–$1.50 above the just-under-$3.00 level on the eve of the U.S.–Israel strikes on Iran. Diesel is about $6.51, a U.S. record. California regular is above $6.15; Texas is under $4.00.¹
The convenient explanation is the Iran war. That is too easy.
Brent has been around $100–$102 a barrel, WTI a few dollars under. America has paid that before — 2008, stretches of 2011–14, 2022. It is not 2008’s ~$147, and it is not the 2022 spike.² Crude is high. It is not at a record. Diesel is. Gasoline has outrun the barrel.
A common rule still works: about $0.24 a gallon at the pump for every $10 on crude, before taxes and a normal refining margin.³ Pre-war futures had 2026 crude in the high $60s to around $70. A $30–$35 premium is $0.70–$0.85/gal if it passed through cleanly, $0.60–$0.75 in practice. That takes ~$3.00 toward the high $3s, maybe $4 with a routine summer crack.
It does not get you $4.48 after Labor Day. It does not get diesel to $6.51. The leftover — about $0.50–$0.80/gal on gasoline above a $100-oil, normal-crack world, and more than that on diesel — is missing refineries. Gasoline cracks have been cited around $40–$50 a barrel, diesel near $100, the 3-2-1 crack near $70.⁴ Refined products rose more than crude.⁵ That is the pump story.
Three shortages sit in that leftover: Gulf plants and product docks hit by the Iran conflict; Russian refineries hit by Ukraine; a U.S. system with no spare capacity. Crude that cannot leave Hormuz is already in the $100 barrel. Don't count it twice. Finished gasoline and diesel that Gulf plants can't make, or can't load, are in the crack.
How a missing refinery shows up at the pump
Oil is not gasoline. When plants have spare room, extra crude becomes extra fuel. U.S. refiners have been at 96–98 percent utilization for much of 2026.⁶ At that rate, an outage, an export pull, or lost foreign product lifts the crack immediately.
They deferred maintenance and, in some cases, ran above nameplate capacity. That does not last. ExxonMobil’s 275,000-barrel-a-day Joliet plant went down in mid-September after a power failure and flooding. Great Lakes prices jumped tens of cents in a week.⁷ No slack, one plant, national news.
U.S. crude output is still near 13.8 million barrels a day.⁸ More crude can change the slate a Gulf plant runs. It does not add a tower. There is enough oil to keep the barrel off a record. There is not enough working capacity to do the same for fuel — diesel especially.
What U.S. capacity actually is
Operable U.S. distillation capacity was 18.2 million barrels per calendar day on January 1, 2026 — down more than 250,000 b/d, about 1 percent, from a year earlier. One hundred thirty operable refineries, two fewer than in 2025.⁹
The recent cuts were lumpy:
- LyondellBasell Houston, ~264,000 b/d, March 2025
- Phillips 66 Los Angeles, ~139,000 b/d, October 2025
- Valero Benicia, ~145,000 b/d, spring 2026
The two California sites were about 20 percent of the state’s capacity. Houston and Los Angeles together removed about 400,000 b/d, only partly offset by small gains elsewhere.¹⁰
The system did not collapse. Capacity peaked near 19 million b/d in early 2020. Net loss since then is about 4–5 percent. There were roughly twice as many plants in 1982. Survivors got bigger. No major new U.S. refinery has been built since 1977.¹¹
$100 oil has lived with spare U.S. refining before. It is not doing that now.
Why the West Coast is the weak link
The Gulf Coast makes more fuel than it uses. The West Coast does not, and little pipe connects them. Lose a California plant and the region imports CARB-spec fuel on ships.¹²
That is why West Coast prices are not “Global chaos plus California taxes.” Same global Brent: Texas under $4, California above $6. The barrel does not explain the spread. Local refineries and logistics do.
The East Coast already walked this road. Regional refinery inputs fell from about 1.3 million b/d in 2008 to about 480,000 by 2025.¹³ The West Coast is on a faster version with less backup. A Western Gateway pipeline is years from solving it.
What tightened the system before the missiles
Covid. About 1.1 million b/d of U.S. capacity left in 2020–21. Some sites became renewable-diesel plants or terminals and will not refine crude again.¹⁴
Old coastal economics. California plants faced high operating and compliance costs against a forecast of falling gasoline use. Phillips 66 said so when it shut Los Angeles.
Soft gasoline demand. Product supplied has been in the high-8 million b/d range, below the pre-pandemic 9.3–9.5 million. Some 2026 summer months look like the weakest June since 2001 outside Covid.¹⁵ Prices are high while Americans burn less gasoline. That is a products shortage, not a driving boom. Diesel and jet demand held up better. Refiners favored those shorter barrels. Exports pulled more fuel out of U.S. tanks.
No new plants. An expansion is still a four-to-five-year path. Capital went upstream, to pipes, and to shareholders.
That left no shock absorber. It did not, by itself, create $4.48 gas. It decided how much of a global product shortage would show up in Illinois and Los Angeles this week.
The global gap is refined products — and it is not only Iran
About 5 million b/d of world refining has been offline at points this year. Enverus puts war-damaged or constrained Middle East and Russian capacity near 7 million b/d, on top of ordinary turnarounds.¹⁶
Ukraine closed a refining region. Strikes have taken out on the order of 30 percent of Russian capacity. Product exports fell to about 1.1 million b/d in mid-summer, a two-decade low. Moscow limited gasoline and diesel exports.¹⁷ Russia was a diesel machine. $100 oil does not explain that. In a $100 year with those refineries running, gas is expensive. Diesel does not automatically set an AAA record. While Zelensky is crowing about recent strikes on Russian refineries.
The Iran conflict closed another — Gulf refineries and product docks, not just crude. Middle East runs are about 7.3 million b/d versus about 9.9 million in February. Pre-war the region exported more than 5 million b/d of products.¹⁸ Q2 runs were even lower, around 6.5 million. Crude “dark transits” recovered some Hormuz oil. Refined cargoes did not recover with them. Product tanks fill, runs get cut, and the missing gallons are diesel and jet, not another tick on Brent.¹⁹
Iran’s own plants are a small slice of that: runs about 2.2 million b/d versus about 2.45 million before the war.²⁰ The U.S. pump feels Saudi, Kuwait, Bahrain, and stranded Gulf product, not Abadan.
That is why U.S. plants can run flat-out and still print fat cracks. The country became the residual supplier to a short distillate world. Inventories tested multi-year seasonal lows. The market clears on price.
What the pump would look like if the Iran conflict’s Refineries ran
Leave oil at $100. Restore Gulf throughput toward February’s ~9.9 million b/d and let product actually load. Leave Russian plants broken. Leave U.S. utilization where it is.
| Fuel | Now | Iran-conflict refining and product exports restored |
|---|---|---|
| Regular gasoline | ~$4.48 | about $4.15–$4.30 |
| Diesel | ~$6.51 | about $5.70–$6.00 |
Call that $0.20–$0.35 off gas and $0.50–$0.80 off diesel. Diesel moves more because the Gulf sold distillate and jet. Gasoline moves less because it is still a U.S. refining market sitting next to Russia and thin stocks.²⁴
You don't get February’s ~$3.00 gas or ~$3.50 diesel. That needs a cheaper barrel and Russian product and spare U.S. capacity. Relief that matters needs Gulf or Russian product flows back. One is not enough.²⁵
Why new U.S. plants will not fix this year
Utilization is already maxed. New refinery capacity takes years. Firms will not bet billions on gasoline demand they expect to keep drifting down under green initiatives. Idled sites get floated. None of them add barrels this fall.
If prices ease, it will be because Gulf docks load, Russian plants run, or driving drops after Labor Day — not because a new CDU appears on the Gulf Coast. EIA still expects fall maintenance to cut runs while distillate stocks stay below the five-year low.²⁷
The prices already made the argument
If crude were the constraint, the barrel would be at or near a record and fuel would track it. The barrel is not near a record. Diesel is. Gasoline is $1.30 above last year because refineries cannot print the gallons.
Iran shut a strait and damaged a refining region. Ukraine damaged another. The United States closed plants and ran the rest at the limit. $100 oil is nothing new. The crack is the new one. Blaming the whole sticker on Iran skips major contributors — and counts Hormuz crude twice. The pump is adding three refined product gaps to a barrel price America has seen many times before.
End notes
- AAA, Sept. 21, 2026: regular about $4.48/gal; diesel about $6.51/gal (record on AAA’s series). Year-ago regular about $3.18. Pre-war regular just under $3.00 on Feb. 28. California above $6.15; Texas under $4.00. Center for American Progress, mid-September AAA comparison: gasoline up about 49 percent vs. Feb. 27; diesel up more than 70 percent.
- Market wraps Sept. 21, 2026 (Mathrubhumi, Vantage, Business Recorder): Brent ~$100–$102, WTI a few dollars lower. 2008 peak ~$147; 2022 spike below that but above today’s print.
- Rough conversion used in California market notes and industry rules of thumb: ~$0.24/gal per $10/bbl crude.
- CNN Business, Sept. 13, 2026 (Auers, Kloza, Woods); Atlantic Council, Sept. 9, 2026; Enverus, Sept. 16, 2026: 3-2-1 near $67–$70; distillate crack ~$95–$100, top percentile of the past 16 years. AAF, Sept. 11, 2026: Gulf Coast diesel crack from ~$20/bbl pre-conflict to above $100/bbl.
- Goldman Sachs, cited in New York Times, Sept. 4, 2026: main refined products rose more than crude over the prior six months.
- EIA weekly utilization as reported by Politico (July 31, 2026), Gulf News/OilPrice (Sept. 2026), IER (June 2026): mid-to-high 90s, at times 97–98 percent.
- The Center Square / GasBuddy (Patrick De Haan), Sept. 17–18, 2026: Joliet ~275,000 b/d offline after power failure and flooding.
- EIA Short-Term Energy Outlook, 2026: U.S. crude production ~13.7–13.8 million b/d.
- EIA, “U.S. refining capacity decreased during 2025,” Today in Energy, June 28, 2026: 18.2 million b/cd as of Jan. 1, 2026; 130 operable refineries.
- Same EIA report: LyondellBasell Houston 263,776 b/cd (March 2025); Phillips 66 Los Angeles 138,700 b/cd (October 2025); Valero Benicia ~145,000 b/d off monthly estimates by March 2026. Combined Houston + L.A. ~400,000 b/d.
- EIA registers compiled in eco3min; CNN Business, Sept. 13, 2026: peak ~18.98 million b/cd on Jan. 1, 2020; no major U.S. greenfield since 1977.
- EIA Today in Energy, June 2026, on limited Gulf-to-West-Coast product pipe and PADD 5 impact. S&P Global, May 2026: California lost a large share of state capacity over five years and shifted toward imports.
- Oil & Gas 360, Sept. 15, 2026, citing EIA: East Coast inputs ~1.3 million b/d (2008) vs. ~480,000 (2025).
- EIA 2021 capacity report; IER, June 2026; eco3min closure tally: ~1.1 million b/d lost in 2020–21, some converted to renewable diesel.
- EIA weekly product supplied ~8.8 million b/d recently. Tom Kloza, Sept. 2026: June 2026 demand among the weakest Junes since 2001 excluding Covid.
- Atlantic Council, Sept. 9, 2026: ~5 million b/d global refining offline. Enverus Intelligence Research, Sept. 16, 2026: ~7 million b/d Middle East and Russian capacity damaged or constrained, excluding routine turnarounds.
- Atlantic Council, Sept. 9, 2026: ~30 percent of Russian capacity hit; product exports ~1.1 million b/d in July–August; export restrictions.
- Kpler, Aug. 20, 2026: Middle East runs ~7.3 million b/d vs. ~9.9 million in February. CGEP, Aug. 28, 2026: pre-war regional product exports exceeded 5 million b/d.
- CGEP, Aug. 28, 2026: Q2 Middle East runs ~6.5 million b/d (−27 percent); crude dark transits recovered to an estimated 40–50 percent of pre-war Hormuz crude; products did not. Kpler: Saudi, Kuwait, Bahrain combine damage and run cuts; UAE and Iran more constrained by product evacuation. Full product tanks force run cuts without a matching crude print.
- Kpler, Aug. 2026: Iranian runs ~2.2 million b/d vs. ~2.45 million pre-war.
- John Auers, Novi Labs, in CNN Business, Sept. 13, 2026: the modest U.S. capacity drop is not the main price driver; combined Russia and Middle East product-supply loss cited around 2 million b/d. Darren Woods, ExxonMobil: pump prices set by refined-product supply and demand, not only crude.
- Brown Climate Solutions Lab / WSJ / Axios: ~$100–$107 billion extra gasoline and diesel since Feb. 28. ITEP, Sept. 21, 2026: ~$727 per household. Moody’s Mark Zandi: ~$115 billion including jet, ~$860 per household. Dallas Fed working paper: gasoline ~+30 percent through late June. AIER / Daily Economy, May 2026: ~$0.85/gal average premium vs. a no-war forecast.
- CREA, “What the Hormuz crisis has cost fossil fuel importers — March to August 2026”: crude ~+35 percent vs. pre-war curve; gasoline ~+43 percent; diesel/gasoil ~+59 percent.
- Bipartisan Policy Center, June 2, 2026: Middle East slates and pre-war seaborne jet/diesel trade explain larger distillate and jet moves. AAF, Sept. 2026: diesel also tight from Russian bans, 98 percent U.S. utilization, and jet yield competition.
- Warren Patterson, ING, in NYT, Sept. 4, 2026: with little spare refining capacity, meaningful relief requires a recovery in Persian Gulf and/or Russian product flows.
- WSJ / To Vima, June 28–29, 2026: during the brief Hormuz-reopening scare, Brent traded near ~$72, close to pre-war. The barrel moved; product cracks did not instantly normalize.
- EIA STEO, Sept. 2026: distillate inventories forecast below 100 million barrels in September and below the 2021–2025 low into 2027; seasonal maintenance cuts runs in September–October.
