Set the $106 crack [2] against the $90.33 crude quoted in the same morning market check [7], and the refining step is worth $15.67 a barrel more than the oil going into it, about 117 percent of the crude price [6], which works out to 54 cents of every dollar of value in the wholesale diesel barrel [3]. Margins like that are supposed to summon steel.
Take it down to the pump. Crude plus crack is $4.67 a gallon across a 42-gallon barrel [1], the retail national average is $5.63 [3], and the 96 cents in between covers tax, pipeline, terminal and forecourt [2]; the refiner's slice of what the buyer hands over is $2.52, or roughly 45 cents on the dollar [4]. Retail is still 17 cents, call it three percent, under the $5.80 record set in 2022 [5], and for a record margin to sit beneath a below-record pump price the non-refining part of the barrel has to cost less than it did then [7]. On the gasoline side the national average held above $4 for the whole of August for the first time ever [5], so the pass-through is already in household budgets.
What stops the signal converting into capacity is the mismatch in tenor: a new refinery is billions of dollars and three to five years of construction, by which point demand may have eased [12], plus several permits from federal, state and local authorities [14], against what Energy Aspects analyst Robert Campbell called three months of record margins, saying nobody makes a huge multibillion-dollar investment on that basis [13]. So the units already standing run near maximum [11] and the margin becomes cash rather than concrete. The industry is capturing the best distillate economics on record without adding a single barrel of new distillation capacity.
It can break more than one way. Gulf de-escalation hands most of this back within weeks, which ING half-concedes in calling middle distillate cracks highly elevated and volatile [6]. Or seasonally stronger distillate demand arrives against a global system ING says has little slack [6], and $5.80 goes. Or the pass-through breaks first, retail stalls short of the record because buyers cut consumption, and the margin holds while the pump does not.
This is probably wrong, but the variable worth pricing is duration, not level: a record crack that lasts two quarters is a cash-flow event for the companies that sat in that White House room [9], and one that lasts two years is a capital-spending event for the whole system, and only the second changes the balance at the end of the decade. The read fails the day one of them commits capital to greenfield capacity on these numbers, because that would mean the operators closest to the margin have decided it is structural rather than a three-month window [13].
