Marathon Petroleum Corporation MPC

Key data

The business

Marathon buys crude oil, runs it through refineries, and sells gasoline, diesel and jet fuel to wholesalers and its own branded stations. That is the engine, called Refining & Marketing, and it is a price-taker: the spread between what crude costs and what fuel fetches, the crack, is set by the market, not by Marathon. When the crack is wide, as it was this spring, the money is enormous. When it is thin, the same plants barely break even. Bolted to the refiner is MPLX, a separately listed pipeline and gas-processing company Marathon controls, which charges fees to move and treat hydrocarbons and pays cash up the chain regardless of the crack. MPLX just paid $4.5 billion of new notes to buy Northwind Midstream, over 200,000 dedicated acres of sour-gas gathering in Lea County, New Mexico. That is the durable half of the company, and it is where the capital is now flowing.

The numbers

Q2 2026 is not a trend, it is a spike.

Profit ran more than four times the year-ago quarter and EPS of $17.73 came in a quarter after $1.73, on a crack that widened hard while a Q1 heavy with turnaround maintenance rolled off. The Street was well behind it, modelling $14.27 and getting $17.73, a 24% beat. Judge the multiple on what the plants earn across a cycle, not on June.

The five-year record is the cycle in plain sight: a 2022 peak of $28 EPS, a 2024 trough of $10 when cracks collapsed, and net income that swung by three-quarters top to bottom. What did not swing was the share count, and that is the whole equity story.

Marathon has retired 44% of its shares since 2022, from 516 million to about 290 million. That is why 2024 profit two-thirds below 2022 still printed $10 of EPS. The 1H 2026 buybacks went out at an average $246 per share against a stock now at $389, accretive, though richer than the low-$100s they paid in prior years. The mix shifted: buybacks halved from their pace while capex climbed to a record, nearly all of it into MPLX. The variant here is simple. The market is capitalizing a June that will not repeat at 8.6x EV/EBITDA, the 92nd percentile of a twelve-year range, above the 8.1x peers fetch. On mid-cycle earnings you are paying a peak multiple on a peak quarter, and the print that settles it is the Q3 2026 refining capture rate.

Management

No insider has bought a share in twelve months. Fifteen open-market sales totalling $9.5 million went the other way, the largest $2.1 million by Michael Henschen on August 12, with plan status not disclosed in the filings, so read them as officers trimming into a record stock rather than a signal either way. Capital allocation is the real tell, and it is disciplined: $6.13 billion still authorized, dividends of $585 million in the half, and a buyback that slows when the stock is dear rather than chasing it. The one thing to watch is the capex pivot, because every dollar into MPLX is a dollar not retiring stock at these prices.

How it fails or surprises you

Margins mean-revert (downside). June's $5.1 billion was a crack event, not a run-rate. If refining spreads normalize toward mid-cycle, quarterly net income falls back toward the $1.0 to $1.5 billion these plants throw off in an ordinary quarter, and an 8.6x multiple set on peak EBITDA compresses fast. First tell: the Q3 2026 R&M margin per barrel.

MPLX compounds into a re-rate (right tail). The midstream arm now earns fee-based cash that covers the dividend whatever oil does, and the Northwind build widens it. If investors start valuing that stream on its own multiple instead of burying it inside a cyclical refiner, the sum of the parts is worth more than the blended 8.6x today, and the capex the market dislikes is what builds it.

The revenue nobody explained (proof-wrong). Revenue jumped 52% sequentially, from $34.2 billion to $52.0 billion, above even the 2022 super-peak quarters, a move too large to be crack alone. Until the driver is clear, treat the quarter's quality as unproven, because that single line is the fact this read explains least.

Closing thoughts

This is a cyclical priced near the top of its own range on its best quarter in history, so the distribution is skewed by where you are standing in the cycle, not by any secret. The downside is a return to normal cracks, which costs you the multiple and a chunk of the earnings at once. The upside that no print yet pays for is MPLX quietly re-rating the whole structure. The fatter near-term tail is the downside, because you are paying a peak price for peak earnings, and one day after the June quarter it still stands as the unrepeated high-water mark while the stock sits within $10 of its record, so the question of whether that print was the peak is exactly where it was: open, with Q3 the first test.

The bet is still that Americans keep burning fuel and Marathon keeps refining it at a wide spread, with the pipeline arm underneath paying cash through the lean years. What breaks it is a mean reversion in the crack, and the pair that tells you first is the Q3 refining capture rate against a share count that has to keep falling to hold EPS. If June proves the high and the buyback slows while the stock holds near $399, you paid up for a price-taker.