The physical market tells a different story
WTI crude fell to $90.34 on Sunday, its lowest since mid-April. Brent dropped to $96.70. Both benchmarks have shed roughly 16% from their late-April peaks as traders price in a US-Iran deal that would reopen the Strait of Hormuz.
But underneath the paper selloff, physical oil is vanishing at a pace never seen before.
The IEA's May Oil Market Report pegged the global inventory draw at 8.5 million barrels per day for the second quarter of 2026. That is not a typo. The agency has never published a number that large. For context, the worst quarter during the 2022 Ukraine crisis saw draws of around 1.4 million bpd.
EIA confirmed the squeeze last week
The U.S. Energy Information Administration reported a 7.86 million barrel draw for the week ending May 15. Analysts had expected 2.5 million. It was the fourth straight weekly decline in U.S. crude stocks.
American inventories now sit at roughly 449 million barrels, down from 457 million just three weeks ago. The five-year average for this time of year is around 470 million.
Those numbers matter because the U.S. is one of the few major consumers that can actually measure its stockpiles weekly. The rest of the world reports monthly, with a lag. When EIA data looks this bad, global figures are usually worse.
How prices can fall while barrels disappear
Futures markets trade expectations, not barrels. Right now, every trader on the planet is betting on some version of the same thesis: the Strait reopens within weeks, Gulf oil flows resume, and the deficit closes before inventories hit critical levels.
That bet has been wrong for three months running. The Strait has stayed shut since February 28. But the market keeps pricing in a resolution because Trump keeps saying one is close.
The result is a widening gap between the paper price and the physical reality. Refiners in Asia are paying steep premiums for delivered cargoes. Spot differentials for non-Gulf grades have blown out. The Atlantic Basin is getting pulled dry as buyers who used to source from the Middle East scramble for alternatives.
What happens if the deal falls apart
This is the scenario nobody on the sell side wants to model. If talks collapse and Hormuz stays closed through June, the math gets ugly fast.
At 8.5 million bpd in draws, global commercial inventories would lose roughly 255 million barrels over a 30-day period. China's strategic reserves, which have been absorbing part of the shock, are finite. Morgan Stanley warned in early May that Beijing's cushion buys time but does not buy a solution.
Aramco's Nasser put it bluntly during a May 13 investor call: without a Hormuz reopening by mid-June, the market faces a supply gap that stretches well into next year. Goldman Sachs modeled a deal-collapse scenario with Brent repricing to $130.
With 14.4 million bpd of Gulf production still offline according to the IEA, the supply hole is deeper today than when WTI was trading above $108 a month ago. Only the expectation has changed.
The backwardation signal
The futures curve tells the same story from a different angle. WTI's July contract trades at $90.34. December sits at $83.42. That $7 spread, a premium for near-term delivery over deferred months, exists because physical barrels are scarce right now regardless of what might happen in sixty days.
If inventories were comfortable, the curve would flatten. It hasn't. The backwardation has narrowed from its April panic levels, but it persists. Traders are willing to pay more for oil they can touch today versus oil that might exist in six months.
What to watch
The next EIA weekly report lands May 28. Another draw above 5 million barrels would put U.S. crude stocks on track to breach 440 million, a level not seen since late 2023. At that point, the gap between falling prices and tightening supply becomes hard for even the most committed deal optimists to ignore.
If Iran signs, the selloff deepens and the inventory crisis resolves itself over weeks. If Iran does not sign, WTI at $90 looks like a gift that the market will not offer twice.