Where are the Rigs?
“The Stone Age didn’t end for lack of stone, and the Oil Age will end long before the world runs out of oil” - Ahmed Zaki Yamani, Minister for Oil for Saudi Arabia
Gas prices in the US are nearly $5 per gallon, the SPR is reaching rock-bottom levels, and the Strait of Hormuz remains closed, yet the response to US oil production remains muted. What gives?
Since the start of the war, Brent has reached $120/bbl, and WTI has risen to $110/bbl. The Hormuz disruption had an even more pronounced effect on products, resulting in a vertiginous rise in jet fuel crack spreads to $80/bbl, vs the 2022 peak of $60/bbl, as Amos pointed out yesterday
To buttress the disruption of oil and product flows, the US in conjunction with the IEA plan to release 400m in equivalent barrels, 172m of which will come from the US. Since March 20th, the US has released nearly 41.3 million barrels, culminating in the largest weekly draw on record.
Now, the sustained rise in US gasoline and oil prices elicits questions around how the domestic oil industry will respond. So far this year, gasoline stocks fell from 259 million barrels to 214 million barrels, while refinery capacity utilization rose from its February trough of 88% to just under 92%. Anecdotally, at least, the US’s midstream capacity is running at near full capacity in the short run. Dwindling product inventories are also complicated by the fact that most of the US’s refinery capacity is generally suited to process heavier grades from Canadian mineral sands and Venezuela, while the vast majority of upstream production is light.
Against a backdrop of falling product inventories, a depleting SPR, and higher prices, one would expect rig counts as recorded by Baker Hughes to increase meaningfully. They aren’t, and as of May 15, Baker Hughes reports that US oil rigs sit at 415—up six since January, but down fifty year-over-year. The Permian, the most productive basin in the world, has shed 36 rigs against last year to 246. Texas is off 26 rigs YoY. Eagle Ford, DJ, and Williston are flat to down.
Interestingly, while rigs have remained relatively flat this year, the Frac Spread Count—the measure of fracking crews operating in the US—increased from 145 in February to 184 for the week ending on May 15. With this increase in crews, wouldn’t that presage more rigs?
At a high level, the relationship between WTI spot prices and oil rigs is well understood, but over the last several years, the correlation has evolved. A simple time-series model (specifically, an autoregressive distributed lag model with structural breaks) that predicts the US rig count using its lags and WTI prices can quantify this evolution. The model identifies five distinct regimes in the WTI elasticity. In the post-2014 recovery (2016–17), a sustained 1% increase in prompt WTI lifted the US rig count by roughly 0.055% over the impact horizon — and close to 1.9% in the long run, after accounting for the autoregressive persistence in rigs. That sensitivity has since compressed to ~0.022% in the short run, about 40% of its 2016–17 peak, reflecting the industry's shift in technology preferences and capex discipline. Put differently, over the long run, a 10% increase in oil prices would historically have driven a ~19% increase in the US rig count, versus only ~6% today.
Higher oil prices lead to more rig counts, but the decision to deploy a rig depends on a whole host of issues related to the shape of the futures curve, a given producer’s breakeven cost, general economic certainty, labor conditions, etc. Notably, since the onset of the war, WTI has experienced extreme backwardation, which has not necessarily incentivized oil production. Instead, a meaningful steepening in contango would draw rigs in on the expectation of higher demand relative to supply.
Another key development since 2022 is the rate of horizontal drilling and well efficiency. Between 2014 and 2024, the share of horizontal drilling rigs rose from 10% to 22%, and average lateral well lengths doubled, reducing the need to deploy additional rigs to produce the same number of barrels. This trend is generally reflected in increased productivity in the Permian, Bakken, and Appalachian basins.
Lastly, a counterargument is that the supply of drilled-but-uncompleted wells enables producers to tap into DUCs, which would take 3-6 weeks, compared with a few months for a rig to drill and complete from start to finish. So far this year, US DUC inventories have declined from 5,021 to 4,972, while field production has remained flat at around 13.7-13.8 million barrels per day. However, the decrease in DUCs remains marginal and is thus a weakly held proposition.







One might consider that everything in AI land is built on projections. That includes data centres. Ive got a niggly feeling that speculation ie buying and selling of financial products in lieu of the growth of IA is driving this. It's better to spot the ACTUAL financing and construction of data centres. Ive noticed a lot of delays recently. Some say AI is a bubble, not quite dutch tulip status but still, i increasingly get the feeling it might be so.
Commodity cycles usually end from incentive destruction before resource depletion.
Years of underinvestment eventually tighten spare capacity, raise geopolitical sensitivity, and increase volatility transmission across the entire inflation complex.
Energy markets are still heavily shaped by prior capex decisions.