The original argument
The Foreign Policy essay by Jason Bordoff and Bob McNally reframes expensive diesel as a shortage of global refining flexibility, rather than simply a shortage of crude oil. Strategic reserves principally release crude, which still needs to pass through a refinery before it becomes diesel, gasoline or jet fuel. The source account of the essay records the loss of nearly 2 million barrels a day of refining capacity across parts of the United States, Europe and Asia since 2020. With many refineries running close to capacity, the remaining cushion is limited; China holds significant spare capacity but has restricted refined-product exports to protect domestic supply.
Against that background, US retail diesel briefly exceeded $6.50 a gallon, while high international refining margins encouraged American refiners to raise production and exports. An export ban appears politically attractive because fuel would initially accumulate at home. The authors acknowledge that prices could fall briefly in some locations, particularly on the Gulf Coast. Their objection concerns what happens next.
A refinery produces a bundle of products
A refinery does not make diesel independently of gasoline and jet fuel. If exports disappear and domestic buyers cannot absorb the output, storage fills and refiners reduce throughput. The initial price relief can therefore be followed by a decline in the supply of several fuels at once. The policy changes production incentives, not merely the destination of an unchanged quantity of diesel.
The longer-term channel runs through investment. If producers expect government to close export markets when prices rise, returns on maintenance, expansion and new capacity become less predictable. Lower refinery utilization also reduces demand for domestic crude and can weaken upstream investment. A measure presented as a way to retain fuel may ultimately discourage both refining and oil production.
The essay invokes the US restrictions and price controls of the 1970s as a warning. Refined-product restrictions were subsequently lifted under the Reagan administration, while the crude-export ban was removed under President Obama and Congress. The historical comparison supports the authors’ argument that supply security depends on production incentives as well as inventories.
Geography and credibility matter
The United States is a major diesel exporter—roughly one-fifth of seaborne trade in the source account—but it is not a frictionless national fuel market. Some coastal regions depend on imports. Diesel retained at a Gulf Coast refinery cannot necessarily reach every domestic market cheaply or quickly; higher world prices may still feed back into import-dependent US regions.
The geopolitical costs would extend to buyers in Mexico, Chile, Ecuador, Peru, Brazil, the United Kingdom and the Netherlands. Those customers would seek alternatives, potentially increasing reliance on Chinese refining capacity. Restrictions could also weaken US objections to other countries’ controls on critical minerals and make buyers and lenders question the durability of American supply commitments. The authors’ wider claim is that energy influence rests on productive capacity, open trade and supplier credibility, not simply on keeping resources inside national borders.
The data center connection
For facilities configured around diesel emergency generation, fuel is a component of the resilience architecture. High availability targets—sometimes expressed as 99.999 percent—depend on more than installed generator capacity. They also require usable on-site inventory, dependable replenishment, equipment readiness and permission to operate. Such targets and designs vary by facility and workload; they are not universal characteristics of AI data centers.
Export restrictions, price controls and emergency allocation can change that architecture’s cost and reliability. The important questions are not limited to the spot price of diesel. Can a facility secure deliveries when transport and refining systems are under pressure? How much fuel should it retain? Do procurement arrangements create firm delivery obligations? In an emergency, how would public authorities prioritize hospitals, military facilities and commercial computing loads?
These questions become more consequential as some AI campuses plan loads in the hundreds of megawatts or at gigawatt scale. A large concentration of backup generation represents contingent fuel demand: ordinarily dormant, but potentially simultaneous during a widespread outage. Installed backup capacity should therefore not be treated as proof that fuel can be delivered to every site at the same time.
The institutional question
The essay offers a caution against static price analysis. A policy can lower a local price today while weakening the production and investment incentives on which future resilience depends. Applied to data centers, this raises a research question rather than an established conclusion: should some AI infrastructure receive critical-infrastructure priority for backup fuel, and, if so, which facilities qualify and who pays for that assurance?
The issue connects fuel-supply agreements, emergency powers, storage requirements and the distribution of scarcity. Any priority granted to a computing facility has an opportunity cost for other users. The relevant comparison is therefore between arrangements for access, cost and risk, not simply between national fuel prices.
About this reading
This commentary brings together a reading of the linked publication and an analysis of its relevance to data center infrastructure. Figures and event descriptions refer to the source’s publication context; the data center implications are identified as analysis rather than attributed findings.