The U.S. Energy Information Administration reported that the front-month futures price of Brent crude oil traded in a wide range in the second quarter of 2026, reaching a high of $118 per barrel on April 29 and falling to a low of $72 per barrel on June 26. That $46 swing in eight weeks, driven by disruptions to international crude oil and petroleum product flows through the Strait of Hormuz according to the U.S. Energy Information Administration, turned diesel backup generation from an insurance line into a margin question for data centers that sized standby capacity to support AI workloads without hedging the fuel underneath.
I track this because power is the binding constraint in AI infrastructure, and the bottleneck migrates. For the past eighteen months the conversation has centered on grid interconnection queues, substation lead times, and utility politics. Diesel generators appeared in capital plans as a bridge technology, a way to light up racks before utility power arrived or to carry load during maintenance windows. The implicit assumption was that fuel cost did not matter because runtime would remain negligible. The Hormuz disruptions in the second quarter of 2026 tested that assumption and found it incomplete.
My read is that the $72 to $118 per barrel range in Brent crude translates directly into diesel costs for standby generation, hitting facilities that banked on cheap backup power without building fuel-cost hedges into their operating models. The U.S. Energy Information Administration noted that uncertainty around reopening the Strait of Hormuz to shipping traffic contributed to highly volatile prices in April and May, with an average daily price swing of $4 per barrel in the Brent crude oil price, compared with $1 per barrel in the same months in 2025. When daily swings quadruple and the absolute range doubles in a quarter, any facility running diesel generators for more than emergency backup starts to see fuel expense move from a footnote to a line worth managing. For AI data centers that leaned on diesel to smooth power gaps or defer grid upgrades, that shift arrived faster than the operating models anticipated.
Density Drove Runtime Up
AI training and inference clusters pack more compute into less floor space than traditional workloads, which means more watts per rack and higher instantaneous draw when those racks light up. Utilities and grid operators set interconnection schedules measured in quarters or years, not weeks. Diesel generators became the gap filler, allowing operators to energize capacity ahead of permanent grid connections or to cover maintenance windows without idling revenue-generating hardware.
That operational choice turned diesel from pure emergency backup into a semi-regular fuel input. The U.S. Energy Information Administration reported that the price of Brent crude oil began the quarter above $100 per barrel, as disruptions to international crude oil flows through the Strait of Hormuz reduced access to crude oil for much of the world and led many countries in the Middle East to shut in crude oil production. A facility planning for diesel runtime of one percent annually can absorb fuel-cost volatility as a rounding error. A facility running diesel generators ten or twenty hours per month to bridge grid shortfalls or demand spikes starts to care about the difference between $72 and $118 per barrel, because that difference flows straight through to the cost of keeping the lights on.
The margin impact depends on two variables: runtime and hedge coverage. Hyperscalers with deep treasury functions and energy-trading desks can lock in fuel costs months ahead or swap exposure into fixed-price contracts. Colocation providers and smaller regional operators often lack that capability, leaving them exposed to spot diesel prices that track crude with a lag. When Brent moves $4 per barrel per day for two months, spot diesel prices follow, and any facility buying fuel week to week sees its energy cost swing in real time.
Supply Shock Became Hedging Lesson
The Strait of Hormuz disruptions created a natural experiment in fuel-cost exposure. The U.S. Energy Information Administration noted that from May 18 to June 17, negotiated ceasefires and growing market anticipation for the resumption of shipping traffic through the Strait of Hormuz led the Brent crude oil price to decline by an average of more than $1 per barrel per day. On June 17, the United States and Iran signed a Memorandum of Understanding that, among other things, sought to resume traffic through the Strait of Hormuz, according to the U.S. Energy Information Administration. The price relief that followed confirmed that the April and May spike was a supply shock, not a demand-driven structural shift.
For data-center operators, the lesson is not that crude will stay high or stay low. An unhedged fuel line can move faster than quarterly budgets accommodate. Diesel generators exist because grid power is not perfectly reliable and interconnection timelines are not perfectly predictable. The cost of that insurance used to be stable enough to ignore. The second quarter of 2026 demonstrated that stability is a function of geopolitical calm and open shipping lanes, neither of which carries a guarantee.
The practical response is to treat diesel fuel as a hedgeable input, the same way utilities hedge natural gas for peaker plants and airlines hedge jet fuel. That means either locking in forward contracts for diesel delivery, buying crude futures or swaps to offset price risk, or shifting backup generation toward fuels with more stable local pricing, such as natural gas where pipeline access permits it. Each option carries cost and complexity, but the alternative is accepting that a two-month supply disruption halfway around the world can move your energy expense by double-digit percentages.
The Counterargument
Most hyperscale facilities run on grid power with diesel generators serving only as emergency backup, so fuel-cost volatility affects capital planning and insurance reserves more than quarterly operating expense. If runtime remains under one percent annually, even a doubled diesel price moves the needle by basis points, not percentage points. The Hormuz disruptions hit during a period when many AI-focused data centers were still ramping, meaning diesel runtime was elevated as a transitional condition rather than a steady-state operating mode.
Once grid interconnections complete and utility power becomes the primary source, diesel generators return to true standby duty, and fuel cost becomes a contingency line rather than a recurring expense. In that scenario, the second quarter of 2026 represents a one-time shock during a buildout phase, not a structural shift in operating economics. The hedge I am arguing for would then be overkill, adding cost and complexity to manage a risk that only materializes during rare supply disruptions and temporary reliance on backup generation.
The weakness in that counterargument is that AI workloads are not static. Training runs and inference clusters scale up and down with customer demand, model releases, and competitive pressure. Diesel generators provide the flexibility to light up capacity fast without waiting for utility approvals or grid upgrades. As long as speed to market and workload elasticity remain competitive advantages, diesel will see more runtime than traditional emergency-backup assumptions predict. Whether operators price that runtime into their cost structure or treat it as a variable they can ignore until the next supply shock arrives is the open question.
Fuel Diversification Gets Attention
The Hormuz episode also raises the question of fuel diversification. Diesel generators dominate data-center backup because they are reliable, widely available, and compatible with existing permitting and safety frameworks. Natural gas offers a lower-carbon alternative with more stable regional pricing in markets with pipeline access, but it requires different infrastructure and often faces longer permitting timelines. On-site generation using natural gas peakers or combined heat and power systems can reduce reliance on diesel, but those projects take quarters to design and build, and they shift the operating model from backup power to distributed generation.
I expect the next six months to show whether operators treat the second quarter as an anomaly or as a signal to diversify fuel sources. If diesel hedging disclosures start appearing in earnings calls or if natural gas peaker projects get announced at major campuses, that would confirm that fuel-cost volatility has moved from a footnote to a planning priority. If those signals do not appear, the market is betting that grid interconnections will finish fast enough and diesel runtime will drop low enough that fuel cost remains immaterial.
The U.S. Energy Information Administration also reported that China's crude oil imports fell in the second quarter, noting that China, the world's largest importer of crude oil, imported less crude oil in the second quarter of 2026 following higher crude oil prices that resulted from disrupted flows through the Strait of Hormuz. China's lower imports reduced global demand, softening the upward price effects from the disrupted supply through the strait, according to the U.S. Energy Information Administration. That demand response illustrates how quickly high prices can curb consumption in price-sensitive markets, but data centers do not have the same demand elasticity. A training cluster or inference endpoint cannot simply shut down when fuel gets expensive; it either runs or it does not. That inelasticity is why fuel-cost exposure matters more for AI infrastructure than for industries that can defer or reduce consumption when prices spike.
On My Watchlist
Diesel fuel hedging disclosures or energy-cost line items in third-quarter 2026 hyperscaler and colocation earnings calls, running from August through October 2026. If operators call out fuel expense or announce hedging programs, that confirms the margin impact was real and the response is underway.
Announced natural gas peaker or on-site generation projects at major campuses over the next six months, signaling a shift from diesel-only backup toward fuel diversification. Permit filings, equipment orders, or partnership announcements with gas utilities would all count.
Utility rate-case filings or power-purchase agreement amendments citing fuel-cost pass-through provisions through year-end 2026. If utilities or large customers start embedding fuel-cost adjustments into contracts, that indicates the market is pricing volatility into long-term agreements rather than treating it as a one-time event.