The Oil Inside the Grid: Transformers, AI and Petroleum Supply Chains

A data center can buy electricity from a grid supplied by several different fuels and technologies. That tells us something important about its power supply, but it does not describe everything needed to build the connection, manufacture the equipment or keep the system working. One particularly concrete petroleum connection sits inside the electrical network itself: mineral insulating oil in transformers.

This is a useful place to examine the continuing case for oil. Petroleum serves both as an energy source and as an industrial input. Its value to a modern power system can therefore remain substantial even where crude oil is not the fuel generating the electricity. The strongest argument starts with equipment and supply chains, then asks how much new demand they actually create.

The equipment behind the electricity

Transformers change voltage so electricity can move through transmission and distribution networks. In its July 2024 Large Power Transformer Resilience report, the U.S. Department of Energy maps crude oil and mineral oil among the inputs to the transformer supply chain. The report also explains how insulating oil provides cooling and electrical insulation. This is a physical use of petroleum within the grid, separate from burning fuel at a power station. See the report’s supply-chain discussion and Figure 3. DOE: Large Power Transformer Resilience

The distinction matters. An insulating fluid is part of an equipment system, not a fuel consumed in direct proportion to every kilowatt-hour delivered. A larger electrical buildout can create demand for equipment and its initial fluid fills; operating that equipment does not mean buying the same volume of fresh oil again every year. Treating those two activities as identical would overstate the petroleum requirement.

AI adds pressure to a very physical network

The IEA’s 2026 Key Questions on Energy and AI reports global data-centre electricity consumption of 485 terawatt-hours in 2025 and projects about 950 TWh in 2030. The latter is a forecast, not an observed result. The agency also describes constraints in grid connections and equipment supply chains, including transformers. Projects proposed on paper do not all become completed, fully loaded facilities. IEA: 2026 energy and AI outlook

Those electricity numbers cannot be converted directly into barrels of oil. A defensible petroleum estimate would need the actual equipment mix, fluid specifications, construction schedule, transport distances and fuel choices. Still, the mechanism is clear: where additional load requires new oil-filled equipment and petroleum-fuelled construction or freight, electricity expansion can support petroleum demand outside the power plant’s fuel bill. That is an equipment-level conclusion, not a claim that AI alone determines the global oil market.

Follow the project through three stages

Original illustration for this article. Categories describe different roles, not equal volumes or measured demand.

Construction and delivery come first. The EIA identifies construction vehicles, trucks and trains among diesel applications. For a new substation or power connection, the relevant questions include how the site is prepared, how heavy components arrive and what powers the installation equipment. Petroleum diesel is one established pathway; renewable diesel, biodiesel blends and electrified equipment mean the fuel specification still matters. EIA: Diesel fuel explained

Manufacturing and equipment inputs come next. Mineral transformer oil is a direct example, while petroleum also supplies feedstocks used in chemicals, plastics and synthetic materials. However, a finished polymer should not automatically be counted as entirely crude-oil-derived: feedstock routes differ, including natural-gas-derived inputs. Steelmaking energy, factory electricity and chemical feedstocks also need separate accounting rather than being grouped into a single invented oil requirement. EIA: Petroleum products and their uses

Operation and maintenance are the third stage. Equipment inspection, service transport, lubrication and eventual replacement continue after construction ends. The Bureau of Reclamation’s Transformer Maintenance manual documents oil testing as part of caring for transformer equipment. That supports the importance of managing the installed fluid, not an assumption that every maintenance visit requires a complete refill. Condition, equipment design and the specific maintenance action determine what is consumed. Bureau of Reclamation: Transformer Maintenance

This same way of thinking extends to generating equipment. DOE-supported wind-drivetrain testing found that lubricant chemistry and additives were important to gearbox reliability. The wind turns the turbine; the lubricant performs a different job inside its machinery. That evidence does not establish that every turbine uses the same fluid, that every design has a gearbox or that every lubricant is mineral oil. Those details must come from the equipment specification. DOE: Wind drivetrain testing and lubrication

Alternatives are real, and specifications matter

There is no need to pretend petroleum is the only possible transformer-fluid option. Cargill markets natural and synthetic ester dielectric fluids, including FR3 fluid derived from more than 99% soybean oil. That is manufacturer evidence of an available alternative, not an independent estimate of its market share or a recommendation for any installation. Fluid selection can change the petroleum content of a project materially. Cargill: Dielectric fluid portfolio

For oil-sector analysis, this makes the question more useful: what are utilities and equipment buyers actually ordering? A headline about electricity demand is a starting point. Purchase specifications, new installations, replacement programmes and maintenance practices provide the connection to real product demand. Substitution and longer equipment life can reduce virgin-petroleum requirements, while more construction and a larger equipment fleet can work in the opposite direction.

Dependable supply must also replace declining output

Demand is only half of the discussion. Extracting crude draws down a physical resource, and existing fields do not maintain output indefinitely without continued work. In its September 2025 decline-rate study, the IEA found an average observed annual post-peak decline of 5.6% for conventional oil fields. This is an average across a varied field population, not a universal rate for every well. IEA: Oil and gas field decline rates

The same study estimates that, if capital investment in existing sources of production stopped immediately, global oil output would decline by an average of about 5.5 million barrels per day each year over the following decade. That is a conditional no-investment scenario, not a prediction that the market will lose that volume while companies continue investing. Its significance is the scale of replacement work required behind a seemingly steady production total. IEA: The no-investment supply scenario

Physical depletion and reported proved reserves are different measures. The EIA defines proved reserves in terms of recoverability from known reservoirs under existing economic and operating conditions. New discoveries, improved recovery, revised geological knowledge and changing economics can therefore increase reported reserves even while extraction continues. A rising reserve estimate does not mean previously extracted oil has reappeared underground. EIA: Proved reserves definition

What this means for the oil-market conversation

The practical pro-oil case is that reliable petroleum supply supports both everyday transport and the construction, equipment and maintenance behind a more electricity-intensive economy. Transformer fluid is a small but revealing example: oil’s usefulness cannot be measured only by its share of electricity generation. A dependable industrial system needs the right products in the right places, as well as enough total energy.

That does not settle the price outlook. Oil prices reflect available supply relative to demand, inventories, spare capacity, production costs and the timing of new output, alongside policy and wider economic conditions. Additional infrastructure-related consumption could tighten a market if supply cannot respond quickly; supply growth, efficiency, substitution or weaker activity could offset it. Field decline creates replacement needs, but it does not by itself guarantee an immediate shortage, a higher oil price or a return to an investor.

A useful checklist is therefore physical rather than promotional: which projects are actually being built, which petroleum products do their specifications require, how much is an initial fill versus recurring consumption, and what production must be replaced to keep supplying the wider economy? Answering those questions makes the case for petroleum more credible and makes the limits of the evidence visible.

Editorial note: This article is industry education, not investment or tax advice. Figures retain their source dates; forecasts and conditional scenarios are identified as such. No oil-per-megawatt estimate is asserted. Cover and in-article graphics are original conceptual illustrations created for this article, not photographs of an operating facility or engineering schematics.