Oracle Uses Trucked CNG to Bridge Pipeline Delays at AI Data Center Projects
Редакция ЦОД:Главное
Oracle is using truck-delivered compressed natural gas (CNG) to support AI data center construction and early operations where pipeline infrastructure has been delayed. The approach has reportedly been used in Utah and Texas and is under consideration for Project Jupiter in New Mexico.
- Oracle has reportedly used trucked CNG in Utah and Texas to mitigate pipeline-related constraints on AI data center development.
- Truck-delivered CNG is being considered, rather than deployed, for Project Jupiter in New Mexico.
- Project Jupiter's stated 2.45 GW figure is total planned capacity, not commissioned capacity.
- S1 estimates that 100 MW supplied through trucked CNG would equal about 4% of Project Jupiter's planned capacity.
- Road-based CNG supply may support limited early load, but it is materially more expensive and operationally difficult to scale than pipeline supply.
Oracle has reportedly turned to trucked compressed natural gas to maintain construction schedules and support early operations at AI data center developments affected by pipeline delays.
According to S1, the model has been used for more than a year at a data center project in the Salt Lake City suburbs, with Certarus, a Superior Plus company, involved in supply. In Shackelford County, Texas, VoltaGrid LLC is supplying the approach for an Oracle campus being built for OpenAI.
Oracle is also considering trucked CNG for Project Jupiter in New Mexico as a temporary option ahead of pipeline availability. Energy Transfer LP has had to redesign the pipeline route after New Mexico regulators rejected the original alignment. Trucked gas is described as a measure under consideration, not as an operating supply arrangement or commissioned capacity at the project.
S1 describes Project Jupiter as having 2.45 GW of total planned capacity. It estimates that 100 MW of power supplied through trucked CNG would represent about 4% of that planned total. This comparison highlights the constraint of using road-based fuel logistics for a large-scale data center campus: it may support an initial, limited load, but does not provide a scalable pathway to full planned capacity.
The economic burden is also material. S1 estimates that truck-delivered CNG can cost about four times the price of natural gas at major pipeline hubs. It further estimates that one large truck would support roughly 40 minutes of electricity supply, implying a sustained and operationally intensive transport requirement. These are third-party estimates cited by S1 rather than confirmed Oracle operating data.
For AI data center developers and infrastructure investors, the situation could reinforce the value of projects with secured gas, grid and interconnection capacity before major deployment commitments. Temporary autonomous or semi-mobile power arrangements may help bring selected loads online earlier, but their economics and logistics could increase delivery risk and the cost of supplying compute capacity at large campuses.
The case illustrates how gas and grid infrastructure constraints can directly affect AI data center delivery schedules. Trucked gas may enable limited early-load operation, but its cost, continuous logistics requirements and limited scalability make it a temporary bridge rather than a substitute for permanent pipeline and grid connections.
- — / 9 октября 2026 г. · КитайскийПервоисточник →
Материал подготовлен редакцией ЦОД: Главное на основе указанных первоисточников.