Oracle Trucks Natural Gas to Temporarily Power AI Data Centers
nashnova research
Oracle is hauling compressed natural gas by truck to power multiple AI data centers, a stopgap already running for over a year in Utah — at roughly four times pipeline cost. AI compute demand is outrunning the infrastructure meant to feed it.
Why truck gas to a data center?
Oracle needs power at several AI data centers before the gas pipelines feeding them are built.
The workaround — compressing natural gas, loading it onto trucks, and driving it to the site — has been running at a facility outside Salt Lake City, Utah for over a year. The same method is in use at a Texas campus being built for OpenAI.
This means → the pipeline can't keep up, so the truck becomes a makeshift fuel line. Start the machines now; let the pipe catch up later.
What went wrong with Project Jupiter?
Project Jupiter in New Mexico is one of Oracle's flagship data center builds, with a planned total capacity of 2.45 gigawatts.
A gas pipeline meant to serve the site was forced to reroute for regulatory reasons. The original summer start date has slipped to next year.
Oracle has issued a force-majeure notice to the developer — a contractual move that could shield it from some delay-related payments.
After the news broke, Oracle shares fell roughly 5.5%. This reflects market anxiety over whether its AI buildout timeline can hold.
How expensive and inefficient is trucked gas?
Energy consultancy East Daley Analytics estimates trucked gas costs about four times the price at a major pipeline hub, once labor, specialized equipment, and truck fuel are factored in.
The logistics chain: extract gas from a pipeline → compress and load → drive to site → decompress into generators — around the clock, without pause.
In plain terms = even powering just 100 megawatts of Project Jupiter — only 4% of its planned capacity — a single large trailer carries enough gas for roughly 40 minutes of operation.
This means → trucking can bridge a gap, but it cannot sustain large-scale, long-term operations.
Who is making money on this stopgap?
The power bottleneck at AI data centers has spawned a niche of "virtual pipeline" service providers — truck fleets standing in for unfinished pipes.
Certarus (a Superior Plus subsidiary) supplies the Utah site; VoltaGrid handles the Texas project.
Oracle publicly praised VoltaGrid for delivering "cost-effective power solutions." Certarus said lengthening timelines for grid and pipeline connections have "created urgent demand for rapidly deployable, scalable energy solutions."
What does this tell us about the bigger picture?
Oracle's free cash flow is currently negative and is expected to stay that way until more data centers come online.
Whether trucked gas can sustain Project Jupiter's early-stage operations remains a key variable for the market.
This reflects a broader reality: AI compute demand is running ahead of power infrastructure. "Pay a premium to switch on now, wait for the cheap solution to catch up" is becoming an industry pattern.
市场有风险，内容仅供研究参考，不构成投资建议。