To meet its AI data center delivery deadlines, Oracle has resorted to unconventional power supply methods to overcome pipeline delays, though high costs and scalability constraints remain concerns.

According to a Bloomberg report on October 8, Oracle is using trucks to transport compressed natural gas to maintain the construction and operation progress of multiple AI data centers, circumventing time losses caused by pipeline infrastructure delays.

This unconventional strategy has already been implemented in Utah and Texas and is being considered for a key project in New Mexico. Oracle's free cash flow is currently negative and is expected to remain under pressure until more AI data centers are completed and operational.

Oracle shares fell 5.5% on Thursday to close at $135.69, marking their largest single-day drop since July 16.

Meanwhile, shares of Bloom Energy, which has a fuel cell power partnership with Oracle, plunged more than 6% the same day, reflecting growing market concerns over the pace of Oracle's data center expansion and the prospects for related partnerships.

Trucking gas has precedent: Utah and Texas led the way. This is not Oracle's first attempt. According to people familiar with the matter cited in the report, at a data center project on the outskirts of Salt Lake City, Oracle has used trucked gas to maintain normal progress for over a year, with services provided by Certarus, a subsidiary of Superior Plus. Certarus stated that the timelines for data centers to connect to pipelines and the grid continue to lengthen, "creating growing demand for quickly deployable, flexibly scalable energy solutions." In Shackelford County, Texas, at a data center campus being built for OpenAI, Oracle also adopted this approach, with supplier VoltaGrid LLC. Oracle wrote on social media: Thanks to the VoltaGrid team for finding a cost-effective power supply solution to ensure customers can access OCI (Oracle Cloud Infrastructure) computing power on time. The most concentrated pressure point is New Mexico. As mentioned by Wall Street CN, the data center project codenamed "Project Jupiter" faces severe pipeline delays. Pipeline operator Energy Transfer LP was forced to reroute after its original route was rejected by New Mexico regulators, with the expected commissioning time postponed from this summer to next year. To protect the project timeline, Oracle is considering delivering compressed natural gas to the data center by truck, allowing the facility to begin early-stage operations before the pipeline is officially connected. The urgency is already evident: Oracle issued a force majeure notice to the project developer last month, which could exempt the company from some payment obligations if further delays persist.

High costs and scale limits: not a long-term solution. Trucking gas is not a cheap option. According to estimates by Jack Weixel, senior energy analysis director at energy analysis firm East Daley Analytics, factoring in labor, specialized equipment, and vehicle fuel costs, the price of compressed natural gas delivered by truck is about four times the price of natural gas at major pipeline hubs. Natural gas must first be extracted from pipelines, compressed and loaded onto trucks, transported for several hours, then decompressed and injected into generators—a process that means round-the-clock truck dispatching for continuously operating data centers. Scale bottlenecks cannot be ignored either. According to calculations by Ellie Holbrook, an analyst at energy analysis firm SemiAnalysis, if Oracle used compressed natural gas trucks to provide 100 megawatts of power for "Project Jupiter," that would only account for about 4% of the project's total planned capacity of 2.45 gigawatts. Each large truck can only support about 40 minutes of power supply, meaning dozens of truckloads would be needed per day for uninterrupted delivery. For this reason, compressed natural gas trucking has previously been limited mainly to remote industrial scenarios such as mining and oil drilling, far from grid connection points. But for Oracle, which has made commitments to customers like OpenAI and urgently needs to deliver computing power on time, this may be the least costly transitional solution under current conditions.