Key Highlights
- Oracle sent a Force Majeure notice to the developer of Project Jupiter, its largest AI Data Center campus, preserving the ability to defer payments if the site misses its 2028 target.
- The notice was tied to delays in a gas pipeline needed to power the New Mexico site, not to any decision by Oracle to cancel or scale back the project.
- Oracle separately reaffirmed a 2.4 GW fuel cell order with Bloom Energy tied to the same campus, indicating the contractual move was a hedge rather than a signal of project failure.
- The episode adds to growing evidence that power delivery, not chip supply, has become the binding constraint on how quickly new AI data centers can be brought online.
What Oracle's Notice Actually Says
Oracle sent a force majeure notice to the developer behind Project Jupiter, its Stargate-branded AI data center campus in New Mexico valued at roughly USD 165 billion, giving itself the contractual option to defer certain payments if the site does not meet its 2028 target. Force majeure clauses exist precisely for situations like this: they allow a party to pause or adjust its obligations when circumstances beyond its control, in this case a delayed gas pipeline needed to power the campus, threaten to disrupt a project's schedule. Sending the notice is not the same as terminating the underlying agreement or abandoning the project, and Oracle underscored that distinction by separately reaffirming its 2.4 GW fuel cell order with Bloom Energy tied to the same site around the same time. Read together, the two moves look less like a company walking away from a commitment and more like one protecting its financial flexibility while the timeline issue plays out.
The Power Wall Behind the AI Buildout
Project Jupiter's pipeline delay is a specific instance of a broader pattern reshaping how AI infrastructure gets built. For the earlier phase of the AI buildout, chip Supply was the primary constraint on how quickly new compute capacity could be deployed. That bottleneck is increasingly shifting toward power: gas pipeline construction, permitting timelines, and Utility interconnection queues now frequently take longer than the data center buildings and hardware they are meant to serve. Companies across the power infrastructure supply chain are racing to Fill that gap through behind-the-meter generation, including Atlas Energy Solutions' pivot from frac sand Mining into gigawatt-scale power generation and Bloom Energy's large-scale fuel cell deployments. Oracle's force majeure notice is a visible, contractual acknowledgment of exactly this dynamic playing out at one of the largest AI campuses under construction.
What It Means for Oracle and the Sector
For Oracle, preserving the option to defer payments provides financial flexibility while the pipeline and permitting issues at Project Jupiter are resolved, without requiring the company to renegotiate its core supply agreements or walk away from its Stargate commitments. For the broader sector, the episode is a useful reminder that contractual hedges like force majeure notices are risk management tools, not default signals, and should be read in the context of a company's other actions, such as Oracle's parallel reaffirmation of the Bloom Energy fuel cell order. Other hyperscalers and data center developers facing similar infrastructure bottlenecks are likely watching how this specific situation resolves, since the underlying power constraints, gas pipeline capacity and permitting speed in particular, are not unique to New Mexico or to Oracle's project alone.
Where Things Stand
The data center boom is hitting a power wall in a real, if partial, sense: the constraint on Project Jupiter is not compute, capital, or customer demand, but the physical infrastructure needed to deliver electricity to the site on schedule. Oracle's force majeure notice reflects that reality without signaling retreat, as shown by its simultaneous reaffirmation of the 2.4 GW Bloom Energy fuel cell order. The broader lesson for the sector is that power delivery timelines, not chip availability, now sit at the center of how fast new AI infrastructure can actually come online.

_11_13_2026_19_06_43_490830.jpg)




Please wait processing your request...