Oracle shares fell 4% on Thursday after Bloomberg reported, citing people familiar with the matter, that the company sent a "force majeure" notice tied to a data center project in New Mexico, according to CNBC's report on the Bloomberg story. The notice is aimed at protecting Oracle from higher expenses on a campus that has not yet come online, and it lands squarely inside the capacity-planning math that cloud and GPU buyers run when they commit to a facility years ahead of workload.
The project in question, dubbed Project Jupiter, is part of the broader Stargate artificial intelligence infrastructure buildout that Oracle is pursuing alongside President Donald Trump, per CNBC. Oracle sent the force majeure notice to Blue Owl Capital, the developer building out the campus, according to the Bloomberg report as relayed by CNBC. The mechanism at play is a contractual one: Oracle is looking to delay payment on the campus if it doesn't come online as expected in 2028, the report says. A force majeure clause is typically invoked to excuse or defer performance under a contract when circumstances outside a party's control make the original terms impossible or newly costly — here, reportedly, to shield Oracle from cost increases rather than to walk away from the deal entirely.
For architects who model capacity around a specific facility and a specific delivery date, this is the mechanics that matter: the 2028 online date is not a soft target, it's the trigger written into the payment structure between the hyperscaler and its developer partner. When that date is at risk, the operator with the compute commitment — Oracle, in this case — has a lever to renegotiate terms rather than simply eat the delay, and that lever is being pulled publicly enough to move the stock four percent in a single session.
CNBC's report frames the New Mexico project as "riddled with setbacks and regulatory hurdles," including local opposition to data centers ahead of the upcoming midterm elections. That is the operational reality behind the contract dispute: this isn't a chip shortage or a power interconnect delay in isolation, it's a build that has run into community and political resistance in the run-up to an election cycle, on top of whatever construction and cost pressures triggered the force majeure notice itself.
None of the reporting carries a dollar figure for the cost increases Oracle is trying to protect itself against, nor a revised timeline if 2028 slips. What's on the record is the 4% share drop, the 2028 target date, the identity of the developer, and the project's place inside the Stargate buildout — not the magnitude of the cost overrun or how long a delay might run. Teams reading this for signal should treat the absence of a number as the signal: a force majeure notice this early, attached to a facility still years from delivery, is a marker of contract stress, not a quantified schedule change.
What breaks in week two for anyone depending on this campus for committed GPU capacity is the assumption that a 2028 delivery date is fixed. A force majeure notice on a data center contract is a renegotiation event, not a cancellation, but it is also not nothing — it is the operator telling its developer, in writing, that the original terms no longer hold. Teams with capacity plans anchored to a single facility and a single delivery year should treat that date as provisional the moment a force majeure notice surfaces, and build a fallback path that doesn't assume Project Jupiter lands on schedule.