top of page

Oracle Force Majeure Notice Turns Stargate's 2028 Deadline Into a Financial Test

2 hours ago
14 min read

Oracle sent a force majeure notice covering Project Jupiter, placing new contractual pressure behind the New Mexico Stargate campus's 2028 delivery target. The Oracle force majeure notice does not cancel the development or establish that it will be late. It preserves Oracle's ability to seek payment relief if infrastructure problems prevent the enormous AI campus from opening on schedule.

That distinction matters because Project Jupiter remains under construction while its power system faces unresolved regulatory and delivery questions. Oracle says the campus remains on schedule. Yet the notice gives the company protection against a scenario that its public messaging continues to reject.

Project Jupiter is not a speculative site on an internal planning map. It is a flagship part of Stargate, the infrastructure program supporting OpenAI's expanding demand for computing capacity. STACK Infrastructure is developing the campus with backing from Blue Owl Capital, while Oracle plans to serve as its main tenant.

The immediate conflict is therefore not Oracle against OpenAI or Oracle against its developer. It is Oracle's confident 2028 commitment against the physical, regulatory, and financial conditions required to keep that commitment.

What the Oracle Force Majeure Notice Actually Changes

The notice preserves Oracle's contractual defenses without withdrawing its commitment to Project Jupiter.

Oracle delivered the notice to STACK Infrastructure, the Blue Owl-backed developer responsible for the New Mexico campus. The original force majeure reporting described the action as protection against payment obligations if the facility misses its planned 2028 opening.

Force majeure is a contract provision covering events outside a party's reasonable control. Depending on the contract, it can suspend duties, extend deadlines, or excuse certain performance failures. Invoking the provision does not automatically prove that a qualifying event occurred.

Oracle also is not attempting to leave the project, according to people familiar with the arrangement. It remains the campus's main tenant and continues to describe Project Jupiter as an active, on-schedule development.

That makes the timing more consequential. Companies usually preserve contractual rights before a disputed event becomes irreversible, not after every consequence has already arrived. Oracle is creating a formal record while major power and permitting questions remain unsettled.

An Oracle spokesperson said such notices are common in developments of this scale. The company said they preserve contractual rights and do not independently establish a delay or change delivery expectations.

Blue Owl offered a similarly calm interpretation. A spokesperson said the parties remain aligned and that the notice does not change their financial commitments to the multiyear project.

Those statements narrow the meaning of the event, but they do not make it irrelevant. A routine contractual safeguard can still identify which risk the parties are preparing to allocate.

According to financing disclosures, Oracle could receive a three-year rent deferral if the parties agree that a qualifying power-related event occurred. Oracle would reportedly continue paying other costs and later owe rent across the full lease term.

That structure would shift timing risk rather than erase the bill. The developer and its lenders could wait longer for rent while still carrying construction and financing exposure. Oracle would avoid paying for capacity it cannot use.

It remains uncertain whether the notice will produce that outcome. The contractual conditions are private, and the public reporting does not establish that STACK accepted Oracle's position. A notice preserves a claim; it does not settle one.

The development also has not entered default because of the notice, according to the reporting. Construction can continue, permits can progress, and the parties can maintain their commercial relationship.

The change is more precise. Project Jupiter's delivery risk now has an explicit contractual channel connecting unfinished power infrastructure to Oracle's payment schedule.

That connection turns a local permitting dispute into a material question for technology investors. It also explains why a notice that changes no current deadline still moved Oracle and Blue Owl shares.

Project Jupiter's Power Plan Is the Real Deadline

Buildings alone cannot satisfy Oracle's Stargate commitment because the campus needs an unprecedented amount of dependable energy.

Project Jupiter is planned for Doña Ana County in southern New Mexico. Its proposed power system would include 2,275 methane-fueled Bloom Energy fuel cells generating approximately 2.46 gigawatts.

That is closer to the output of a major regional power system than a conventional corporate facility. The campus cannot simply connect a few buildings and declare the project operational. Servers, cooling equipment, networking systems, and supporting infrastructure require continuous power at extraordinary scale.

Oracle and Yucca Growth Infrastructure revised the energy plan during 2026. They replaced an earlier turbine and diesel-centered proposal with a fuel-cell microgrid, an on-site system designed to produce power independently from the wider grid.

Oracle says the revised design would reduce nitrogen oxide emissions by about 92 percent compared with the January application. It also says the fuel cells would use minimal water after an initial startup fill.

Those claims address two sources of local concern, but the new plan introduced its own dependency. Thousands of fuel cells require a large and reliable natural gas supply.

The proposed Green Chile pipeline is intended to deliver that fuel. Energy Transfer rerouted the project after New Mexico's State Land Office rejected rights-of-way across state trust land. The pipeline's expected service date then moved from August 2026 to February 2027.

That date still precedes the campus's planned opening. However, it reduces the buffer available for commissioning, testing, construction problems, and further regulatory delays.

The air-quality process adds another dependency. The New Mexico Environment Department is reviewing the microgrid under a separate permit proceeding. Its public permit docket includes the application, revisions, modeling materials, draft permit documents, and hearing notices.

Oracle emphasizes that the data center buildings and the microgrid are separate facilities on adjacent sites. The campus construction underway today does not require the microgrid's air-quality permit, according to the company.

That legal separation supports continued building work, but it does not eliminate the operational link. A completed data hall has limited value if its planned energy system cannot run when computing equipment arrives.

The permitting process has also faced procedural disputes. New Mexico Attorney General Raúl Torrez supported challenges alleging that state agencies compressed review timelines and limited opportunities for discovery.

The attorney general's September filing described the proposed fuel-cell installation as an unprecedented concentration of the technology. It also cited more than 100 million gallons of groundwater pumping in a separate water dispute.

These claims come from a litigating state office and should not be treated as a final ruling. They nevertheless demonstrate that opposition is operating through formal legal processes, not only public protests.

New Mexico's Supreme Court temporarily interrupted parts of the permitting process before lifting the relevant stays in September. The air review resumed, but the litigation showed how quickly procedural conflict can affect the schedule.

Oracle's updated energy design might ultimately receive approval. The pipeline might also enter service during 2027 without compromising the 2028 target. Neither outcome was settled when Oracle delivered its notice.

This is the core reversal behind the story. The company publicly maintains that Project Jupiter is on schedule while contractually preparing for a power-related delay.

That is not necessarily contradictory. Executives must keep projects moving while legal teams protect the company against adverse scenarios. However, the combination provides a clearer measure of how Oracle now assesses the risk.

The decisive milestone is no longer how quickly visible buildings rise near Santa Teresa. It is whether the project can complete its permitting, fuel delivery, commissioning, and power integration in the remaining window.

The 2028 Commitment Now Carries Financing Pressure

Any delivery slippage would move costs and revenue across a financing structure built around timely rent and usable computing capacity.

Project Jupiter is supported by approximately $18 billion in construction financing from roughly 20 banks, according to people familiar with the transaction. That makes the campus part of a much wider test for debt-funded AI infrastructure.

The debt reportedly traded below 90 cents on the dollar around the force majeure disclosure. Such pricing does not prove that lenders expect default. It indicates that investors demand a discount for uncertainty surrounding repayment, timing, or project execution.

The force majeure notice did not trigger a financing default, according to reporting. Yet it gave debt investors another reason to examine the relationship between the tenant lease, construction schedule, power agreements, and loan documents.

A delayed campus can transmit pressure through each layer. Developers must keep funding work. Lenders wait longer for the expected cash flow. Oracle might defer rent. OpenAI waits for computing capacity that supports future model development and commercial services.

Oracle's position is different from that of a developer. It wants access to operational capacity, not an unfinished property. Paying full rent before the site can deliver contracted computing infrastructure would leave Oracle absorbing a delay it did not create.

STACK and Blue Owl face the inverse problem. They are responsible for delivering a highly specialized campus whose economics depend on a major tenant. A long rent deferral could extend the period before the development generates its expected contracted income.

This is why the Oracle force majeure notice matters even if every party continues supporting the project. It identifies who might bear the carrying cost if the schedule slips.

Public-market investors responded quickly. Oracle shares closed 3.5 percent lower on September 24 after recovering from larger losses earlier in the session. Blue Owl shares ended 3.6 percent lower.

Daily share moves have many causes, so those declines cannot be assigned entirely to Project Jupiter. They still show how sensitive investors have become to evidence that AI infrastructure revenue might arrive later than expected.

William Blair analyst Sebastien Naji said the near-term effect on Oracle appeared limited. His broader concern was that data center delays could postpone revenue recognition while investors evaluate Oracle's delivery capacity.

The distinction between total demand and revenue timing is essential. OpenAI can need more computing power while Oracle still struggles to deliver it on the schedule embedded in contracts and forecasts.

Oracle has accumulated substantial commitments associated with future cloud infrastructure. Those agreements become revenue only when capacity is delivered under the relevant terms. Construction progress cannot replace a functioning service.

The financing market must therefore evaluate two different propositions. Demand for AI computing remains strong, but the infrastructure serving that demand can still arrive late.

That separation challenges an assumption underlying the AI construction boom. A long-term customer contract reduces demand risk, but it does not remove permitting, power, equipment, labor, or community risk.

It can also concentrate exposure. A developer building for one dominant tenant has less flexibility if that tenant invokes contractual protections. A tenant relying on a few enormous campuses has fewer substitutes if one project falls behind.

The same pattern reaches lenders. Project financing often depends on detailed agreements that assign specific risks to developers, operators, suppliers, and customers. If one contract permits payment delays, lenders must understand whether other agreements still cover interest and construction costs.

Oracle's notice tells the market to read those documents more carefully. Headline commitments do not reveal when payments begin, which events suspend them, or who carries costs during an interruption.

That scrutiny will extend beyond New Mexico. Banks have financed multiple large AI campuses, while technology companies continue signing long-duration power and capacity agreements.

Future transactions could demand larger contingency reserves, stronger completion guarantees, or higher borrowing spreads. Developers might also seek narrower force majeure language after seeing how power disputes affect tenant obligations.

None of those changes requires Project Jupiter to fail. Contract markets often reprice risk after a warning rather than an actual loss.

Stargate's Promise Is Meeting Local Infrastructure Reality

Stargate can secure customers and capital nationally, but every campus still depends on local permits, utilities, land, and public consent.

OpenAI launched Stargate in January 2025 with Oracle, SoftBank, and MGX as initial equity funders. The initiative initially described an intention to invest up to $500 billion in American AI infrastructure over four years.

OpenAI and SoftBank took the lead roles, while Oracle became a central technology and infrastructure partner. The program began in Texas and expanded toward additional campuses, including the New Mexico site.

The original Stargate announcement presented the buildout as a national competitiveness and economic-development program. Project Jupiter now illustrates the difference between announcing aggregate capacity and delivering one specific campus.

At a national level, gigawatts can look like entries in a portfolio. On the ground, each gigawatt requires generation equipment, fuel, transmission, water planning, construction labor, environmental review, and community agreements.

New Mexico also expects material benefits. Oracle said in July that Project Jupiter had generated nearly $80 million in state and county tax revenue. It reported nearly 700 New Mexico residents working at the site.

The company projected more than 7,000 construction jobs and 1,500 ongoing project-supported roles after completion. It also estimated more than $4.7 billion in long-term economic impact from taxes, investments, and related activity.

Those figures come from Oracle and its supporting economic analysis. They remain projections rather than guaranteed outcomes. The company acknowledges in its materials that estimates should not be treated as precise measurements.

Oracle has also committed funding for local infrastructure and services. Its community projections include support for water infrastructure, schools, county services, and workforce programs.

Supporters see those commitments as evidence that the development can produce benefits before operations begin. Construction employment and tax receipts provide a tangible argument for keeping the project moving.

Critics focus on a different accounting. They question the proposed gas consumption, emissions, water effects, permitting process, and public obligations surrounding the campus.

The debate creates pressure on both Oracle and public officials. Accelerated approvals invite allegations that agencies are favoring a politically important corporation. Longer reviews strengthen procedural legitimacy but reduce the schedule buffer.

Project Jupiter's microgrid was designed partly to avoid waiting for enough conventional grid capacity. That approach promised faster energy deployment, yet it transferred dependencies to fuel cells, air permitting, and a dedicated gas pipeline.

This tradeoff is becoming common across the AI data center industry. On-site generation can bypass one bottleneck while creating new ones involving fuel access, emissions, equipment performance, and local acceptance.

Oracle's notice shows that those issues have reached the contract layer. Community opposition is no longer merely a communications challenge for the developer. It can affect the timing of rent, revenue, and debt performance.

The comparison with Stargate's Texas campus is instructive. The first Abilene phase entered operation on Oracle Cloud Infrastructure in 2025. Its progress demonstrated that large AI campuses can move from plans into production.

Abilene does not guarantee the same outcome in New Mexico. The sites have different developers, energy arrangements, permits, and political environments. Project Jupiter must satisfy its own chain of conditions.

There is also no evidence that Oracle's notice reduces OpenAI's underlying need for computing capacity. OpenAI said in 2026 that its planned American infrastructure already exceeded half of its 10-gigawatt target for 2029.

The risk concerns delivery concentration and timing. If a flagship campus misses its window, OpenAI and Oracle must find substitute capacity, delay workloads, or accept a slower infrastructure ramp.

Competitors face similar physical constraints. Microsoft, Google, Amazon, and Meta are all pursuing large data center and energy programs. Their ability to spend does not grant automatic access to ready power.

Project Jupiter could therefore become a precedent regardless of its ultimate opening date. An on-time launch would support the argument that private microgrids can accelerate AI infrastructure despite difficult permitting.

A significant delay would support the opposite conclusion. It would show that bypassing conventional utility timelines does not bypass environmental review, fuel logistics, or community resistance.

The Notice Protects Oracle, but It Cannot Produce Power

Contract language can redistribute losses from a delay, but it cannot complete a pipeline, approve a permit, or energize computing equipment.

This limit is the skeptical angle investors should keep in view. Oracle's legal preparation improves its negotiating position, but it does not solve Project Jupiter's operational dependencies.

A force majeure clause also does not automatically cover every regulatory setback. The parties must evaluate the exact language, notice requirements, causation standards, mitigation duties, and events excluded from protection.

A developer could argue that predictable permitting difficulties are ordinary project risks. Oracle could respond that state decisions or legal interruptions fall outside its reasonable control.

Public information cannot resolve that dispute because the relevant contract is private. Reports describing a possible three-year rent delay explain the potential remedy, not its guaranteed availability.

The boundaries matter because the project changed its power plan after development began. Questions could arise about whether later design choices contributed to the conditions behind any delay.

Oracle and STACK currently present a united front. Neither has publicly accused the other of causing a failure, and both say the project remains on track.

That alignment can continue even while the parties preserve competing rights. Large projects often require cooperation during construction and clear contractual positions if a milestone later slips.

The greater uncertainty concerns the amount of contingency built into the schedule. A February 2027 pipeline service date leaves time before 2028, but commissioning a multigigawatt system is not a single-day event.

Fuel cells must arrive, be installed, connect to gas infrastructure, pass testing, and integrate with the campus. Data halls must also receive computing hardware, networking equipment, cooling systems, and operating approvals.

The reported deadline is described broadly as a 2028 opening target. Public sources do not provide enough detail to determine which buildings, power blocks, or capacity levels must be available first.

That ambiguity can conceal partial success. Project Jupiter could begin operating in stages while later sections remain under construction. Whether partial capacity satisfies Oracle's lease depends on terms that are not public.

The reported force majeure position also focuses attention on payment timing rather than total project economics. Deferring rent would not remove expenses already incurred by Oracle, STACK, Blue Owl, or lenders.

Oracle would still need computing capacity somewhere. The developer would still manage construction obligations. Banks would still monitor collateral and projected cash flow.

A delay could also raise equipment questions. AI accelerators improve quickly, so hardware selected years before opening can lose relative value before installation.

The facility itself has a longer useful life and can receive newer systems. However, postponed capacity can complicate procurement schedules, deployment plans, and expected returns.

It would be equally misleading to conclude that the notice predicts cancellation. Oracle has strong incentives to complete the campus, and construction has already produced jobs, spending, and tax revenue.

The project also has institutional support from county officials and business partners. Its scale makes rerouting all intended workloads to alternative sites difficult.

The most defensible conclusion sits between reassurance and panic. Oracle has not abandoned Project Jupiter, but it has formally prepared for a scenario in which the promised delivery date collides with unresolved power infrastructure.

That is a more meaningful signal than either party's public optimism alone. It reveals where the commercial consequences will appear if technical and regulatory work falls behind.

Three Signals Will Decide Whether Project Jupiter Stays on Track

The air permit, pipeline schedule, and project debt will show whether Oracle's notice remains precautionary or becomes economically active.

The first signal is New Mexico's decision on the microgrid's air-quality permit. Approval would remove a central regulatory obstacle, although opponents could continue challenging the process or its outcome.

A denial, major revision, or renewed legal interruption would weaken Oracle's schedule. It would also strengthen the argument that the force majeure notice responded to a concrete delivery threat.

The second signal is measurable progress on the Green Chile pipeline. Investors should watch for final federal authorization, construction milestones, and confirmation of its February 2027 service target.

If the pipeline advances on time, Project Jupiter retains a meaningful commissioning window before 2028. Another delay would compress that window and increase the importance of alternative power arrangements.

Oracle has indicated that grid power could provide another path under some conditions. Any formal change in the energy plan would deserve close attention because it could alter both timing and project economics.

The third signal is the price and treatment of Project Jupiter's debt. Stable or improving loan prices would suggest lenders view the notice as manageable contract administration.

Further deterioration would indicate that investors see rising completion or cash-flow risk. New AI infrastructure loans could provide another comparison through their interest spreads, protections, and completion guarantees.

Oracle's own disclosures also matter. Investors should listen for changes in capital spending, cloud capacity delivery, remaining performance obligations, and the timing of revenue from major AI contracts.

The company should not be judged only by whether it repeats that Project Jupiter remains on schedule. More useful evidence would include energized capacity, completed power blocks, equipment installation, and customer-ready service dates.

OpenAI's response will matter as well. Additional Stargate capacity at other locations could represent ordinary portfolio expansion, or it could provide insurance against delays in New Mexico.

Readers should resist treating every new site as evidence that Project Jupiter has weakened. The stronger signal would be an explicit shift in workload plans or a revised New Mexico delivery commitment.

The Oracle force majeure notice has already clarified one important point. AI infrastructure risk does not end when a customer signs a contract and banks provide construction financing.

The difficult work happens between commitment and operation. Power must arrive, permits must survive scrutiny, equipment must work, and communities must accept the terms of development.

Project Jupiter remains capable of meeting its target. Oracle and Blue Owl both say the commercial commitments remain intact, and neither has announced a revised opening schedule.

Yet the project's 2028 promise now has a documented fallback. Oracle has protected itself against paying rent for capacity that cannot come online as expected.

Watch the permit decision first, the pipeline second, and the debt market third. Together, those signals will show whether the notice stays in a legal file or begins reshaping Stargate's economics.

Give every agent the context to do better work

Connect your agents to the knowledge, decisions, and history already organized in remio.

remio currently supports Windows 10+ (x64) and Macs with Apple silicon.

Your AI Partner at Work
Get more done with remio

Plan. Create. Deliver.
All in one place.

bottom of page