Oracle ($ORCL) and Blue Owl Capital ($OWL): Who Pays When the Power Is Late?

A data center can have a tenant, financing and a very impressive announcement. It can still be unable to earn a dollar if the power is late. The lenders, unfortunately, tend to keep their calendars.

That is what makes Oracle’s Project Jupiter worth examining. AI demand could remain strong while the companies building its infrastructure face an uncomfortable gap between when costs arrive and when revenue begins.

Oracle sells database software, business applications and cloud computing capacity. Its AI expansion requires a much larger physical footprint: buildings, servers, cooling and reliable electricity. Blue Owl Capital manages alternative investments, including capital backing the project’s developer. For investors, the important question is how a construction problem travels through those financial relationships.

Project Jupiter is a planned 2.45-gigawatt campus in New Mexico, supported by an $18 billion construction loan. Bloomberg reported that Oracle sent the developer a force-majeure notice intended to preserve payment relief if the project misses its planned 2028 opening. Oracle says the notice itself establishes no delay; Blue Owl says commitments remain unchanged. Reporting describes potential rent deferral, with other costs still payable, rather than cancellation of the lease.^[1]

The power plan includes a huge methane-fueled fuel-cell installation. Air and water permitting have faced legal challenges. This is a substantial engineering and approval process, with consequences for when the site can operate.^[2]

I find the allocation of those consequences more interesting than another argument about whether AI is a bubble. If the facility cannot serve its customer, revenue may move further out. Financing and other carrying costs may continue. The exact contracts determine whose cash gets squeezed.

We have reported descriptions of those contracts, rather than a complete contractual review. That limits how confidently anyone can assign the eventual loss.

Still, the underlying mechanism matters. AI demand can keep growing while investors lose money if financing costs arrive before the infrastructure can earn revenue.

Oracle’s latest figures show both its funding capacity and its dependence on it. At August 31, cash and equivalents were $36.4 billion. Quarterly operating cash flow was $23.1 billion against $28.5 billion of capital expenditures, leaving negative free cash flow of $5.4 billion. The company also raised $20 billion of common equity before commissions.^[3]

Within operating cash flow, $11.4 billion came from increased deferred revenue associated with customer prepayments carrying a significant financing component. Customer advances are useful funding, but they also come with future service obligations. Investors should examine how much expansion is supported by recurring cash generation, how much by advances, and how much by fresh capital.^[3]

There is real business growth alongside that spending: cloud infrastructure revenue increased 121% to $7.4 billion in the quarter.^[3] The investment question is whether that growth produces attractive cash returns after the infrastructure bill is paid.

The obligations extend well beyond the current construction cycle. Oracle disclosed $288 billion of additional lease commitments at August 31, substantially for data centers, generally commencing between fiscal Q2 2027 and fiscal 2029 and lasting 15–19 years. They appear in the financial-statement footnotes, outside recognized balance-sheet lease liabilities.^[4]

Those multiyear amounts are not directly comparable to present-value debt. Their significance is the duration of the commitment. A long lease provides capacity security, while leaving less flexibility if utilization, pricing or customer economics disappoint.

Private-credit withdrawals add another consideration, but the evidence deserves careful handling. Ares Strategic Income Fund reported repurchase requests for 13.1% of shares. Its disclosed net new requests, however, fell approximately 60% from the preceding quarter. Blackstone’s BCRED estimated net outflows at roughly 3% of NAV, consistent with the previous two quarters.^[5][6]

Those funds face liquidity demands. Their withdrawal figures do not establish insolvency, accelerating new exits or the end of AI infrastructure financing. Nor should investors assume every private-credit portfolio owns the same risks. The borrower, collateral and funding structure matter.

For readers, I would use this situation to tighten the investment test. With $ORCL, follow the conversion from contracted capacity into operating facilities, customer billing and cash. A growing backlog is encouraging; the cost and timing of fulfilling it determine the return.

With $OWL, trace the exposure from the individual project to its investment vehicles and then to the listed manager. A loss borne by fund investors, a reduction in performance fees and a direct corporate balance-sheet loss affect shareholders differently. We need that bridge before attaching a loss estimate to the stock.

The case would improve with resolved power approvals, credible delivery milestones, customer payments that support construction and better cash generation as facilities enter service. It would weaken with repeated delays, higher carrying costs, customer renegotiations or additional funding needs without corresponding revenue progress.

I would keep both companies on a research watchlist at this stage. This evidence alone does not establish an attractive entry price or a short position. A compelling risk story still needs a valuation—and a view of how much concern the market already reflects.

I remain interested in what AI can make possible. The buildout also gives us a practical way to separate ambitious plans from productive assets. Announced gigawatts become economically valuable when they power equipment that customers pay to use. Following that conversion should tell us considerably more than the size of the next announcement.

Sources And Verification

1. Bloomberg, republished by Insurance Journal: Project Jupiter force-majeure notice, September 25, 2026. Contract descriptions are attributed reporting; the complete contract was not reviewed.
2. New Mexico Department of Justice: Project Jupiter permitting challenges, September 3, 2026. Describes the attorney general’s position, not a final judicial determination.
3. Oracle: fiscal Q1 2027 earnings release and financial statements, September 10, 2026. Quarter ended August 31; figures rounded.
4. Oracle: fiscal Q1 2027 Form 10-Q, Note 6, Leases and Other Commitments.
5. Ares Strategic Income Fund: shareholder letter, September 24, 2026. Management-reported figures; requested repurchases are distinct from completed outflows.
6. Blackstone Private Credit Fund: tender update, September 3, 2026. Q3 estimates, subject to finalization.