Oracle weighs Apollo and Goldman deal to fund a large AI chip purchase as industry debt rises

Oracle is reportedly in talks with Apollo and Goldman Sachs to arrange financing for a major purchase of AI chips, as overall debt levels in the market continue to climb. The arrangement, as described in a recent market report, would likely involve investors funding a distinct company that buys the chips and then leases them to Oracle, shifting part of the capital burden away from Oracle’s balance sheet.

The report frames the proposal as a financing structure rather than a direct sale-and-purchase arrangement. In that model, the lender-backed purchaser would hold the chips and monetize them through lease payments. For Oracle, the practical appeal would be access to computing hardware for AI workloads without tying up as much of its own capital up front, though the precise economics and risks would depend on the lease terms and chip delivery schedule.

Apollo, known for credit and investment strategies across corporate lending and structured finance, and Goldman Sachs, which provides investment-banking services and can participate in structured financing, are named as potential backers. The report does not specify the size of the transaction Oracle would be seeking from lenders, beyond characterizing the broader credit market as having debt above a very large threshold.

The same report points to a market context in which debt is already at elevated levels. While that backdrop can help explain why financiers are looking for new assets to underwrite, it also raises the importance of deal specifics such as interest rates, collateral protections, and the lease contract’s duration. For an AI chip purchase, contract duration matters because chip procurement is tightly linked to rapidly changing hardware roadmaps.

Oracle’s interest in AI-related infrastructure has been consistent with the broader industry trend toward building data center capacity for training and inference workloads. Still, the company has not, in the materials provided here, disclosed any definitive deal terms, the chip brands or models, the expected delivery timetable, or the internal end-use configuration for the financed assets.

Companies in the AI infrastructure supply chain often face a timing challenge: training cycles and model rollouts can require hardware sooner than some corporate procurement cycles allow. Financing structures such as asset purchases held by a special-purpose entity and then leased back can address that timing, but they can also add complexity, including operational dependencies on the lessor and potential constraints if hardware is upgraded or swapped.

As of now, key details remain unreported. The market report does not confirm whether Oracle and these financiers have reached agreement, nor does it provide valuation assumptions, interest-rate spreads, or a breakdown of how much of the cost would be debt versus equity. It also does not disclose whether Oracle would have an option to purchase the chips at lease end, nor what would happen if chip supply is delayed or if demand shifts faster than expected.

What to watch next is whether Oracle issues a clarification through investor channels, or whether the proposed structure appears in filings or deal announcements from the participating finance firms. Investors and industry observers will likely focus on whether the arrangement is purely financial, such as a leaseback, or if it also includes commitments tied to future capacity, procurement volume, or service-level performance for AI deployments.