Oracle Stock Is Down 60% From Its High. Here's Why I'm Still Not Buying.

Oracle shares are down nearly 60% from their September 2025 high despite OCI revenue surging 121% year over year to $7.4B and remaining performance obligations reaching $664B. The concern is financing: Oracle plans $95B of capex this year, generated negative $5.4B of quarterly free cash flow, and had $125B of long-term debt plus $288B of off-balance-sheet data-center lease commitments. A potential delay at the Project Jupiter campus could leave Oracle liable for lease payments before related AI-compute revenue begins, amplifying downside if AI demand or construction execution weakens.

Analysis

ORCL is being valued less like a software company and more like a highly levered infrastructure developer, but without the balance-sheet flexibility of MSFT, AMZN, GOOG, or META. The key earnings risk is not reported cloud growth; it is the conversion of contracted backlog into powered, revenue-producing capacity before fixed lease, interest, and depreciation charges begin. A mismatch between construction timing and customer deployment would create negative operating leverage, forcing either further debt/equity issuance or a capex reset that undermines the AI-growth multiple.

The non-obvious spillover is to Oracle's financing counterparties and data-center ecosystem. Higher perceived execution risk can widen ORCL credit spreads, raising the hurdle rate on incremental facilities and potentially slowing orders for suppliers such as VRT, ETN, PWR and data-center REIT/developers with concentrated Oracle exposure. Conversely, MSFT and GOOG gain competitively if enterprise customers view OCI capacity as less reliable or if Oracle must prioritize cash preservation over aggressive GPU availability and pricing.

Near-term sentiment is likely already impaired, so the next 1-3 month catalyst is not another large contract announcement but evidence on cash conversion: operating cash flow, capex guidance, committed financing, power-delivery milestones, and whether customer prepayments offset construction exposure. Over 6-18 months, the thesis is falsified if OCI margins expand while free-cash-flow burn narrows materially despite elevated build spending, demonstrating that utilization and contractual economics—not merely backlog—support the capital structure. The contrarian case is that a deeply discounted ORCL may re-rate sharply if it secures non-recourse project financing and converts a small portion of backlog into live capacity; that outcome requires verification rather than reliance on headline RPO.

Key Decisions for Investors

• Initiate a 3-6 month relative-value position: short ORCL / long MSFT in beta-neutral dollars. The pair isolates Oracle's financing and execution exposure while retaining AI infrastructure demand; reassess if ORCL reports improving free-cash-flow conversion or narrows credit spreads materially versus large-cap software.
• For directional exposure, use ORCL 6-9 month put spreads rather than outright short stock after any contract-driven rally. Define the trade around a 10-15% premium-funded downside structure; the principal risk is a financing announcement or capacity-go-live milestone triggering a sharp short-covering re-rating.
• Avoid treating backlog as deployable revenue until management discloses customer concentration, cancellation rights, prepayment terms, facility-level completion dates, and expected utilization. Make these disclosures a watch-item before adding to either bearish or contrarian long exposure.
• Monitor ORCL bond spreads and data-center construction milestones over the next two quarters. A sustained spread widening alongside unchanged capex guidance would strengthen the short thesis; successful refinancing, delayed lease obligations, or capex moderation without cloud-growth deceleration would reduce it.
• Maintain relative preference for MSFT and GOOG over ORCL within AI infrastructure allocations for the next 6-18 months: both can fund capacity internally and can absorb deployment delays without the same dilution or refinancing sensitivity.