Oracle Stock Is 54% Below Its High. The Last Time It Fell That Far, Getting Back Took More Than a Decade.

Oracle shares closed at $158.78, 54% below their $345.72 record high, as investors weigh an AI infrastructure spending surge against sharply negative free cash flow. Fiscal 2026 revenue rose 17% to $67.4B and cloud infrastructure revenue jumped 77% to $18.1B, while remaining performance obligations increased to $638B from $138B. However, capex surged 162% to $55.7B, exceeding operating cash flow by about $24B and driving free cash flow to negative $23.7B; management still targets roughly $90B of fiscal 2027 revenue and $8.05 in adjusted EPS.

Analysis

The key underwriting issue is not demand visibility but conversion of contracted AI capacity into acceptable returns on invested capital. ORCL is effectively shifting from a software-margin model toward a capital-intensive infrastructure model; a sustained gap between buildout spending and customer utilization would pressure both FCF credibility and the multiple. The market will likely focus less on reported backlog growth than on contract duration, customer concentration, prepayment terms, cancellation rights, and whether incremental cloud gross margin is improving as new capacity comes online.

Near term, the Sept. 10 earnings event can produce a sharp relief rally if management quantifies financing capacity and reiterates the path from negative FCF to self-funded expansion. Over the next 1-3 months, however, debt issuance, lease obligations, or weaker-than-expected capex discipline could keep ORCL trading like a leveraged AI infrastructure proxy rather than a defensive enterprise-software compounder. MSFT, AMZN and GOOGL are relatively advantaged if enterprise buyers prioritize multi-cloud flexibility, while NVDA benefits only if Oracle's deployment pace remains intact; a financing-led slowdown would be a second-order negative for AI server, networking and data-center supply chains.

Consensus may be too anchored to the historical valuation comparison. A lower earnings multiple does not eliminate risk when the earnings base excludes the economic burden of a rapidly expanding asset base; the relevant comparison is forward return on incremental invested capital versus hyperscaler alternatives. Conversely, the selloff may be overdone if contracted capacity is substantially prepaid and customer commitments are investment-grade, because that would reduce the feared funding and utilization tail risk.

Key Decisions for Investors

• Remain neutral-to-underweight ORCL into earnings; do not add directionally until management discloses customer concentration, contract cancellation protections, and a credible 12-24 month FCF inflection. A post-results move higher without improved capex-to-revenue or capex-to-bookings conversion is a potential short-entry setup.
• Use a defined-risk bearish event structure: buy ORCL 1-2 month put spreads, funded only partially with lower-strike puts, rather than outright short equity. Thesis is a downside repricing if financing needs or cloud-margin timing disappoint; exit if guidance is reaffirmed with materially lower forward capex intensity or meaningful customer prepayments.
• For AI exposure, prefer long NVDA versus short ORCL in a 1-3 month pair only after earnings if Oracle indicates capacity additions remain robust but its FCF recovery is deferred. This isolates supplier demand from ORCL's balance-sheet and execution risk; cover if Oracle demonstrates accelerating cloud gross margin and reduced external funding needs.
• Set a diligence alert on incremental debt, finance leases and working-capital movements at the earnings release. A large increase in financing commitments without corresponding disclosed contracted cash receipts would falsify the benign-backlog interpretation and supports a more aggressive ORCL underweight.