Down 55% From Its High, Is Oracle a Buy?

Oracle’s fiscal 2026 revenue rose 17% to $67B+ and cloud infrastructure revenue jumped 75% YoY, with backlog reaching $638B for strong revenue visibility. However, aggressive AI data-center build-outs drove free cash flow sharply negative and increased debt, contributing to the stock’s ~55% decline from its $345 peak. With the stock around $145 (roughly $200 below the all-time high) and a ~17x forward P/E plus sub-1 PEG, the article frames the repricing as an attractive entry point despite capex/FCF and demand risks.

Analysis

ORCL is no longer being traded as a pure software compounder; the market is pricing it as a capital-intensive AI landlord with equity-duration risk. That matters because the upside case now depends less on demand quality and more on whether management can convert backlog into cash before financing costs and depreciation eat the spread. In that setup, NVDA is the cleaner beneficiary: every incremental Oracle buildout cycle translates into GPU/server demand without forcing NVDA to warehousing-like balance-sheet expansion.

The real second-order loser is Oracle’s own equity multiple if investors keep demanding proof of free-cash-flow discipline. A debt-funded buildout can still be rational, but only if utilization ramps fast enough; otherwise the stock behaves like a levered call option on AI demand with a long maturity and negative carry. Over the next 1-3 months, the key catalyst is not revenue growth but guidance on capex moderation, debt trajectory, and whether management can stop the market from treating backlog as non-monetizable. Over 6-18 months, sustained high utilization would support a re-rating; failure would invite multiple compression even if top-line growth stays strong.

Contrarian view: consensus is assuming backlog equals visibility, but backlog is only valuable if the implied margins survive the funding cycle. The market may be underestimating how quickly capital intensity can force ORCL to sacrifice buybacks, balance-sheet flexibility, and ultimately pricing power versus better-capitalized cloud peers. Falsifiers are straightforward: a visible FCF inflection, capex plateau, or leverage stabilization would reduce the bear case quickly; absent that, ORCL remains more of a credit story than an easy long equity.

Key Decisions for Investors

• Prefer NVDA over ORCL for 1-3 months: long NVDA / short ORCL as a relative-value expression on AI capex monetization. Risk/reward improves if Oracle keeps funding buildout with debt while NVDA captures the spend without balance-sheet drag.
• Do not chase ORCL on a bounce; if the stock rallies 8-10% without an accompanying improvement in free-cash-flow guidance, consider a starter short or 1-2 month put spread. Falsifier: next update shows capex flattening and FCF turning positive.
• If you need AI exposure but want lower equity risk, prefer Oracle credit over Oracle equity and monitor spread behavior for entry. The thesis is that debt gets paid off by backlog before common equity gets paid by FCF.
• Set an earnings-alert on ORCL for capex and leverage commentary over the next 1-2 quarters. If management signals capex discipline and utilization remains high, reverse bias and cover shorts; if not, expect multiple compression to continue.