Oracle Has $638 Billion of Contracted Backlog and a $430 Billion Stock

Oracle ended fiscal 2026 with $638B of remaining performance obligations, up 363% YoY, but only expects 12% of the backlog to convert to revenue within 12 months. While cloud infrastructure revenue accelerated to as high as 93% YoY in the year, free cash flow swung to -$23.7B as the company built out AI data centers, funding the gap via $43B of net debt plus a planned ~$40B more in fiscal 2027 (including a $20B at-the-market equity program). The article frames the stock’s valuation as supported by backlog (revenue growth guided up ~34% in fiscal 2027) but warns that AI contract economics and the balance-sheet strain are still unproven.

Analysis

Oracle is increasingly behaving like a financed infrastructure intermediary rather than a pure software compounder. That shifts the economic rent away from equity holders: the AI demand is real, but the incremental dollars are likely shared with GPU vendors, power/electrical contractors, and lenders before they reach residual cash flow. In that setup, NVDA is the cleaner beneficiary of the capex cycle, while ORCL becomes a lower-quality way to express AI demand because it carries execution, utilization, and funding risk.

The key market mechanism is not revenue growth but return on capital. If Oracle keeps funding buildout with debt and equity, the stock can stay cheap for longer even if backlog converts as promised; what matters is whether gross margin and FCF improve faster than financing costs. Near term, the stock will trade with every update on capex, financing, and any sign that AI customers are prepaying enough to de-risk the balance sheet; over 1-3 months, spread widening or another equity raise would be a negative catalyst, while evidence of better-than-feared unit economics would be the main upside surprise.

The contrarian point is that backlog size alone may be a valuation trap: the market is not disputing demand, it is discounting dilution and low incremental ROIC. A 6-18 month bull case only works if Oracle proves that data-center assets can be monetized at high utilization and that financing needs plateau; otherwise, each added contract can expand revenue but compress equity value. That argues for treating ORCL as a show-me story, not a cheap AI proxy.

Key Decisions for Investors

• Prefer NVDA over ORCL on a 3-6 month horizon: long NVDA / short ORCL pair to isolate the difference between selling picks-and-shovels and funding the data-center buildout. Risk: if Oracle proves unusually high-margin AI infrastructure economics, the short leg can squeeze.
• If initiating ORCL exposure, wait for financing-driven weakness rather than chasing strength; look for any bounce toward post-results highs as a better short-entry zone. Thesis is invalidated if management meaningfully raises FCF guidance or cuts external funding needs.
• Watch ORCL credit spreads and any new debt/equity filing as the next catalyst set; a widening in credit while equity holds would be a bearish confirmation for the next 1-3 months.
• Use ORCL as a sell-the-rip name rather than a core long until the company shows positive free cash flow after capex. If the stock loses the market's tolerance for dilution, downside can re-rate faster than revenue growth supports it.
• For more conservative accounts, stay out of ORCL and express AI demand through NVDA instead; it captures the hardware spend without the balance-sheet overhang.