This AI Infrastructure Company Has a $638 Billion Backlog and Is Trading Near an 18-Month Low

Oracle touts a $638B remaining performance obligation (RPO), up 363% YoY, with $67B in new AI infrastructure contracts signed in the latest quarter. Over half the backlog is tied to OpenAI, including a reported $300B+ (5-year, starting 2027) deal, while only ~12% is expected to convert to revenue within a year. The key risks flagged are customer ability to fund commitments and Oracle’s funding needs, with FY capex estimated at $95B requiring ~$40B new capital and pushing long-term debt above $100B as free cash flow turns negative. At ~18x forward earnings and with guidance for 28% revenue growth (midpoint) in the current quarter, the article frames the setup as an attractive risk-reward if execution holds.

Analysis

Oracle should be viewed less as a classic software re-rating and more as a financing-sensitive AI infrastructure proxy. The market is discounting that a large share of the economic value now depends on two balance sheets staying open: Oracle’s and its largest customer’s. That makes ORCL equity behave like a levered call option on AI demand, with upside if utilization ramps quickly, but disproportionate downside if funding or execution slips.

The cleaner beneficiaries are the upstream picks-and-shovels names. NVDA, AVGO, and AMD monetize the buildout earlier in the cycle, with less working-capital strain and less refinancing risk, while AMZN can absorb demand through a broader cloud mix. Oracle, by contrast, faces a classic margin paradox: revenue can grow fast while free cash flow stays negative because capex arrives before cash receipts, which can compress the equity multiple even if the top line looks strong.

The key catalyst window is the next 1-3 months, when the market will care more about financing terms, customer fundraising, and any change to capex guidance than the headline backlog. Over 6-18 months, the real test is ROIC and utilization; if capacity lands below expectations, the market will likely reprice ORCL as a debt-funded buildout rather than a durable growth compounder. The contrarian miss is that the backlog itself is not the asset — the asset is access to capital and the ability to turn reserved capacity into cash without destroying margins.

Key Decisions for Investors

• Prefer a 3-6 month pair trade: long NVDA or AVGO vs short ORCL. The thesis is that AI spend accrues first to the chip/systems layer, while ORCL carries the funding and execution overhang. Falsify this if ORCL issues debt on attractive terms and reaffirms capex without a spread blowout.
• Avoid chasing ORCL on the first bounce; if initiating, do it only after the next financing/capex update confirms no further balance-sheet surprise. Use small size given the asymmetric downside if customer funding weakens.
• For investors already long ORCL, hedge with a put spread into the next earnings or financing catalyst. The setup is a classic event-driven risk: upside may be gradual, while a funding miss can re-rate the stock quickly.
• If looking for cleaner AI infra exposure, rotate incrementally from ORCL into NVDA/AVGO on any ORCL strength. That expresses the view that the market is overpaying for the financing risk embedded in the backlog story.
• Set an alert on ORCL credit conditions and capex commentary; if debt markets demand materially wider spreads or management trims expansion plans, that is the first sign the equity case is breaking.