Prediction: Oracle's $638 Billion Cloud Infrastructure Backlog Could Make It One of the Best-Performing AI Stocks Through 2028

Oracle reported fiscal 2026 revenue of $67.4 billion, up 17%, as cloud infrastructure revenue surged 77% to $18.1 billion; its remaining performance obligations reached $638 billion, up $500 billion year over year. Management expects 12% of RPO to convert within 12 months and 34% over the following two years, supporting projected fiscal 2027 revenue growth of 34% to more than $90 billion and improving gross margins. The article argues that prepaid and customer-provided-hardware contracts should ease capital-spending pressure, while EPS could accelerate from an estimated 5% gain in fiscal 2027 to 43% growth by fiscal 2029; Oracle trades at 19x forward earnings versus 25x for the Nasdaq-100.

Analysis

ORCL’s valuation debate is shifting from demand visibility to financing quality and execution cadence. The critical variable is not the headline backlog but how much is tied to creditworthy counterparties, minimum-spend commitments, cancellation rights, and customer-funded hardware; a backlog that converts with upfront cash materially reduces the bear case around incremental leverage and free-cash-flow deficits. The first investable proof point is likely quarterly operating cash flow and capex intensity, not reported revenue, over the next 1-3 quarters.

Competitive dynamics are more nuanced than a generic AI-infrastructure shortage. ORCL can win workloads where database proximity, sovereignty, and bespoke capacity matter, but its largest hyperscale rivals—MSFT, AMZN, and GOOGL—retain distribution advantages and can use cloud credits or bundled software to pressure pricing. If Oracle’s buildout is genuine incremental capacity rather than displaced hyperscaler demand, NVDA, AVGO and networking/power suppliers benefit; if customer-owned equipment becomes a larger mix, ORCL may improve cash conversion while shifting some hardware procurement economics to customers.

Consensus may be too focused on a multi-year earnings endpoint and too dismissive of the intervening balance-sheet risk. A rerating requires evidence that gross margin expands despite GPU depreciation, power costs, and any customer concentration; revenue recognition alone will not resolve concerns if capex remains elevated. Falsify a constructive view if capex-to-revenue fails to decline over the next four quarters, operating cash flow does not inflect despite conversion, or management reduces cloud growth/remaining-obligation recognition timing.

Key Decisions for Investors

• Initiate a staged long ORCL position over the next 1-3 months, sized below a core mega-cap allocation; add only after the next two reports show improving operating cash flow and no deterioration in cloud gross-margin commentary. Target a 12-18 month rerating from de-risked funding needs; exit on a material capex increase without corresponding prepaid commitments or a cloud-growth guidance cut.
• Express the idiosyncratic thesis as long ORCL / short IGV or a basket of MSFT, AMZN and GOOGL, 6-12 months. This isolates Oracle’s potential execution-driven multiple recovery from broad AI/software beta; close if Oracle’s cloud growth premium versus hyperscalers narrows for two consecutive quarters.
• For defined downside, use a 9-12 month ORCL call spread rather than outright long-dated calls: buy moderately out-of-the-money calls and sell calls near a valuation level consistent with peer software multiples. The setup only works if cash-flow conversion validates the backlog before time decay becomes material.
• Set an event alert around quarterly disclosure of prepaid/customer-funded capacity, debt issuance, and free-cash-flow guidance. If these disclosures remain qualitative rather than quantified, treat the backlog narrative as unverified and avoid increasing exposure.