Oracle Just Posted Its Best Quarter in 15 Years -- Here's Why 2026 Could Get Even Better

Adjusted revenue and EPS both grew at least 20% YoY in Oracle's Q3 FY2026 (ended Feb. 28), with cloud infrastructure revenue up 84% YoY and remaining performance obligations increasing by $30B to $553B. Trailing-12-month free cash flow is negative $24.7B due to heavy AI-related capex on chips and data centers, but management cites a 60% reduction in chip-to-revenue lead time and a 'halo effect' as AI customers adopt additional services. Analysts model operating profit rising from $25B in FY2025 to $46B by FY2028; the stock trades around 20x FY2027 EPS and remains over 52% below its 52-week high, suggesting potential upside if investments convert to recurring cloud cash flows.

Analysis

Oracle’s AI-driven data center build creates an asymmetric moat: owning the stack (infrastructure + apps) lets it harvest higher lifetime value per customer versus best-of-breed vendors, and that stickiness compounds as customers consolidate procurement. The more interesting second‑order winners are chipset and systems suppliers who can stay ahead of lead times — GPU makers and board/system integrators will see outsized order visibility, while legacy CPU suppliers face a mix of incremental demand and margin pressure as Oracle prioritizes the highest-throughput silicon.

The main operational risk is cadence mismatch between heavy up‑front hardware commitments and the timing of enterprise AI deployments; a modest slowdown in customer model training budgets or an erosion in AI model pricing could push the cash conversion inflection well beyond one year. Macro variables matter: a sustained rise in real yields or tighter vendor financing would amplify funding costs for data center rollouts and could force either slower capacity growth or unanticipated asset-light deals with third parties.

Practical trade construction should respect both the optionality of the growth story and the short-term cashflow noise: use time‑spread and collar structures to capture asymmetric upside while limiting drawdown from further capex skepticism. Monitor three near-term readouts as trade triggers — equipment lead times, contracted ARR cadence from large customers, and any shift in lease/financing terms — which will signal whether capex is being converted into recurring cash faster or slower than the market assumes.

Contrarian view: the market is pricing the capex as permanent profit destruction rather than a temporary investment hump; if Oracle’s supply‑chain improvements keep realization lag to under 12 months, the company can re‑leverage margins faster than consensus models that assume multi‑year payback. The mirror risk is a capacity glut or aggressive price competition from hyperscalers that would compress returns on those very same data centers; set hard stop-losses tied to contracted revenue growth, not headline spend narratives.

Key Decisions for Investors

• Long ORCL equity exposure (size 2–3% of portfolio) on any pullback: use Jan-2028 LEAP calls ~30–45 delta to capture multi-year re‑rating, hedge with selling near-term calls or buying 9–12 month puts to cap downside; target 30–50% return in 12–24 months, cut loss at 20% of option premium.
• Pair trade — long ORCL / short a large hyperscaler (1:1 notional, size 1–2%): run 6–12 months to play margin and share gains from full‑stack adoption; exit if hyperscaler gross margins widen or Oracle’s contracted ARR growth decelerates for two consecutive quarters.
• Directional hardware play: accumulate NVDA (size 1–2%) or buy 6–12 month NVDA calls to play sustained GPU demand from Oracle’s buildouts; trim into strength and watch shipment lead times as exit signal.
• Capital‑structure hedge: buy 12 month out-of-the-money puts on ORCL equal to ~0.5–1% portfolio risk or construct a collar if holding stock — protects against funding/market‑rate shocks that would delay cash‑flow conversion beyond 18–24 months.