Down 21% This Year: Is Oracle the Most Undervalued AI Stock on the Market?

Oracle has pivoted to an AI hyperscaler strategy and raised fiscal-year capex to $50 billion; FCF was approximately negative $44 billion through the first three quarters of fiscal 2026 and long-term debt rose to about $148 billion (nearly 50% higher vs end-2025). Management raised financing via $25 billion of debt and $25 billion of equity/convertible issuance, project backlog surged 325% YoY to $553 billion, and revenue guidance is ~$67 billion for fiscal 2026, ramping to ~$90 billion in fiscal 2027. The stock is down ~21% YTD — the plan offers potential first-mover advantages in AI data centers but carries significant execution and balance-sheet risk.

Analysis

Oracle’s pivot to owning and operating bespoke AI-scale facilities changes competitive dynamics beyond software vs. cloud. The immediate second-order winners are GPU and accelerator vendors (greater multi‑year unit demand and contractual pricing power) and industrial power/thermal contractors (long-cycle, high-margin build projects). Conversely, traditional colocation landlords and some enterprise SaaS partners face displacement risk on large, bespoke AI workloads that prefer vertically-integrated stacks.

Credit and capital structure are now an active part of the equity thesis rather than background noise. The market should trade the equity as a levered project developer: visible near-term revenue cadence will be driven by conversion of multi-year backlog into cash, while downside is dominated by execution and refinancing risk if macro rates and chip supply lines tighten. Key catalysts that will reprice risk are quarterly backlog conversion rates, announced multi-year chip/power supply agreements, and any sovereign contract cancellations or wins.

Near-term (weeks–quarters) the stock is sensitive to operational miss and any headlines around anchor-customer concentration; medium-term (6–24 months) it’s binary on delivery economics and FCF normalization. That makes structured, asymmetric exposure preferable to naked long equity: you want exposure to upside from successful onboarding and pricing power, but protection against capex overruns, bond‑market repricing, or anchor-client churn. Monitor corporate credit spreads and GPU supply agreements as leading indicators — they will lead equity re-rates by one to three quarters.

Key Decisions for Investors

• Asymmetric equity: Buy ORCL 9–12 month call spreads (buy 1yr ~ATM call, sell 1yr +25–35% OTM call) to capture a multi‑quarter conversion rally while funding premium with the sold call; target 20–30% upside vs 3–5% premium paid (roughly 4–8x payoff if backlog converts).
• Relative value pair: Long NVDA 9–12 month 20% OTM calls financed by selling ORCL 9–12 month 10% OTM calls (delta‑hedged). Rationale: capture secular GPU demand upside while hedging against project execution risk in the Oracle build‑out; unwind if NVDA/accelerator lead indicators (fab allocations) soften.
• Credit hedge: Buy protection on ORCL senior credit (5‑yr CDS or buy puts on 7–10yr bonds) with a 12–24 month horizon. Pay modest premium to protect against a >150–300bp widening event tied to capex overruns or covenant pressure — a tail insurance trade against equity downside.
• Event trade: If management announces a material multi‑year anchor contract being delayed or canceled, short ORCL equity into the gap and cover on any confirmed renegotiation within 3–6 months; conversely, if Oracle secures long‑dated, non‑cancelable supply agreements for accelerators, add to the call‑spread leg.