Oracle Stock Has Lost More Than Half Its Value in 6 Months. It May Finally Be Time to Buy.

Oracle reported fiscal Q3 revenue of $17.2B (+22% YoY) and EPS of $1.79 (+21% YoY), with cloud infrastructure revenue of $4.9B (+84% YoY). Remaining performance obligations surged to $553B (+325% YoY), driven by large-scale AI contracts that are often prepaid or customer‑supplied, reducing Oracle’s incremental funding needs. Management guided to roughly $50B of capex for fiscal 2026 to support AI buildout, a material near‑term cash burden, but the stock trades at ~19x forward EPS (vs. market ~21x), implying an attractive valuation if backlog converts while profitability holds.

Analysis

Oracle’s prepaid-heavy AI contracts change the capital dynamics: a large portion of future revenue is now accompanied by up‑front customer funding or customer‑supplied GPUs, shifting cash‑flow risk off Oracle and converting what would have been onerous capex into working‑capital driven receipts. That structure compresses near‑term financing risk but creates lumpy margin recognition and execution risk as Oracle must deploy or host third‑party hardware reliably — a service/ops challenge more than a pure technology one.

Second‑order winners include GPU OEMs and systems integrators that win Oracle’s volume economics; losers are the small cloud incumbents and legacy CPU vendors if Oracle secures long‑duration capacity commitments from hyperscale AI users. Network, power, and real‑estate players that can flex capacity to Oracle’s footprint become strategic partners — expect accelerated demand for colocation and for power‑dense racks over the next 12–36 months.

Key catalysts and risks: successful recognition of the backlog (timing over quarters), GPU supply allocation (weeks–months cadence) and realization of targeted utilization (12–24 months). Tail risk is two‑fold: (1) customer renegotiations or cancellations if AI project economics sour, and (2) operational failures on customer‑supplied equipment that force Oracle to assume replacement costs — either would re‑price credit and equity risk quickly.

Key Decisions for Investors

• Core long ORCL (4–6% portfolio) — buy equity now, target +35–50% over 12–24 months if backlog converts and utilization ramps; set a hard stop at -25% to protect against capex/operational execution failure. Consider buying 9–12 month protective puts (ORCL Sep/Dec 2026) to cap tail risk while holding the core position.
• Pairs trade: Long ORCL / Short INTC equal notional for 12 months — aim to capture GPU‑centric infrastructure wins versus legacy CPU exposure. Reward: asymmetric if Oracle locks in long AI spend; risk: Intel rebound or Oracle execution misses; close at 20% pair‑move or after 12 months.
• Long NVDA via call‑spread (buy Jan‑2028 400C / sell Jan‑2028 650C) to limit premium outlay — 18–36 month horizon to profit from sustained incremental GPU demand driven by Oracle and other hyperscalers. Risk: regulatory/competitive shock or weaker AI capex; position sizes should be limited given IV and concentration risk.
• Event hedge: Buy ORCL short‑dated puts (3–6 months) sized to cover 20–30% of the core equity allocation ahead of key execution print quarters; if Oracle reports execution glitches, these puts pay off and free up capital to add to long positions on weakness.