The Headline That Doesn’t Make Sense Until It Does
Oracle hit $345 per share in September 2025. It’s trading at $149 today. That’s a 57% decline from peak while the company was simultaneously posting record revenue, winning AI infrastructure contracts, and getting embedded into the Stargate joint venture alongside OpenAI and SoftBank. The average analyst price target sits at $252 — implying roughly 70% upside from here. Something in this picture is priced wrong.
Understanding ORCL means understanding both why the stock fell that far and why the business didn’t follow it down.
What the Business Actually Looks Like
Oracle is no longer just a database company that happened to build a cloud. By FY2026, it has reorganized itself around three distinct growth engines operating at the same time:
1. Oracle Cloud Infrastructure (OCI) — Oracle’s hyperscale cloud platform, competing with AWS, Azure, and Google Cloud but with a differentiated architecture. OCI’s RDMA networking and GPU cluster density made it the preferred platform for AI training workloads that require extreme inter-node bandwidth. While it remains smaller than the three giants, it has been gaining share specifically in the AI training segment where architectural differentiation matters most.
2. The Stargate Joint Venture — Oracle is one of three founding partners (alongside OpenAI and SoftBank) in Stargate, the largest AI infrastructure project ever announced. Initial capacity is anchored at Abilene, Texas, with multi-site expansion planned across the U.S. The market read Stargate’s announcement as a structural endorsement of OCI’s capabilities. The financial impact shows up in Oracle’s RPO (remaining performance obligations) — the contracted-but-not-yet-recognized revenue backlog that has grown to levels that imply years of visibility.
3. Cloud Applications + Database — Fusion ERP, HCM, and NetSuite continue to compound at double-digit rates. Oracle also struck MultiCloud partnerships with AWS, Azure, and Google, meaning Oracle Database runs natively inside competing hyperscaler environments — turning what was a zero-sum fight into a growth engine. The Cerner healthcare acquisition, which had a painful integration period, is now contributing constructively.
FY2026 Numbers
| Metric | Result | vs. Estimate | YoY |
|---|---|---|---|
| Q4 Revenue | $19.18B | Beat ($19.10B) | +20.6% |
| Q4 EPS (adj.) | $2.11 | Beat ($1.96) | +24% |
| Full-Year Revenue | $67.4B | Record | +17% |
| Stock Reaction (Q4) | −10% | Despite beat | — |
Why the Stock Fell 57% While the Business Grew
The disconnect has three layers:
Layer 1 — It was priced for perfection. At $345, Oracle was trading at a multiple that priced in both perfect execution on Stargate and immediate FCF conversion. When Q2 FY2026 (December 2025) showed mixed results and guidance that didn’t accelerate as aggressively as bulls hoped, the multiple compressed fast. The January 2026 selloff took another 15%+ off on top of that.
Layer 2 — DeepSeek and AI efficiency concerns. When DeepSeek demonstrated that frontier AI models could be built with a fraction of the assumed compute, the market repriced the entire AI infrastructure buildout narrative. Companies like Oracle — which had been valued on the assumption that AI demand would grow exponentially and indefinitely — got hit disproportionately. The thesis didn’t break; it just got questioned.
Layer 3 — Capex ate free cash flow. Oracle has been aggressively issuing debt and spending to build data center capacity for Stargate and OCI expansion. FCF compressed as that spending accelerated. The market hates paying growth multiples for a company burning cash on construction. Q4 FY2026 beat on revenue and EPS, then fell 10% anyway because capex guidance and the conversion timeline didn’t satisfy investors.
The AI Restructuring Story Nobody Talks About
Oracle cut 21,000 employees in fiscal 2026 — roughly 13% of its entire workforce. Part of this is AI automation replacing roles inside Oracle’s own operations. This is a company that is simultaneously selling AI infrastructure to the world and using AI to restructure its own cost base. When the capex cycle matures and FCF recovers, the leaner headcount structure sets up margin expansion that isn’t fully priced in.
Where the Stock Stands Today
At $149, Oracle is trading at roughly 25x TTM earnings on $5.82 EPS, with a $430B market cap and $67.4B in annual revenue. The 52-week range tells you everything about the sentiment journey: from $134.57 (near the lows) to $345.72 (the AI-hype peak) and back to $149. The average analyst target of $252 implies the stock would need to rally 70% just to reach consensus. Next earnings are September 9, 2026.
The dividend is $2.00 annually (1.31% yield) — not why you own this, but it reflects a management team that is still returning capital while funding a massive buildout.
Position Framework
- Starter: $140–$155 — near current price, reasonable entry for the backlog thesis
- Add: $125–$135 — if macro or AI spending concerns push it toward the 52-week low
- Trim: $220–$240 — as it approaches analyst consensus PT; re-evaluate guidance at September earnings
- Target: $250–$280 over 12–18 months if Stargate backlog converts and capex normalizes
- Invalidation: Stargate JV restructures or delays materially; OCI growth decelerates below 15%; FCF doesn’t recover by FY2027
Oracle is a company that spent the better part of a decade being dismissed as a relic, quietly built a cloud platform that now hosts frontier AI training workloads, and got embedded into the largest AI infrastructure project in history. The stock peaked when the market priced all of that in at once, then gave half of it back when the timeline stretched. At $149, you’re not paying for the hype — you’re paying for a real business that grew revenue 17% last year and has years of contracted backlog ahead of it. The question is patience.