ORCL -- Earnings Review Q3 FY26

Date: 2026-03-11 | WSM


Verdict

This was a very good quarter, better than I expected, and the key point is that Oracle is starting to look like a real AI infrastructure revenue conversion story, not just an AI capex story.

I still would not call it a top-tier WSM name because the capital intensity, leverage, and lower-quality legacy mix make it less clean than NVDA/CRDO/AXON. But the quarter was decisively bullish.


The Table

Metric Q2 FY26 Q3 FY26 Read
Revenue $16.06B $17.19B +22% YoY, about +7% QoQ
Non-GAAP EPS -- $1.79 Beat
Total Cloud Revenue ~$7.97B $8.90B +44% YoY
OCI / IaaS ~$4.10B $4.90B +84% YoY, +19.5% QoQ
SaaS / Cloud Apps ~$3.87B $4.00B +13% YoY
RPO $523B $553B +325% YoY, +$29B QoQ
GAAP Op Margin -- 32% Solid, not elite
Non-GAAP Op Margin -- 43% Down slightly YoY
Q4 Revenue Guide -- +19% to +21% YoY Strong follow-through
Q4 Cloud Guide -- +46% to +50% YoY Very strong
FY26 Revenue Guide -- $67B Held
FY27 Revenue Guide prior lower $90B Raised
FY26 CapEx Guide -- $50B Huge

What Matters

1. OCI is the story

OCI at +84% YoY is the number. Not the legacy database business. Not the applications. OCI.

And sequentially, +19.5% QoQ at this scale is real acceleration. That says backlog is converting.

2. RPO is absurdly strong

$553B RPO, up 325% YoY. Eish!! That is not normal.

Now, I do want to be careful here: backlog is not revenue, and Oracle has a history of investors getting excited about future promises. But when OCI is also accelerating, the RPO matters much more. The leading indicator is finally lining up with current revenue.

3. The market's main fear eased

The fear was: Oracle is levering up and spending wildly on AI infrastructure without enough demand visibility.

This quarter answered that in three ways:

  1. Revenue beat
  2. OCI acceleration
  3. Management saying much of the AI capacity is customer-prepaid or customer-funded

That last point matters a lot. It makes the capex look less reckless.

4. It is still not a clean WSM-style business

The problem is obvious:

So this is not AXON. It is not CRDO. It is not even NVDA in terms of quality. It is a more complex story.


What Management Said That I Care About

Topic Commentary My Read
AI infrastructure demand Capacity coming online over 3 years, mostly partner-funded Bullish if true; lowers funding risk
Multicloud traction 33 Microsoft, 14 Google, 8 AWS regions live; more coming Strengthens Oracle's role as neutral AI/data layer
AI infra margin Q3 delivered AI capacity at ~32% gross margin, above 30% guide Better than feared
AI in software Ellison/Sicilia argued Oracle benefits from AI rather than gets disrupted by it Narrative helpful, but OCI numbers matter more

My Read Through The WSM Lens

Lens Assessment
Revenue trajectory Improving
Leading indicators Very strong -- RPO, multicloud, AI demand
Margins Good enough, not exceptional
FCF Weak because of capex
Simplicity Middling -- too many moving parts
Management credibility this quarter Improved
Thesis Intact-to-strengthening post print

Bottom Line

I like the quarter. Much more than the stock historically deserved from me.

Would I buy ORCL over NVDA, CRDO, AXON or even FIGR here? No.

Would I dismiss ORCL as an old legacy company missing AI? Also no. That would be a mistake after this print.

The correct conclusion is: Oracle is executing far better than the market expected, and the OCI/RPO combination is now too strong to ignore. But for me it still sits in the "good company, not best idea" bucket because of capex intensity and business complexity.

If OCI drops back toward 50-60% growth while capex stays enormous, I would get cautious quickly. If OCI can hold high growth and backlog keeps converting, then the rerating can continue.

-wsm