Capital Wealth
FRI CLOSE · SEP 11   S&P 500 7,656.98 ▲0.86%  ·  DJIA 52,573.29 ▲0.98%  ·  NASDAQ 26,333.04 ▲0.96%  ·  10-YR 4.975%  ·  2-YR 4.572%  ·  WTI $99.99 ▼2.4%  ·  GOLD $4,390.00 ▲0.6%  ·  VIX 15.84 ▼11.2%
Markets & The Fed · The AI File

Oracle Booked a $664 Billion Backlog and Plans to Spend $95 Billion This Year to Serve It

Cloud infrastructure sales more than doubled and quarterly profit jumped 60%. The contracted future revenue now runs to $664 billion — and so does the capital spending required to deliver it.

By Sean Anees Saifi · Capital Wealth · Published Friday, September 11, 2026 · Source: The Wall Street Journal, September 11, 2026 edition
Key Points
$664B
Contracted future revenue, or backlog
+60%
Quarterly profit, to $4.76 billion
$90–$95B
Capital spending planned this year
-5.4%
Thursday close, before the numbers
A long corridor of server racks inside a data center, lights receding toward a bright doorway at the far end
Cloud-infrastructure demand tied to artificial intelligence has become the main driver of Oracle's revenue. The contracts behind it stretch years out; the spending needed to honor them lands now.
In one line: Oracle's AI backlog reached $664 billion, but the capital spending to deliver it arrives years before the revenue does — which is the whole question.

Oracle (ORCL) spent Thursday getting sold. The shares closed about 5.4% lower at $152.94, the kind of slide that usually means somebody knows something. Then the numbers landed after the bell and the stock jumped more than 7%. Quarterly profit was up 60%, to $4.76 billion, on revenue of $19.34 billion — and neither of those is the number that matters.

The backlog is the headline

Oracle books what it calls remaining performance obligations, which is a long way of saying contracted revenue it hasn't delivered yet. That figure now stands at $664 billion, up $209 billion in a year, with more than $30 billion of new AI cloud contracts signed in the quarter alone. Cloud revenue rose 62% to $11.61 billion, infrastructure sales more than doubled, and the legacy software business shrank 3% as customers kept migrating. Full-year adjusted earnings guidance went up to at least $8.10 a share.

Here's the part that doesn't fit in a headline. Delivering that backlog takes data centers, and Oracle still plans $90 billion to $95 billion of capital spending this year — against full-year revenue guided at least $90 billion. The company is building an enormous future out of a borrowed present, and its credit sits one notch above junk. The revenue is contracted. The spending is immediate. Those two facts live on different calendars, and the gap between them is filled with debt.

One quarter doesn't settle it

A backlog is a promise to pay, not a payment, and promises running into the 2030s carry counterparty risk, delay risk and refinancing risk on the way. That's why Oracle is held in the book and kept on the watch list rather than added to: a very good quarter is evidence, not a verdict. The round trip from down 5.4% to up 7% inside one evening tells you how thin conviction is on both sides of this name.

If a single holding has quietly grown into an outsized share of a portfolio because it's been right, that's worth an hour and a spreadsheet before it's worth a celebration. Bring the statement to your next review and count what one company is actually carrying.

What It Means For Your Portfolio

Watch — a real backlog funded with a lot of borrowed money

An AI backlog of $664 billion is a genuine asset and a genuine obligation at the same time: the revenue arrives over years, the capital spending arrives now, and the debt sits in between.

General planning principles, not advice for anyone in particular: a company can post a spectacular quarter and still carry real balance-sheet risk, and one earnings report rarely settles a long-term thesis in either direction. When a single holding has run hard, position size deserves at least as much attention as the story behind it.

In the book, Oracle is held and stays on the watch list — the position isn't being added to on the strength of one quarter. The house bias this week is cautious, defensive on duration and neutral on stocks, and nothing new is being bought: the September letter's conditions for tactical adds were not met after Friday's report.

Book a 15-Minute Review → Back to Edition No. 168 →