Oracle Update: A Debt Plan, and the $20 Billion to Fund It
In July the debt kept us cautious on Oracle. On 28 October Oracle capped new share and bond sales at $20 billion through fiscal 2030, set a debt target that counts its leases and gave a date for positive cash flow. Five days later it raised the full amount. We're less worried now. Here's why, and what you'd have to believe either way.

In March we wrote a full 5D case study on Oracle. There was a lot to like. Margins were strong, cloud revenue was growing fast and most analysts were positive. Our DCF had it well above the price. But we called it a leveraged bet on AI infrastructure, because of the debt.
By our July update the customer side looked better, but the balance sheet looked worse, and we were more cautious.
The worry
Oracle is spending far more on AI data centres than it brings in. It has been covering the gap by borrowing and selling new shares. In July S&P cut its credit rating to one notch above junk. The debt kept climbing, and Oracle kept saying more share sales were coming without saying how many.
That made a share hard to value. If you don't know how many new shares are coming, you don't know how big your slice is. And there was no number to hold Oracle to on the debt.
What Oracle announced
At its investor day on 28 October, Oracle said it will raise no more than $20 billion of new money from bond and stock investors through fiscal 2030, which ends in May 2030. Half would be shares and half debt. Refinancing old bonds doesn't count. That's on top of the $20 billion of stock it sold in the quarter to August.
It also gave dates for the cash. Spending on data centres peaks this year and next, then steps down. Free cash flow, the cash left after that spending, turns positive in the second half of fiscal 2028, which runs from December 2027 to May 2028. It reaches at least $30 billion in fiscal 2029.
And it set a target for the debt. Net debt should be below 3.5 times adjusted earnings before interest, tax, depreciation and amortisation (EBITDA) in fiscal 2029. That counts leases and supplier financing as debt, so it covers what sits outside the $20 billion cap too. Keeping its investment-grade rating comes first in Oracle's list of priorities.
Five days later, on 2 November, Oracle raised the full $20 billion in one go. It sold 86 million new shares at $116, and $10 billion of bonds paying 7% to 8.65% a year. All three rating agencies kept their ratings on the new bonds.
The stock jumped 10.7% on investor day and fell 8.8% on the day of the raise. It closed at $123.77 on 4 November.
What it means
The share sales are done. Oracle has about 8% more shares than it had in May, and under its plan there are no more share sales to raise money before mid-2030. Employee stock awards will add a little each year, and February's preferred shares convert into about 1% more.
Customers are carrying more of the cost. In the quarter to August they paid $11.4 billion upfront, and cash from the business nearly tripled from a year earlier, to $23.1 billion. Customers who pay ahead for capacity want it.
Demand is still growing. Cloud revenue grew 62% in that quarter and total revenue 30%. Oracle's backlog of signed contracts reached $664 billion, up from $455 billion a year earlier.
What could still go wrong
The debt keeps rising until the cash turns. Free cash flow was still minus $5.4 billion in the quarter to August. Oracle has also committed to $288 billion of data centre leases that haven't started yet. Those count toward the 3.5 times target, so earnings have to grow as the leases start.
A large share of the backlog reportedly comes from one customer, OpenAI. And S&P has said it could cut Oracle again if debt stays above about 4.5 times earnings on its own measure, which is about where it expects Oracle this year. S&P also wants a path to positive cash flow by fiscal 2029, and Oracle's plan gets there a year earlier.
What our valuation says
Our discounted cash flow (DCF) model puts Oracle at about $332 a share as of the 5 November run, against a price of $123.77. That's more than twice the price, so our page flags it as low confidence. Nearly 90% of the value comes from cash Oracle earns more than ten years out.
The more useful question is how much growth the price already assumes. Here's how the model value moves when we change only revenue growth:
| Revenue growth, per year for 5 years | Model value | vs $123.77 price |
|---|---|---|
| 32% (analyst estimate) | $332 | +168% |
| 25% | $228 | +84% |
| 20% | $172 | +39% |
| 15% | $126 | about the same |
| 10% | $89 | -28% |
At today's price the model needs about 15% a year for five years, the same as in July. Analysts expect 32%.
The model doesn't use Oracle's spending plan. It eases spending down slowly over ten years, so its free cash flow turns positive in fiscal 2031, three years after Oracle's date. If spending falls when Oracle says it will, the cash comes sooner than the model assumes.
Our DCF uses Oracle's average share count for its latest quarter, to August. That leaves out some of the shares sold during the quarter and the 86 million sold on 2 November. Counting every new share and the cash it raised takes the value from about $332 to about $323.
On the same earnings and sales multiples as other tech companies, Oracle comes out at about $210 a share, and against its own history about $149. In March the peer multiples gave about $143, close to the price. Now they sit above it too.
Where the five dimensions stand now
| Dimension | March 2026 | Now |
|---|---|---|
| Business quality | 3.0 | 2.9 |
| Peers (sector / industry) | 3.1 / 2.9 | 3.3 / 3.6 |
| Valuation (DCF / peer multiples vs price) | $304 / $143 vs $139.66 | $332 / $210 vs $123.77 |
| Analyst sentiment | 3.6 | 3.7 |
| Holdings | 1.7 | 4.0 |
Quality barely moved. Leverage still scores 1.5 out of 5, as in March, while margins score 4.75. Peers look better, though both comparison groups have grown since March, so it isn't like for like. Analysts are about where they were.
Holdings moved the most. Insiders sold far less than usual in the September quarter, and on 29 September a director, Stephen Rusckowski, bought 25,000 shares at $139, the only insider purchase this year. More institutions added Oracle than cut it in the June quarter.
Where this leaves us
In March and July our DCF had Oracle well above the price, and it still does. By July the worry was how Oracle was paying for its build-out: no limit on new shares, and no target for the debt.
Oracle has now addressed both. The share sales are capped, and the debt has a target that counts the leases, with dates for when the cash turns. So the main reason for our caution is smaller, and we're less worried than we were in July.
What's left is delivery. Customers have to keep paying for the capacity Oracle is building, OpenAI most of all. If they do, the debt plan holds. If AI spending slows before 2028, the debt keeps rising, the cash comes late, and S&P's 4.5 times line comes into play.
You can test your own assumptions, including keeping spending high for longer, on our Oracle valuation page.
Sources: Oracle's 28 October 2026 investor day release (Form 8-K, exhibit 99.1); its 2 November 2026 bond pricing term sheet and preliminary share and bond prospectuses, with the $116 share price from news reports of the final pricing; its quarterly report for the three months to 31 August 2026 (Form 10-Q); and Stockoscope's data as of the 5 November 2026 run. S&P's rating triggers are from public reporting of its 9 July 2026 report. OpenAI's share of the backlog is from news reports; Oracle's filings don't name the customer. Prices are closes to 4 November 2026.