Oracle's $20 Billion War Chest Meets a Political Roadblock

Oracle shares fell 7.1% on Wednesday, caught in a broad selloff that Reuters says is heading toward the tech sector's steepest one-day drop since June. Worries about AI spending and data-center buildouts are weighing on the whole industry — not on any single company announcement. That distinction matters, because the more revealing story at Oracle sits beneath the sector-wide noise: the company has just finished raising a lot of money, and it has just lost a marquee contract to politics.

The raise is done; the return is not

Oracle closed a $20 billion financing on Tuesday, split evenly between $10 billion of new equity priced at $116.00 per share and $10 billion of bonds. The terms were announced Monday to fund the company's AI and cloud infrastructure expansion.

The senior notes come in five tranches with coupons running from 7.000% to 8.650% and maturities stretching from 2029 to 2046. Those fixed coupons are a contractual obligation; the revenue the funded infrastructure is meant to generate still has to be earned. Capital in hand enables investment — it does not settle the question of what that investment will eventually return.

Free cash flow is the real test

On October 28, Oracle reaffirmed guidance for fiscal 2027 of at least $90 billion in revenue and adjusted earnings of $8.10 per share. That forecast is the nearer checkpoint for the buildout.

The free cash flow outlook explains better why broader AI-spending skepticism lands on Oracle. Growth and self-funding are different yardsticks, and revenue targets alone obscure the stretch in which capital must be put up front. Oracle expects positive free cash flow in the second half of fiscal 2028 and has penciled in at least $30 billion for fiscal 2029 — a target, not a surplus already booked. It marks the intended turn from capital-heavy expansion to a business that generates cash.

By fiscal 2030, Oracle is aiming for at least $225 billion in revenue, an adjusted operating margin above 40%, and adjusted earnings per share of at least $21. The scale of the ambition is clear. What the targets cannot do is substitute for proof that spending, growth and cash generation fit together.

A $7 billion deal undone by Washington and Beijing

Capital is not the only constraint. Reuters reported Tuesday, citing the Financial Times, that Oracle scrapped a chip-leasing agreement with Tencent after U.S. officials raised concerns. The five-year deal was valued at roughly $7 billion and would have given Tencent access to about 100,000 advanced AI chips in Oracle data centers across Southeast Asia.

According to people familiar with the matter, the chips were unavailable in China because of restrictions imposed by both Washington and Beijing, and officials in Guangdong had pressed Tencent to throw more support behind domestic chip suppliers. The transaction was thus under political pressure from both sides. Reuters could not independently confirm the report; Oracle, Tencent and the White House did not respond to requests for comment. The reported contract value should not be confused with revenue already on the books.

A $664 billion backlog only counts if it converts

Oracle reported a backlog of $664 billion for the quarter ended August 31, 2026 — a figure that speaks to substantial future business. The Tencent episode shifts attention to the conditions under which that business gets delivered. Nothing in the cancellation points to a broad drop in demand, just as a fat backlog does not guarantee that every signed project will proceed unhindered.

The spending burden remains heavy: Oracle plans capital expenditures of $90–95 billion a year, and free cash flow was minus $5 billion in the most recent quarter. Outlays are running ahead of the cash the company hopes to generate, which raises the stakes on durable customer contracts — they need to document demand now and produce revenue later.

Oracle reports second-quarter results in December. Beyond growth, investors will be watching how the backlog converts. The Tencent case makes the point plainly: a financed buildout and secured revenue are two separate prerequisites for the AI bet to pay off.