Opinion: Oracle bet on AI... and the bill is now due

Fourteen months ago, Oracle’s stock was the hottest in the technology sector. In September 2025, following the announcement of new cloud services contracts worth hundreds of billions of dollars, the company’s share hit a record high of $323.43. However, on November 2, 2026, the stock closed at $119.22, having fallen by more than 60 percent.

The spark came from the company’s disclosure of its plan to raise an additional $20 billion through the issuance of shares and bonds to fund its expansion in artificial intelligence, just five days after informing investors that this amount would be the maximum funding required through fiscal year 2030. The stock plummeted by 8.8 percent in a single day.

For decades, Oracle was akin to a cash-flow machine, selling databases and business software that customers rarely abandoned. But artificial intelligence has changed its model. The company decided to build the massive data centers required by firms such as OpenAI, signing a contract with the latter in 2025 valued at approximately $300 billion. The value of contracted work, which had not yet generated revenue, jumped from $138 billion to $664 billion in just over a year.

This growth is real: revenues rose by 30 percent to $19.3 billion in the quarter ending in August, and cloud infrastructure revenues more than doubled. However, Oracle is required to build data centers before collecting revenues, prompting it to rely heavily on borrowing to finance this expansion.

In the fiscal year ending in May 2026, capital expenditures surged by 162 percent to $55.7 billion, and free cash flow turned into a deficit of $23.7 billion. Total debt reached approximately $130 billion. Oracle plans to spend between $90 billion and $95 billion on capital expenditures during the current year. Standard & Poor’s Ratings Agency expects a cash deficit of nearly $42 billion, to which it adds approximately $260 billion in data center lease obligations that begin to take effect by 2029.

In July, Standard & Poor’s downgraded Oracle’s rating to BBB-, one notch above junk bond status, acknowledging that it had “underestimated the scale of required investments.” Meanwhile, Oracle cut approximately 21,000 jobs during fiscal year 2026, equivalent to 13 percent of its workforce.

The deeper source of concern lies in the concentration of obligations with a single client. According to Standard & Poor’s, OpenAI accounts for roughly half of Oracle’s contracted work, yet the company is not profitable. In April, The Wall Street Journal reported that OpenAI had failed to meet internal revenue and user growth targets, and that its chief financial officer had expressed concern about its ability to pay for future computing contracts. OpenAI dismissed the report as “nonsense,” but later stated that it had reduced its computing spending plans through 2030 to approximately $600 billion, down from $1.4 trillion.

If OpenAI fails to pay, Oracle could find itself stuck with lease obligations it cannot shed. The company is seeking to diversify its business, announcing that it will allow the use of Google’s TPU chips starting in 2027, and expects its free cash flow to turn positive in fiscal year 2028. Some analysts share this view; on November 2, Stephens Inc. upgraded its recommendation for the stock.

Oracle will not disappear from the market, as its software operates within a large number of banks, governments, telecommunications companies, and airlines worldwide. However, the behavior of a supplier facing financial pressures may change. Customers should prepare for more stringent audits of usage licenses, higher technical support prices, pressure to migrate to Oracle Cloud, and the possibility of reduced support teams following layoffs.

The Gulf region appears particularly vulnerable to these risks. Oracle has committed to investing $14 billion over ten years in Saudi Arabia, where it operates cloud service regions in Jeddah and Riyadh. It has also pledged to quintuple its investments in Abu Dhabi, and is a partner in the “Stargate UAE” project, a one-gigawatt artificial intelligence complex being built in collaboration with G42 and OpenAI.

Governments and Gulf companies building national services on Oracle’s infrastructure should ask direct questions: What will happen to prices and service levels if the company scales back its operations? Is there a realistic exit path from reliance on it? Are critical workloads distributed across multiple cloud service providers? This is not about inciting panic, but about fundamental supplier dependency risk management.

Not every AI investor is bound to Oracle’s fate. According to Morgan Stanley, companies such as Microsoft, Alphabet, Amazon, and Meta will spend approximately $700 billion on AI infrastructure this year, but most of this spending will be funded from existing profits. Oracle, by contrast, has resorted to borrowing despite having a weaker cash base, and has tied a significant portion of its bet to a single client.

Nevertheless, the gap is narrowing. Morgan Stanley expects AI-linked debt issuance this year to more than double its previous level, approaching $570 billion. This week, Oracle, Alphabet, and Meta all returned to the bond markets. The more AI is financed through debt, the more critical it becomes that revenues arrive on schedule.

This is the lesson Oracle offers. While AI may indeed transform the global economy, promised revenues over a decade do not pay this year’s debt interest. Every board betting on AI should ask the question now preoccupying Oracle’s investors: If the expected returns are delayed, can we afford to wait?

* CEO of DC Capital, visiting lecturer in entrepreneurship, and founder of the Islamic Global Chamber of Online Commerce and Industry. He is also a global speaker in strategy and startup development, known for his insights on economics and building innovative, ethical, and scalable business systems.