Oracle’s aggressive move into AI infrastructure has turned from a source of market excitement into a test of investor confidence. The company is spending heavily on chips and data centers, with much of the strategy tied to supplying computing capacity to OpenAI, the maker of ChatGPT.
The concern on Wall Street is not simply that Oracle is investing in AI. It is that the company is borrowing at enormous scale, taking on long-term commitments, and relying heavily on contracts linked to a small number of AI companies.
A fast pivot into AI infrastructure
Oracle, the US software group founded by Larry Ellison, entered the AI race with a much larger infrastructure push than many investors had expected. The company has committed to spend hundreds of billions of dollars in the next few years on chips and data centers.
Those investments are largely connected to deals to provide computing capacity to OpenAI. Oracle has said its deals with OpenAI would generate $300 billion of revenue between 2027 and 2032.
That headline revenue figure helps explain why the stock had gained attention. But it also explains the pressure now building around the company. AI infrastructure is expensive, and the market is increasingly focused on whether large technology companies can turn massive data center spending into durable value.
Oracle’s shares are down 25 percent in the past month, nearly twice the fall of the next worst-performing hyperscaler, Meta. The decline has reversed more than $250 billion of gains in market value since the Texas-based group disclosed its deals with OpenAI in September.
Why investors are worried
The sell-off reflects a broader concern about lofty technology valuations and very large capital expenditure by a small group of companies. Those worries are sharper when AI demand depends on lossmaking AI start-ups such as OpenAI and Anthropic delivering on their promises for the technology.
Oracle is drawing particular scrutiny because it shifted from business software to cloud computing later than its rivals. Its current strategy is more concentrated around an all-out AI infrastructure bet, and that bet is pinned largely to the success of OpenAI.
Alex Haissl at Rothschild & Co Redburn described the shift bluntly: “This is a completely different business model to what investors prize in cloud services.” He added: “The deals look fantastic when you look at the revenue figures, but they are very capital-intensive so create very little value.”
The market reaction has also appeared in Oracle’s debt. A Financial Times index tracking the price of Oracle’s debt has fallen about 6 percent since mid-September, significantly worse than any of its major peers.
Oracle shares fell 4.2 percent on Thursday as the NASDAQ tumbled 2.3 percent, before recovering some of those losses on Friday. Even after the recent pressure, Oracle’s shares are still up 30 percent this year.
Debt is at the center of the debate
Oracle has been aggressive in tapping debt markets to build AI capacity quickly. The group has about $96 billion of long-term debt, up from $75 billion a year ago, according to Bloomberg data.
Morgan Stanley forecasts that figure will rise to about $290 billion by 2028. Oracle also sold $18 billion of bonds in September and is in talks to raise $38 billion in debt financing through a number of US banks.
That borrowing is creating a sharper contrast with other hyperscalers. Of the five hyperscalers, which include Amazon, Google, Microsoft, and Meta, Oracle is the only one with negative free cash flow. Its debt-to-equity ratio has surged to 500 percent, far above Amazon’s 50 percent and Microsoft’s 30 percent, according to JPMorgan.
JPMorgan analysts pointed to a “tension between [Oracle’s] aggressive AI build-out ambitions and the limits of its investment-grade balance sheet.” Barclays analysts also downgraded their rating of Oracle’s debt from market neutral to underweight, warning that AI infrastructure spending had outpaced free cash flow.
Credit rating agency Moody’s has flagged significant risks from Oracle’s reliance on a small number of AI companies. S&P Global warned that a third of Oracle’s revenues will be tied to a single customer by 2028, referring to OpenAI.
The OpenAI concentration risk
The core question is whether Oracle’s OpenAI-linked contracts can justify the financial strain required to serve them. OpenAI faces questions about how it plans to meet its commitments to spend $1.4 trillion on AI infrastructure over the next eight years, and it has also struck deals with several Big Tech groups, including Oracle’s rivals.
Andrew Chang, a director at S&P Global, said: “That is a huge liability and credit risk for Oracle. Your main customer, biggest customer by far, is a venture capital-funded start-up.”
Analysts have also noted a timing mismatch in Oracle’s commitments. Oracle’s data center leases are for much longer than its contracts to sell capacity to OpenAI.
The company has signed at least five long-term lease agreements for US data centers that will ultimately be used by OpenAI. Those agreements create $100 billion of off-balance-sheet lease commitments. The sites are at varying stages of construction, and some are not expected to break ground until next year.
Leadership and the next phase
Oracle’s AI pivot has also brought leadership changes into focus. Safra Catz, Oracle’s sole chief executive from 2019 until she stepped down in September, had resisted expanding the cloud business because of the vast expenses required.
She was replaced by co-CEOs Clay Magouyrk and Mike Sicilia as Oracle moved into a new era focused on AI. Catz is now executive vice-chair of Oracle’s board.
Oracle’s executives argue that the rewards will justify the risks because AI demand is intense and accelerating, while existing supplies of computing power remain insufficient. Oracle’s infrastructure business is forecast to increase revenues by more than 10 times by 2029, according to estimates compiled by S&P Visible Alpha, and the bulk of Wall Street analysts remain bullish on the stock.
That is the split now defining Oracle’s market story. The upside case rests on scarce computing capacity, huge OpenAI-linked revenue, and fast infrastructure growth. The risk case rests on debt, negative free cash flow, customer concentration, and long commitments that may be difficult to unwind if AI demand fails to meet expectations.