Oracle’s stock has fallen 50% from its September high despite 41 of 51 Wall Street analysts rating it a buy with 43% implied upside. The disconnect centres on Oracle’s $300 billion Stargate deal with OpenAI, where investors see circular financing risk, OpenAI is simultaneously a customer and an investment of the companies funding it, while analysts argue AI infrastructure demand is fungible and Oracle is deeply undervalued.TL;DR
Oracle’s stock has fallen nearly 50 per cent since hitting a record in September and dropped 14 per cent in the six sessions through Thursday, its worst stretch in months. Of the 51 Wall Street analysts tracked by Bloomberg who follow the company, 41 have buy ratings and only one rates it a sell. The consensus price target implies a 43 per cent rise over the next twelve months, one of the highest projected upsides among Oracle’s large-cap peers. One portfolio manager described the stock as offering “an incredibly attractive entry point” with a price target double the current share price. Another called the worries about Oracle’s debt “ridiculous.” The analysts are bullish. The investors are selling. The divergence is not about Oracle’s quarterly numbers, which showed a 95 per cent jump in net income and remaining performance obligations of $523 billion, up 433 per cent year on year. It is about one customer.
Oracle’s growth thesis rests substantially on a $300 billion, five-year partnership with OpenAI to supply cloud computing infrastructure for Stargate, the AI infrastructure joint venture between OpenAI, SoftBank, and Oracle announced in January 2025. SoftBank has assembled a record $40 billion loan to finance part of the venture. Oracle’s role is to build and operate the data centres that will house the computing power OpenAI needs to train and run its models. The $300 billion figure made Oracle’s stock when it was announced. It is now the reason the stock is falling. The question investors are asking is not whether Oracle can build the infrastructure. It is whether OpenAI can afford to pay for it.
OpenAI has contracted for hundreds of billions of dollars in cloud infrastructure across multiple providers, with commitments that assume revenue will grow from approximately $25 billion today to $280 billion by 2030. The company expects to burn through $25 billion in cash in 2026 against a revenue target of $30 billion. OpenAI missed its internal revenue and user growth targets according to a report this week, and the company’s finance chief warned colleagues that if revenue growth does not accelerate, OpenAI could face difficulty funding future compute agreements. Oracle’s shares fell 4.1 per cent on the day the report was published.
Investors have identified a structural problem with Oracle’s OpenAI relationship that analysts have largely dismissed. OpenAI is simultaneously a customer of the companies it is funded by. It has raised capital from Microsoft, which provides Azure infrastructure. It has contracted with Oracle, which has invested in the Stargate venture. It has committed $100 billion to AWS over eight years. The companies building OpenAI’s infrastructure are, in several cases, the same companies funding OpenAI’s operations. If OpenAI’s revenue growth slows and its ability to service these contracts weakens, the companies that lose a major customer are the same companies that lose an investment. The concentration risk is not just that Oracle depends on OpenAI for revenue. It is that the entire financing structure is circular, and a slowdown at OpenAI would simultaneously reduce Oracle’s revenue, impair its investment, and lower the value of the infrastructure it built specifically for the partnership.
The 💜 of EU tech
The latest rumblings from the EU tech scene, a story from our wise ol' founder Boris, and some questionable AI art. It's free, every week, in your inbox. Sign up now!
OpenAI has already paused its Stargate UK data centre project, citing high electricity costs and an unfavourable regulatory environment around AI copyright. The Stargate data centre planned for Abu Dhabi faces geopolitical risk from the Iran conflict. When Oracle and OpenAI scrapped plans to expand a data centre in Abilene, Texas, in March, Meta quickly considered leasing the site, with Nvidia helping to facilitate the discussions. The fact that Meta stepped in is cited by Oracle bulls as evidence that AI infrastructure demand is fungible: if one customer cannot fill a data centre, another will. The bears note that Meta is already a customer of CoreWeave and Nebius, not Oracle, and that stepping in to consider leasing a site is not the same as signing a $300 billion contract.
Wall Street’s argument is straightforward. AI infrastructure demand is real and growing. Oracle has cut up to 30,000 employees to redirect $8 billion to $10 billion in annual cash flow to data centre construction, a move that analysts at TD Cowen describe as accelerating the company’s AI pivot. The infrastructure Oracle is building, data centres equipped with Nvidia GPUs and high-bandwidth networking, is needed by every company training or running AI models, not just OpenAI. Larry Ellison, who has run Oracle for 40 years, “sees a vision and executes it,” said one portfolio manager. The company’s cloud business is growing, its remaining performance obligations are at record levels, and the AI infrastructure buildout is in its earliest years.
The bearish counterpoint is equally straightforward. Oracle’s stock trades at 21 times forward earnings, above its 10-year average of 18 times, despite having fallen 50 per cent. The cost of protecting Oracle’s debt against default recently rose to its highest closing level on record, surpassing the 2008 financial crisis. The company’s debt load, taken on to fund AI data centres, has made it vulnerable to any slowdown in the infrastructure cycle. And the AI disruption of the software sector that Oracle also operates in, its enterprise database and application business, represents an additional risk that investors are pricing but analysts are not. The question is not whether AI infrastructure is valuable. It is whether Oracle’s specific exposure to one customer’s specific financial commitments makes the stock a concentrated bet on OpenAI’s ability to grow revenue tenfold in four years.
Oracle’s stock snapped its six-session losing streak on Friday, jumping as much as 5.5 per cent in early trading after OpenAI’s finance chief said the company sees “a vertical wall of demand” for its products and is meeting its objectives. The market’s reaction illustrates the problem: Oracle’s stock now moves primarily on OpenAI news, not on Oracle’s own performance. A single customer’s quarterly commentary has become the dominant input to Oracle’s valuation. This is exactly the concentration risk that investors are pricing and that analysts say the market is overweighting. The test will come over the next three to four quarters, as Oracle begins to report revenue from the Stargate-related infrastructure it is building. If OpenAI’s commitments convert to actual spending, the analysts are right and the stock is deeply undervalued. If the commitments do not convert, because OpenAI’s revenue growth slows or because the circular financing structure that funds the commitments unwinds, Oracle will have spent tens of billions building data centres for a customer that cannot afford them. The infrastructure may be fungible. The $300 billion contract is not.
Get the most important tech news in your inbox each week.