OpenAI just committed to spend $300 billion on Oracle cloud services over the next five years: a figure equal to the GDP of Finland. The contract, one of the largest in cloud history, would give Oracle roughly $60 billion in annual revenue on average, more than Google Cloud makes today. Oracle’s shares surged more than 40% on the news, and analysts initially applauded. “We’re all kind of in shock, in a very, very good way,” said Brad Zelnick of Deutsche Bank.
Something that’s hard to ignore: Oracle’s future cloud growth is tied overwhelmingly to one customer. Unlike AWS, Azure, or Google Cloud, which built their businesses on diversified client bases, Oracle is concentrating its bets on OpenAI, a company that is still losing billions each year.
Oracle itself acknowledged the concentration in its filings. In June, it disclosed a cloud agreement with one customer that would deliver more than $30 billion in annual revenue starting in 2027, widely assumed to be OpenAI. That contract alone would account for a large share of Oracle’s projected growth.
“Listen, even I am sort of blown away by what this looks like going forward,” said John DiFucci of Guggenheim Securities, who described the numbers as unprecedented. But UBS analysts flagged that reliance on a single buyer could create “margin pressure” if costs of delivering compute rise or if terms are renegotiated.
The risk is magnified by OpenAI’s financial profile. The company generated roughly $10-12 billion in revenue this year, yet committed to average payments of $60 billion annually. OpenAI told investors it does not expect to turn a profit until 2029 and will lose about $44 billion before then.
Reuters reported that OpenAI expects to burn $115 billion through 2029, driven primarily by escalating compute costs. In 2024, it lost $5 billion on $3.7 billion in revenue, and its ChatGPT Pro subscription still runs at a loss.
For Oracle, this means staking tens of billions in future revenue on a customer whose own cash flow is precarious.
The obligations fall on Oracle too. The OpenAI contract requires 4.5 gigawatts of data center capacity, the output of two Hoover Dams or enough to power four million homes. To build it, Oracle will need to purchase vast quantities of GPUs and other hardware, likely taking on additional debt.
That is not a comfortable position for a company with a debt-to-equity ratio of 427%, compared with 33% for Microsoft. Over the past year, Oracle’s $27.4 billion in capital expenditures exceeded its $21.5 billion in operating cash flow, according to S&P data.
The stock market has cheered the growth story. Oracle’s backlog surged to $455 billion, up 359% from a year earlier. Larry Ellison’s net worth soared by $100 billion to $393 billion, briefly surpassing Elon Musk as the world’s richest man.
But some analysts are already warning that the stock is overheated. “Many are blown away,” CNBC reported, but others “are urging caution, warning the unprecedented success has made the stock overvalued”.
Oracle’s boldness cannot be dismissed. Project Stargate, its joint initiative with OpenAI and SoftBank, envisions $500 billion in new U.S. data center investments, positioning Oracle at the center of global AI infrastructure. Oracle also continues to expand into adjacent areas, from rumored interest in TikTok to Ellison’s financing of an $8 billion Paramount acquisition bid by his son’s Skydance Media. These moves underscore Ellison’s ambition to make Oracle indispensable to both AI and media ecosystems.
Yet ambition does not erase concentration risk. AWS and Azure spread growth across millions of workloads, from e-commerce to government contracts. Oracle is relying disproportionately on the fortunes of one unprofitable startup.
For investors, three variables now matter most: whether Oracle can finance the data center build-out without eroding margins; whether OpenAI can grow beyond its losses and honor its commitments; and whether power constraints, chip shortages, or regulatory pressure derail the ramp-up.
The Oracle-OpenAI contract is historic in scale, and in potential reward. But it is also exposed. If OpenAI’s growth falters, Oracle’s empire will rest on a single, shaky pillar.
