OpenAI has inked a contract with Oracle, committing to purchase $300 billion worth of computing power over five years, according to The Wall Street Journal report citing sources familiar with the matter. Starting in 2027, this agreement dwarfs OpenAI’s current annual revenue, reflecting a bold bet on AI’s future.
The deal requires a staggering 4.5 gigawatts of power capacity, equivalent to the electricity generated by over two Hoover Dams or used by about four million homes, added the WSJ report.
This massive contract helped Oracle’s shares surge up to 43%, adding $317 billion in future contract revenue during the quarter ending August 31. Oracle Chairman Larry Ellison’s net worth soared past $400 billion, rivalling Elon Musk among the world’s richest individuals.
Risks And Rewards: A High-Stakes Play For Both Companies
The OpenAI-Oracle contract, set to begin in 2027, represents a high-stakes gamble for both companies. OpenAI, currently operating at a loss, revealed in June that it generates about $10 billion in annual revenue—less than a third of the roughly $60 billion it will need to pay Oracle each year on average. Meanwhile, Oracle is betting heavily on a single client for a significant portion of its future revenue and will likely need to incur debt to finance the AI chips required to power the massive data centres.
The agreement was first hinted at in a June Oracle filing projecting over $30 billion in annual revenue from cloud services starting in 2027. OpenAI publicly announced the 4.5-gigawatt deal but did not disclose the contract’s full value, states the WSJ report.
OpenAI CEO Sam Altman is simultaneously juggling ambitious projects, including developing custom chips with Broadcom, creating a potential iPhone competitor, and launching a new cloud venture named Stargate. As per the report, the startup doesn’t expect profitability until 2029 and may lose $44 billion before reaching that milestone.
Meanwhile, a critical challenge facing OpenAI and the wider AI sector is an ongoing shortage of computing resources, which limits product rollouts and new model development. Industry-wide investment in chips, servers, and data centres is projected to reach $2.9 trillion by 2028, fueling what some describe as a “gold rush” for Wall Street financiers offering debt financing, noted the WSJ report.
