In one of the largest cloud computing deals ever, OpenAI has contracted with Oracle for $300 billion in processing power over five years starting in 2027. Oracle has committed to 4.5 gigawatts of capacity. A typical nuclear plant caps at 1 gigawatt of output generated at any given instant. The deal involves risk for both companies. OpenAI’s annual revenue of about $10 billion is far short of the amount needed to cover this tab. Oracle’s exposure comes with depending on a small number of large customers for so much revenue and the expense of expanding infrastructure to fulfill the obligation.

OpenAI is “burning more money than virtually any other startup on the planet,” and does not expect to generate a profit until 2029, according to The Wall Street Journal, which sums up the “seemingly impossible … host of business challenges” the company is facing such as “trying to build custom chips with Broadcom, create an iPhone competitor and launch a new cloud company called Stargate.”

Oracle — along with SoftBank, Nvidia and CoreWeave — is a partner in Stargate, OpenAI’s data center branding. The company plans to plow $500 billion into Stargate over five years. “After signing its deal with Oracle, OpenAI has secured commitments for more than half of that goal,” The New York Times reports.

WSJ says the $30 billion 2027 commitment Oracle disclosed in July is part of this agreement, which will scale up “over time as more data centers come online.” Stargate is intended to meet OpenAI’s own processing needs for model training and inference as well as serving outside clients.

The agreement coupled with a strong quarterly earnings Tuesday rocketed Oracle shares by up to 43 percent, enough to swell by $101 billion the value of co-founder Larry Ellison’s 41 percent stake to more than $392 billion, allowing him to briefly surpass Elon Musk as the world’s richest individual.

Oracle CEO Safra Catz said in an earnings release that the company “signed four multi-billion-dollar contracts with three different customers in Q1,” its most recent fiscal quarter. During that period, “multicloud database revenue from Amazon, Google and Microsoft grew at the incredible rate of 1,529 percent,” added Ellison, the company’s chairman and CTO.

At Oracle AI World, October 13–16 in Las Vegas, the company plans to introduce a new cloud infrastructure service called the Oracle AI Database that will let Oracle’s “tens of thousands” of enterprise customers choose their preferred LLM from a selection that includes Google’s Gemini, OpenAI’s ChatGPT, xAI’s Grok and more and use it “directly on top of the Oracle Database to easily access and analyze” their existing data.