Larry Ellison, CEO of Oracle, said he had dinner with Elon Musk and Jensen Huang to ask Nvidia CEO to provide AI GPUs.
During an earnings call last week, Oracle founder Larry Ellison admitted that he and Musk had to “beg” Nvidia CEO to sell them the company’s latest GPU.
“At Nobu Palo Alto, I had dinner with Elon Musk and Jensen Huang . I would describe that dinner as Elon and I begging Jensen to sell GPUs: Please take our money. No, take more. You’re not getting enough, we need you to take more of our money, please,” he said
The dinner was considered very productive for Ellison and Oracle as the company recently announced that it will create a Zettascale AI supercluster with 131,072 Nvidia GB200 NVL72 Blackwell GPUs, delivering 2.4 Zettaflops of AI performance, more powerful than Musk’s Memphis AI cluster, and xAI with 100,000 Nvidia H100 AI GPUs.
Oracle’s AI plans require a huge amount of electricity, so the company has secured permits to build three modular nuclear reactors to meet its power needs. However, deploying nuclear reactors in data centers could take years, so the company could take a page from Elon Musk and use giant mobile generators to augment local power supplies if needed.
Although smaller than other large technology companies providing data center services such as Amazon Web Services, Microsoft Azure and Google Cloud, Oracle Cloud Infrastructure (OCI) has an advantage. According to a Barron’s report, OCI offers flexibility that can meet the specific requirements of customers.
Despite its smaller size, Oracle is ramping up its investment in AI. Ellison said advanced AI models will cost $100 billion to train over the next three years, and OCI wants to be one of the leaders in AI processing.
Nvidia was founded in 1993 by three engineers, Jensen Huang, Chris Malachowsky, and Curtis Priem, who predicted that as computers evolved, they would need more sophisticated graphics processing units (GPUs). The bet paid off when they won their first major contract in 2000: manufacturing chips for Microsoft’s Xbox gaming console. Since then, Nvidia has been known for its GPUs for computers. When the cryptocurrency mining boom hit, the company benefited from the thirst for graphics cards.
In early 2022, the company announced the H100 – the most powerful GPU ever built and still one of the most expensive chips today at $40,000. At that time, experts said that Nvidia chose the wrong time to announce it, because businesses were looking to tighten spending and lay off staff. However, reality proved the opposite when OpenAI launched ChatGPT and the generative AI trend exploded in the world, making Nvidia’s chips sought after and constantly in short supply.
