Elon Musk & Larry Ellison’s Bold Move: 3 Secrets to Securing Nvidia AI GPUs

Elon Musk and Larry Ellison managed to get Nvidia AI GPUs for Oracle’s formidable supercomputer after having dinner with Jensen Huang, the Nvidia Company CEO.

Larry Ellison, the father of Oracle, told a surprising story during the company’s most recent earnings call. They asked Nvidia CEO Jensen Huang for more of the company’s powerful AI GPUs at a dinner at Nobu Palo Alto. “Elon and I felt like we were pleading with Jensen for GPUs.” We even begged you to take more of our money. While on the call, Ellison made a joke. In the end, what they asked for worked.

Oracle has since stated its desire to commence the construction of a massive AI system a Zettascale AI supercluster. That will incorporate 131,072 of Nvidia’s newest GB200 GPUs in it. This setup will give AI better performance than Musk’s xAI’s which Memphis Supercluster has 100,000 Nvidia H100 GPUs now.

Oracle is investing money into AI in addition to purchasing three portable nuclear reactors to power its data centers. However, as nuclear power may not be available immediately for Oracle, Musk suggested that Oracle may opt to make up for the shortfall by bringing mobile generators.

Even though Oracle isn’t as big as Amazon, Microsoft, or Google in the cloud, it has a big edge. Barron’s says that Oracle Cloud Infrastructure (OCI) gives customers with specific needs more security and flexibility choices, such as offline servers for better protection.

Another thing Ellison said was that the price is going up because AI is getting better. He also said that it might cost $100 billion over the next few years to teach the next crop of AI models what they need to know. Oracle has put a lot of money into the “AI race,” and Ellison really thinks that being “first” is important. In short, he said, “Everyone will be talking about it: it will be a timed operation where one party gets an edge.”