In Oracle’s latest earnings report, Larry Ellison, the company’s founder, admitted that he and Elon Musk practically begged Nvidia’s CEO, Jensen Huang, to provide the company with the GPUs they needed.
According to the BARRON’S website, Larry Ellison apparely had a dinner with the CEO of Nvidia not long ago, and of course he was not alone; Elon Musk was also prese at this banquet, and ierestingly, he also desperately needed Nvidia artificial ielligence GPUs. Ellison says:
“At Nobu Palo Alto, I had dinner with Elon Musk and Jensen Huang. I describe this dinner as Elon and I begging Jensen to buy Nvidia GPUs. (we used to say) please take our money; No, take more money, that’s not enough; Please accept more money. (Of course) the good work we ahead and paid off.”
Oracle’s Zeta-scale supercluster project is powered by Nvidia chips
It seems that this dinner really worked and Oracle offered the CEO of Nvidia good money so that he can get a lot of GPUs. Oracle recely announced that it will build a Zeta-scale AI supercluster consisting of 131,072 Nvidia GB200 NVL72 Blackwell GPUs, delivering 2.4 ZettaFLOPS; A much more powerful performance than the Memphis xAI supercluster, which currely has 100,000 Nvidia chips.
Oracle’s AI applications require an incredible level of power; For this reason, the company has already received the license to build 3 modular nuclear reactors to meet the electrical needs of its facilities. However, since deploying nuclear reactors in data ceers is likely to take several years, the company could follow Musk’s AI plans and use massive mobile generators to boost the local power supply if needed.
Despite being smaller than other major web services companies such as Amazon, Microsoft and Google, Oracle Cloud Infrastructure (OCI) has a distinct advaage over these gias. According to Barron’s, OCI provides greater flexibility and can meet specific requiremes that some customers request. It can even offer offline servers running on its own network infrastructure to maximize customer security.
