The Dazzling Negotiations Between Oracle and Tesla Over Nvidia's GPUs

The focus of a recent dinner in Palo Alto was on two of the world's wealthiest individuals — Larry Ellison and Elon Musk. Host Jensen Huang, CEO of Nvidia, witnessed an unusual scene: Ellison and Musk formally asked for more GPUs, which are crucial for their advanced data centers and artificial intelligence projects. Under Ellison's leadership, Oracle has dedicated itself to building ultra-fast and cost-efficient data centers, with plans to potentially increase its current 162 data centers to over 2,000. The company has developed RDMA network technology that enables faster data movement, thus reducing the cost of computing power usage. These infrastructural advantages attract renowned AI startups like OpenAI, Cohere, and Musk's xAI. Nvidia supplies the industry's best graphics cards, but currently, high demand exceeds supply. This also affects Tesla, led by Musk, who plans to optimize his self-driving vehicles with a cluster of 50,000 GPUs. Meta Platforms is also entering the AI race with plans to expand the capacity of its LLM model to 600,000 GPUs. While Oracle impressed with a 45% revenue increase in the Oracle Cloud Infrastructure segment in the last quarter of fiscal year 2025, growth potential is limited by the constrained availability of GPUs. Competition for these scarce resources is fierce – aside from Oracle and Tesla, Microsoft, Amazon, and Alphabet are also vying for the coveted Nvidia components. Oracle and Tesla are planning significant investments in their infrastructure. While Oracle aims to double its capital expenditures for the next year, Tesla plans to spend over $10 billion. However, these figures pale in comparison to the gigantic investments by Microsoft and Amazon. Nvidia greatly benefits from this demand, recently recording a 154% increase in data center revenue to $26.3 billion. The trend seems to continue – Ellison predicts a decade full of AI investments and technological arms races.