Oracle's 97.9% Utilization Rate Sparks Fears of Nvidia Shortage
Oracle's latest earnings report revealed that its AI chips ran at a staggering 97.9% utilization rate in the first quarter of fiscal 2027, with nearly every chip busy serving customers.
The company delivered 850 megawatts of AI capacity to customers in Q1, containing over 300,000 graphics processing units (GPUs), and most of these chips are Nvidia's.
Oracle's GPU fleet is massive, with hundreds of thousands of Nvidia GPUs connected to its cluster design for the largest AI deployments, targeting up to 800,000 chips. The company's campus in Abilene, Texas, which serves OpenAI, received 131,000 GPUs in Q1 alone.
The utilization rate indicates that demand for these high-performance computing resources is extremely high, with customers paying a premium to keep even four-year-old chips running at capacity.