Bernstein Estimates OpenAI's Compute Spending to Reach $614B by 2030
Bernstein analysts estimate that OpenAI's cumulative compute infrastructure spending could reach $614 billion by 2030 based on disclosed contracts, which is lower than the previously rumored $750 billion. The analysis shows that annual spending is projected to peak at $184 billion by 2031 as facilities gradually become operational. This projection provides crucial financial context on the immense scale of artificial intelligence investments and multi-year commitments required by leading foundation model providers. It highlights the vast capital deployment flowing into major cloud providers while signaling potential long-term pressures for specialized cloud vendors. OpenAI's disclosed infrastructure agreements include approximately $300 billion with Oracle, $250 billion with Microsoft, $138 billion with Amazon, and up to $22.4 billion with CoreWeave, spanning terms of five to eight years. Actual spending may vary due to construction delays, project cancellations, or hardware adjustments.
## BACKGROUND
CoreWeave is an American AI-native cloud computing company specializing in high-performance GPU infrastructure for artificial intelligence and machine learning workloads. Such specialized data centers are vital for training and running advanced generative AI models like large language models.