Story perspectives

OpenAI and Broadcom Invest $10B in Custom AI Chips

9/8/2025

Summary

Story summary

OpenAI and Broadcom are collaborating to create custom AI chips, investing $10 billion to lessen dependence on Nvidia's GPUs.
The partnership aims to enhance efficiency and reduce costs for AI workloads.
The first chips are set to launch in 2026, potentially transforming the AI hardware sector.
Following the announcement, Broadcom's stock rose nearly 11%, indicating strong investor confidence.
OpenAI's computing power spending is projected to reach $115 billion from 2025 to 2029, emphasizing infrastructure needs.

History

Historical Context

Nvidia's launch of its first deep learning GPU in 2011 transformed AI processing, enabling the company to dominate over 90% of the GPU market by 2023. This shift prompted OpenAI and Broadcom to invest $10 billion to diversify AI hardware sources, indicating a significant competitive evolution in the AI landscape that may reshape future market dynamics and innovation.

1 of 2

Story summary

OpenAI and Broadcom are collaborating to create custom AI chips, investing $10 billion to lessen dependence on Nvidia's GPUs.
The partnership aims to enhance efficiency and reduce costs for AI workloads.
The first chips are set to launch in 2026, potentially transforming the AI hardware sector.
Following the announcement, Broadcom's stock rose nearly 11%, indicating strong investor confidence.
OpenAI's computing power spending is projected to reach $115 billion from 2025 to 2029, emphasizing infrastructure needs.

1 / 2