OpenAI 'Unsatisfied' With Some Nvidia Chips, Has Been Looking For Alternatives For Months: Report

Tensions have risen between Nvidia and OpenAI, driven by OpenAI's dissatisfaction with Nvidia's chips. OpenAI has been searching for alternative chip solutions since last year, particularly for AI inference tasks. Nvidia's planned $100 billion investment in OpenAI appears stalled, despite prior announcements of an agreement. CEO Jensen Huang of Nvidia has expressed concerns about OpenAI's strategy and competition. OpenAI's shift focuses on chips optimized for AI inference, a growing area of competition. This decision tests Nvidia’s dominance, as OpenAI explores options from companies like Cerebras and Groq. OpenAI's interest in alternative chips stems from the need for faster inference performance for applications like Codex. OpenAI seeks chips with increased memory integrated into the silicon for speed advantages in handling user requests. The focus on inference speed is evident in OpenAI’s new Mac app dedicated to its Codex AI coding agent.