OpenAI dismisses reports of large-scale adoption of Google’s AI chips

OpenAI has officially responded to recent speculations regarding the potential large-scale adoption of Google’s proprietary artificial intelligence chips. Reports from various media outlets, including Reuters, had suggested that the AI research organization was planning to utilize Google’s Tensor Processing Units (TPUs) to address its growing computational requirements. However, a representative from OpenAI, the entity behind the popular ChatGPT, clarified on Sunday that while they are engaged in preliminary testing with Google’s TPUs, there are no current plans to implement them widely. The spokesperson firmly stated, "We have no plans to adopt TPUs broadly at this stage." In the realm of AI, it is not uncommon for companies to experiment with various hardware setups. However, transitioning to a new chip infrastructure across operational systems typically necessitates extensive architectural modifications and software adjustments, a process that demands considerable time and resources. At this moment, OpenAI primarily depends on Nvidia’s graphics processing units (GPUs), which are regarded as the industry standard for AI applications. Additionally, the company is also leveraging advanced chips from AMD to satisfy the increasing demand for computational power. Simultaneously, OpenAI is advancing the development of its own AI chip, which is anticipated to enter the critical "tape-out" phase later this year, marking the point at which the chip design is finalized for manufacturing. In a related development, earlier this June, it was reported that OpenAI had engaged Google Cloud services, hinting at an unexpected collaboration between the two tech rivals. Nonetheless, sources indicate that a significant portion of OpenAI’s cloud computing operations still relies on servers managed by CoreWeave, a rapidly growing company providing GPU-powered solutions. Meanwhile, Google has recently started offering its custom-designed TPUs to external clients, moving beyond their previous exclusive internal applications and attracting major customers including Apple, as well as AI startups like Anthropic and Safe Superintelligence, both founded by former OpenAI executives.