OpenAI Affirms No Plans To Use Google-made Chips

OpenAI has no active plans to use Google-made chips (photo; x @notreload_ai)

JAKARTA - OpenAI stated that it had no active plans to use Google's chips in operating its products, just two days after reports from Reuters and other media stated that the artificial intelligence laboratory was considering using an AI chip from its competitors to meet growing demand.

In a statement on Sunday, June 29, an OpenAI spokesperson said that although they were conducting preliminary testing on Google's TPU's Tensor Processing Unit (TPU), there were currently no plans to implement it on a large scale. Google declined to comment on this.

Testing of various types of chips by AI laboratories is commonplace. However, the implementation of new chips on a large scale takes longer because it requires different architectures and software support. Currently, OpenAI is actively using Nvidia's graphics chip (GPU) and AMD's AI chips to meet their growing computing needs.

OpenAI is also developing its own chips, which are targeted to reach the milestone of this year's 'tape-out', which is a stage when chip designs are completed and sent for manufacturing processes.

Earlier, Reuters exclusively reported that OpenAI had signed up to use Google Cloud services to meet its growing computing capacity needs.

This collaboration was surprising because it involved two major competitors in the AI sector. Even so, most of the computing power used by OpenAI will still come from the GPU server owned by the company "neocloud" CoreWeave.

Google itself has recently expanded the availability of its internal AI chip (TPU) for external parties, which was previously only used for the company's internal needs. This move has succeeded in attracting a number of large clients, including Apple, as well as several startups such as Anthropic and Safe Super Intelligence "two competing companies ChatGPT" founded by former OpenAI officials.