OpenAI exploring developing its own AI chips: Report

OpenAI, the company behind ChatGPT, is considering the development of its own artificial intelligence (AI) chips and has explored potential acquisition opportunities, as reported by Reuters. 

According to sources familiar with the company's plans, OpenAI has been addressing the challenge of acquiring expensive AI chips, a crucial component for its operations. While no final decision has been made, discussions on possible solutions have been ongoing since last year. 

Options considered include creating its own AI chips, forming closer partnerships with chip manufacturers like Nvidia, and diversifying its suppliers beyond Nvidia. CEO Sam Altman has emphasised the need to acquire more AI chips, expressing concerns over the scarcity of graphics processing units (GPUs), a market predominantly controlled by Nvidia.  

As per the report, the pursuit of additional chips is motivated by two primary concerns identified by Altman: a shortage of advanced processors to power OpenAI's software and the significant expenses associated with maintaining the hardware required for its projects and products. 

Since 2020, OpenAI has relied on a massive supercomputer, constructed in partnership with Microsoft, which utilises 10,000 of Nvidia's GPUs. Operating ChatGPT is notably costly, with each query costing approximately 4 cents, Reuters reported. If ChatGPT's queries were to reach a scale one-tenth that of Google search, it would initially require approximately $48.1 billion worth of GPUs and around $16 billion worth of chips annually to remain operational.
 
OpenAI's potential pursuit of developing its own AI chips would align it with a select group of major tech companies, including Google and Amazon, who have opted to design their own chips to meet their specific needs. Meta has also faced challenges with its custom chip efforts, leading to the abandonment of some AI chips. The company is now working on a new chip intended for various AI applications. 
 
In April, Reuters reported that Microsoft is also in the process of developing its artificial intelligence chip, codenamed "Athena," to power AI chatbots like ChatGPT. These chips will play a critical role in training large-language models and supporting inference, both essential for generative AI like ChatGPT, which processes extensive data, identifies patterns, and generates human-like conversation.