OpenAI, the organisation behind ChatGPT, is reportedly contemplating the development of its own artificial intelligence chips to counter the ongoing shortage of advanced processors essential for its operations. According to sources familiar with the matter, OpenAI has even considered potential acquisition targets in its quest to secure a more stable supply chain.
Although no final decision has been made, internal discussions within OpenAI have revolved around tackling the scarcity of high-performance chips that are crucial for running their AI applications efficiently. The shortage, especially of graphics processing units (GPUs), has been a significant concern for CEO Sam Altman, who has emphasised the pressing need for more AI chips. Nvidia, a dominant player in the GPU market, currently controls over 80% of the global market share for chips tailored for AI applications.
OpenAI’s dependence on GPUs, specifically in running ChatGPT, has proven to be costly. Analysts estimate that each query on ChatGPT costs approximately 4 cents. If the scale of ChatGPT queries were to reach a fraction of Google search, it would necessitate billions of dollars worth of GPUs initially and an ongoing annual investment of billions in chips to sustain operations.
The prospect of developing custom AI chips places OpenAI among a select group of tech giants, including Google and Amazon, who have ventured into designing their own chips to align with their unique business needs. While this move could offer OpenAI greater control and potentially reduce costs in the long run, it would also represent a substantial strategic initiative and a significant financial investment, potentially reaching hundreds of millions of dollars annually.
The company has considered the possibility of acquiring an existing chip company to expedite its custom chip development efforts, a strategy previously employed by Amazon when it acquired Annapurna Labs in 2015. However, the exact identity of the potential acquisition target remains undisclosed.
Developing custom AI chips is a complex endeavour that demands extensive resources and expertise. Even if OpenAI commits to this initiative, it would take several years to materialise, leaving the company reliant on commercial providers like Nvidia and Advanced Micro Devices in the interim.
While other major tech companies, such as Meta, have faced challenges in their custom chip efforts, OpenAI’s pursuit reflects the growing demand for specialised AI chips, particularly since the launch of ChatGPT. Microsoft, one of OpenAI’s primary backers, is also reportedly working on a custom AI chip that is being tested by OpenAI, signalling potential developments in the near future.
OpenAI’s exploration of custom AI chips underscores the organisation’s commitment to addressing supply challenges and ensuring the continuity of its innovative AI applications, although the final path forward remains to be seen.