OpenAI has officially announced its entry into the AI chip sector, collaborating with chip giant Broadcom to develop its first custom AI inference chip. This strategic move not only signifies a significant step towards supply chain autonomy for OpenAI but also directly challenges NVIDIA's dominant position in the current AI chip market.
According to Reuters, OpenAI has assembled a team of approximately 20 specialized R&D professionals, including a chief engineer who previously worked on Google's Tensor processor. Broadcom will be responsible for chip design, with TSMC handling manufacturing, and mass production is expected to begin in 2026. Notably, OpenAI had previously considered establishing a chip foundry but ultimately opted for a chip design strategy due to high costs.
Image source note: The image is generated by AI, authorized by Midjourney
Two key drivers underpin OpenAI's move: alleviating the shortage of AI chips and reducing operational costs. CEO Sam Altman has publicly stated that the company faces dual pressures of processor shortages and high hardware costs. As one of NVIDIA's largest GPU buyers, OpenAI has relied since 2020 on a Microsoft supercomputer equipped with 10,000 NVIDIA GPUs to develop its generative AI technology.
To reduce reliance on NVIDIA, OpenAI has begun exploring a diversified supply strategy, planning to use AMD chips for model training through the Microsoft Azure cloud platform. This decision is closely related to AMD's breakthrough progress with its MI300AI chip launched last year.
Although this collaboration may cause dissatisfaction at NVIDIA, OpenAI has stated it will continue to maintain good relationships with existing chip manufacturers, particularly regarding the use of new-generation Blackwell chips. Following the announcement, Broadcom's stock price rose, and AMD continued its upward trend, indicating a positive market outlook for intensified competition in the AI chip sector.
This strategic shift marks OpenAI's entry into the self-developed chip camp of tech giants like Meta and Google. However, facing mature products like Google's TPU and Microsoft's Maia100, OpenAI will need to invest heavily to establish a foothold in the AI chip market. The market generally believes that this move will reshape the AI chip supply landscape and promote healthy industry competition.