San Francisco: OpenAI has planned to launch its first artificial intelligence chip in 2026 in partnership with US semiconductor firm Broadcom. The chip has been designed for OpenAI’s internal use, rather than for commercial distribution, as the company seeks to strengthen its computing infrastructure.
The ChatGPT developer has relied heavily on substantial computing power to train and operate its AI models. Developing an in-house chip has been seen as a strategic move to diversify supply sources and reduce costs. Sources have indicated that OpenAI is in the final stages of chip design and is expected to send it for fabrication to Taiwan Semiconductor Manufacturing Company (TSMC).
Broadcom Chief Executive Officer Hock Tan has said that the company expects its artificial intelligence revenue for fiscal 2026 to improve significantly. Hock Tan confirmed that Broadcom has secured more than $10 billion in AI infrastructure orders from a new customer, without naming it. Market analysts believe OpenAI is among the key clients contributing to this demand.
The development has come after months of exploration by OpenAI to secure reliable chip supply amid growing competition in the generative AI space. The firm has evaluated multiple strategies to strengthen computing capacity, with custom chips now forming a cornerstone of its long-term infrastructure plan.
Earlier this year, Tan had noted that Broadcom was working with several major companies to design custom chips. With OpenAI now moving forward as a confirmed customer, the collaboration is expected to enhance Broadcom’s role in the expanding AI hardware market.
The global demand for AI-focused semiconductors has risen rapidly, driven by applications such as natural language processing, computer vision, and generative AI. For OpenAI, the move to develop its own chip marks a milestone in reducing dependency on third-party suppliers and ensuring more sustainable growth.
MOST READ | Malaysia urges TikTok to enforce age verification for minors
