OpenAI links up with Broadcom to produce its own AI chips

In a significant move to bolster its computing capabilities, OpenAI is poised to roll out its own artificial intelligence chips next year. This decision comes as the company seeks to meet the soaring demand for computational power while decreasing its reliance on Nvidia, a leading chip manufacturer. The upcoming chips will be co-designed with Broadcom, a prominent player in the semiconductor industry. Sources familiar with the partnership indicate that these chips are expected to be delivered in the coming year. Broadcom's CEO, Hock Tan, recently hinted at a new customer making a substantial commitment, worth $10 billion in orders, although he did not disclose the customer's identity. This strategy mirrors that of tech giants like Google, Amazon, and Meta, who have all developed specialized chips tailored for AI tasks. The tech sector has been experiencing explosive growth in demand for the computing resources needed to train and deploy AI models. Reports suggest that OpenAI intends to utilize these chips internally rather than releasing them to external clients. Last year marked the beginning of an initial collaboration between OpenAI and Broadcom, but the timeline for the mass production of a viable chip design had remained uncertain until now. During a recent earnings call, Tan revealed that Broadcom had welcomed a fourth major customer into its custom AI chip division, as the company reported financial results that exceeded analysts' expectations. While Broadcom typically keeps the identities of its clients under wraps, it has been confirmed that OpenAI is indeed the new addition. Both OpenAI and Broadcom have opted not to comment on the partnership.