According to a U.S. tech news website, OpenAI is projected to shell out US$115 billion in cash by 2029, which is an increase of US$80 billion compared to previous estimates. Moreover, as reported by the UK’s Financial Times on the 6th, OpenAI plans to team up with U.S. semiconductor behemoth Broadcom to kick off mass production of its self-developed AI chips. Multiple sources with knowledge of the situation disclosed that the chips co-designed by the two parties will be ready for delivery next year. At present, OpenAI heavily leans on off-the-shelf GPUs from NVIDIA. Nevertheless, with the burgeoning demand from AI companies for training intricate models, chip makers and data center operators are finding it hard to keep up with market demands. The Financial Times analysis posits that OpenAI’s collaboration with Broadcom signals the industry’s shift toward seeking customized alternatives to NVIDIA’s chips.