OpenAI Turns to Broadcom as Musk, Zuckerberg Snatch Up Nvidia AI Chips

Points

• OpenAI is in talks with Broadcom and other chip makers to expand its AI operations.
• AI models require massive clusters of GPUs for training, with Nvidia’s H100 being the most popular.
• Elon Musk’s xAI and Mark Zuckerberg’s Meta are also competing for Nvidia GPUs.

OpenAI is reportedly in talks with Broadcom and other chip makers in an urgent effort to expand its artificial intelligence operations. AI models like OpenAI’s ChatGPT and Meta’s Llama typically require massive clusters of graphical processing units （GPUs） or similar computer chips for training, with Nvidia’s H100 being the most popular.

Cointelegraph

Increasing Demand for AI Chips

The H100 can cost anywhere from $15,000 to $30,000, depending on the number purchased and current market conditions. Training a single model can require tens of thousands of these AI chips, with even more needed for larger, more robust systems. Despite OpenAI’s leading position in the generative AI space, much of its hardware is owned by its partner Microsoft.

Broadcom Partnership

According to a report from The Information, OpenAI is in discussions with Broadcom and other chip makers to develop its own chip. While OpenAI has not confirmed the report, it has acknowledged ongoing conversations with industry and government stakeholders about increasing access to infrastructure.

Altman’s Funding Quest

Back in February, The Wall Street Journal reported that OpenAI CEO Sam Altman was actively seeking investors to raise a $5 trillion- $7 trillion war chest for developing more chips. Altman has met with government officials, private investors, and corporate leaders to discuss the initiative.

Competition from Musk and Zuckerberg

Part of the urgency for OpenAI’s chip expansion is driven by competition from Elon Musk’s xAI and Mark Zuckerberg’s Meta.

Both Musk and Zuckerberg have pledged to purchase as many Nvidia GPUs as possible, potentially contributing to the shortage and rising prices. In January, Meta signaled its intent to bring its GPU count up to 350,000 by the end of 2024, while Musk has gone so far as to redirect 12,000 GPUs intended for Tesla to xAI instead. Musk claimed the GPUs weren’t needed by the automaker at the time, but shareholders subsequently sued.

Conclusion

OpenAI’s discussions with Broadcom and other chip makers underscore the intense competition for AI hardware as companies race to develop and deploy advanced AI models. The collaboration could help OpenAI secure the necessary infrastructure to maintain its leadership position in the generative AI space.

解説

• AI Hardware Demand: The high demand for GPUs highlights the critical role of hardware in training and deploying advanced AI models. Companies like OpenAI, Meta, and xAI are competing to secure these resources.
• Strategic Partnerships: OpenAI’s potential partnership with Broadcom could diversify its supply chain and reduce dependency on Nvidia, ensuring access to the necessary hardware for AI development.
• Market Impact: The competition for AI hardware and the involvement of major tech companies underscore the significance of AI in the future of technology, driving innovation and investment in the sector.