OpenAI has reportedly partnered with Broadcom to create its own AI processors, signaling a shift away from reliance on Nvidia’s GPUs. Reports suggest that this collaboration involves a significant deal where Broadcom has secured a substantial $10 billion in AI system orders.

Background

The relationship with Broadcom is motivated by OpenAI’s need to reduce costs associated with using Nvidia chips for model training and inference. This decision highlights the evolving landscape of AI hardware supply chains, where using proprietary processors may ultimately lower operational costs.

“This is a strategic move that reflects the need for more sustainable and cost-effective solutions in the AI sector,” said Hock Tan, the CEO of Broadcom. “Esta es una decisión estratégica que refleja la necesidad de soluciones más sostenibles y rentables en el sector de la IA.”

While OpenAI’s current systems rely heavily on Nvidia’s technology, this partnership signifies a step towards developing self-sufficient capabilities in AI processing.

Conclusion

As OpenAI moves towards building its custom processors, it remains uncertain how this will impact the pricing and availability of AI technologies they currently provide.