OpenAI has entered into a strategic partnership with Broadcom to develop its first proprietary artificial intelligence chips—a move that, according to analysts at NEWSCENTRAL, could radically shift the balance of power in AI infrastructure. The project involves creating specialized processors with a total capacity of approximately 10 gigawatts—equivalent to the energy consumption of eight million American homes. The first deliveries are expected in the second half of 2026, with full deployment by 2029.
According to Freddy Miller, Senior Analyst at NEWSCENTRAL, OpenAI aims not only to reduce its dependence on Nvidia but also to build its own technological vertical—from models to hardware. “OpenAI is doing what once seemed impossible for a software company: turning computing into a strategic asset,” the expert notes.
Under the plan, Broadcom will handle chip design and production, as well as Ethernet-based network infrastructure, challenging Nvidia InfiniBand systems. If the companies achieve comparable performance, this would become one of the largest technological breakthroughs in recent years.
The market has already responded: Broadcom shares rose more than 10%. However, as NEWSCENTRAL emphasizes, such a scale requires massive investments and carries high risks. The timeline through 2029 appears ambitious. Even with support from Microsoft and strategic capital, OpenAI will need to balance innovation with the sustainability of its financial model.
Simultaneously, OpenAI is diversifying supply by signing an agreement with AMD for 6 gigawatts of computing capacity. This demonstrates that the company is building a multi-layered ecosystem where its own chips will coexist alongside partner solutions.
According to NEWSCENTRAL, the OpenAI-Broadcom project is not an attempt to instantly displace Nvidia but a strategic move for the next decade. It sets a new trend in which AI model manufacturers begin to control the physical foundation of their computing.
The outcome of this deal could result not only in cost reduction and increased efficiency for OpenAI but also in a reconfiguration of the entire accelerator market. According to our analysts, in a world where intelligence becomes infrastructure, those who control not only the code but also the silicon will win.