**Broadcom to Develop Custom AI Chips for OpenAI**
*By Dwaipayan Roy | Oct 13, 2025, 07:54 PM*
OpenAI has announced a strategic partnership with Broadcom to develop its first in-house artificial intelligence (AI) processors. This move is part of OpenAI’s plan to keep pace with the rapidly growing demand for its AI services.
Under the agreement, OpenAI will take the lead in designing these custom chips, while Broadcom will handle their development and deployment starting in the second half of 2026.
**Massive Energy Consumption**
The upcoming custom AI chips are projected to consume a total of 10 gigawatts (GW) of power. To put this into perspective, that’s enough electricity to power over eight million U.S. households — roughly five times the output of the Hoover Dam.
This deal highlights the massive investments being poured into AI chip technology across the tech industry, reflecting an ongoing race to build computing systems capable of matching or surpassing human intelligence.
**Expanding Strategic Partnerships**
This partnership follows closely on the heels of OpenAI’s recent 6GW AI chip supply agreement with AMD, which also includes an option for OpenAI to acquire a stake in the chipmaker. Additionally, NVIDIA recently announced plans to invest up to $100 billion in OpenAI and supply data-center systems with at least 10GW of capacity.
“Partnering with Broadcom is a critical step in building the infrastructure needed to unlock AI’s potential,” said Sam Altman, CEO of OpenAI.
**Aligning with Industry Leaders**
By teaming up with Broadcom, OpenAI aligns itself with major cloud-computing giants like Google and Amazon, both of which are developing custom AI chips to address soaring demand and reduce reliance on NVIDIA’s high-cost, limited-availability processors.
However, some similar initiatives by companies like Microsoft and Meta have faced delays or failed to match NVIDIA’s performance, highlighting the challenges in developing competitive AI hardware.
**Deployment Timeline and Technical Edge**
The deployment of these new AI chips is expected to be completed by the end of 2029. This effort will build upon OpenAI’s existing co-development and supply partnerships.
The new systems will be scaled entirely using Broadcom’s Ethernet and networking technologies, potentially giving them an edge over competitors such as Marvell Technology. Moreover, this strategy aims to challenge NVIDIA’s dominant InfiniBand networking solutions in the AI data-center market.
—
OpenAI’s collaboration with Broadcom marks a significant milestone in the evolving AI hardware landscape, signaling robust growth and fierce competition in the race to power next-generation artificial intelligence.
https://www.newsbytesapp.com/news/science/openai-partners-with-broadcom-to-develop-custom-ai-chips/story