OpenAI is getting ready to provoke mass manufacturing of its personal AI chips in 2026, marking a step in direction of assembly the rising demand for computing energy whereas reducing its dependence on Nvidia, reported The Monetary Instances.
The brand new chip, developed in partnership with Broadcom, a US semiconductor firm, is predicted to be shipped within the coming yr, a number of sources accustomed to the collaboration advised the publication.
Broadcom’s CEO, Hock Tan, not too long ago talked about {that a} new undisclosed buyer that has dedicated to $10bn in orders.
This transfer aligns OpenAI with different expertise corporations equivalent to Google, Amazon, and Meta, which have additionally created their very own specialised chips to deal with AI workloads.
In line with insiders, OpenAI intends to utilise the chips internally moderately than providing them to exterior purchasers.
The partnership with Broadcom started in 2024, though the timeline for the mass manufacturing of a profitable chip design had not been beforehand clarified.
Throughout a name with analysts, Tan revealed that Broadcom had secured a fourth buyer for its customized AI chip division.
Whereas Broadcom doesn’t disclose the identities of its purchasers, sources confirmed that OpenAI is the brand new buyer.
Each corporations have chosen to not touch upon the matter, the report stated.
Tan famous that this settlement has enhanced Broadcom’s progress outlook by producing “quick and pretty substantial demand,” with plans to start transport chips for this buyer “fairly strongly” from subsequent yr.
Beforehand, Broadcom collaborated with Google to develop its customized “TPU” AI chips.
OpenAI’s CEO, Sam Altman, has constantly highlighted the necessity for elevated computing energy to accommodate the rising variety of companies and shoppers utilising merchandise equivalent to ChatGPT, in addition to for coaching and executing AI fashions.
In August 2025, Altman indicated that the corporate is prioritising computing assets “in mild of the elevated demand from [OpenAI’s latest model] GPT-5” and goals to double its computing fleet “over the subsequent 5 months.”
Join our each day information round-up!
Give your enterprise an edge with our main business insights.
