OpenAI on Wednesday launched its first “intelligence processor” called “Jalapeño”, marking its entry into the $13.5 trillion chip market dominated by Nvidia. Developed in partnership with Broadcom, the chip is being advertised as the first AI accelerator in what will expand into a multi-generation compute lineup for the coming years.
The designing part for the chip was undertaken by OpenAI. The company said it used its experience and understanding of LLM fundamentals to support the upcoming advanced versions of AI models more efficiently. Broadcom worked with Canada-based electronics manufacturing company Celestica to handle the hardware design and scalable production.
OpenAI said the architecture of the chip is built to optimize data flow and resource balance for near-peak performance.
“By designing more of the stack ourselves, we can serve more intelligence with greater efficiency and keep pushing advanced AI toward broader access,” said Greg Brockman, President and Co-Founder of OpenAI.
The Sam Altman-led AI company is still testing Jalapeño to measure the final performance. The initial testing said the chip’s per watt substantially has shown to be better than the current state-of-the-art system.
“The architecture reduces data movement and balances compute, memory, and networking resources to achieve realized utilization much closer to theoretical peak performance. Broadcom’s silicon implementation and networking technologies, including Tomahawk networking silicon, help bring the platform to large-scale production,” OpenAI explained.
The development comes just weeks after OpenAI submitted a confidential S-1 filing for a review by the SEC. The company is aiming to sell shares at $135 each and raise about $75 billion.
Ahead of the IPO, the company is arming its team with high-profile AI experts. Just last week, former White House AI advisor Dean Ball announced his entry into the company to lead its frontier lab initiative dubbed Strategic Futures.
OpenAI claims that in the coming future it wishes to make AI inference accessible to more students, developers, businesses, and researchers.
Last year, OpenAI shifted from a non-profit model to a for-profit one citing the need for capital in order to advance its AI offerings. Elon Musk, who was one of the original foudners of OpenAI, called the move against OpenAI’s foundational values of making AI to help humans. After a high stakes court battle, a federal jury tossed out Musk’s case in May.
During the course of the trial, Altman maintained that OpenAI needs funds to go forward with its R&D for advanced LLMs — which was becoming difficult under the non-profit model.
With its chip initiative, OpenAI has outpaced rivals like Anthropic to go native in chip designing breaking free from the rat race. It now essentially enters the league tech giants like Google, Amazon, and Meta—all of whom use custom silicon to break Nvidia’s monopoly.
