OpenAI Has Unveiled Its First AI Chip

OpenAI and Broadcom have unveiled their first custom artificial intelligence chip, marking the ChatGPT maker’s move into designing its own silicon as it seeks to reduce reliance on external suppliers and expand its computing capacity.

The chip, called Jalapeño, will be manufactured by Broadcom and used by OpenAI for inference, the computationally intensive process of running AI models for users across ChatGPT and other applications.

OpenAI President Greg Brockman said the chip was designed from end to end in nine months with assistance from the company’s own AI models.

“The degree to which our models have been able to accelerate it was very surprising to us,” Brockman told CNBC.

The launch is part of OpenAI’s broader effort to “build the full stack behind its models and products” as demand for artificial intelligence services continues to surge.

“By designing more of the stack ourselves, we can serve more intelligence with greater efficiency and keep pushing advanced AI toward broader access,” Brockman said in a statement announcing the chip.

Broadcom has emerged as one of the major beneficiaries of the generative AI boom, helping large cloud providers and AI companies develop custom chips designed for specialised workloads. The company’s shares have risen sharply as investors bet on continued demand for AI infrastructure.

OpenAI has faced growing pressure to secure more computing resources as demand for its AI models expands. Since the launch of ChatGPT in 2022, the company has been one of the largest buyers of Nvidia’s graphics processing units (GPUs), the dominant hardware used to train AI systems and handle large-scale workloads.

The company has since expanded partnerships beyond Nvidia, including a deal with Amazon Web Services for access to its Trainium AI chips, as well as agreements with Advanced Micro Devices and AI chipmaker Cerebras.

OpenAI and Broadcom announced in October that they had been working together for 18 months on plans to develop and deploy racks of OpenAI-designed chips, with a long-term goal of building systems requiring 10 gigawatts of power.

The Jalapeño chip is an application-specific integrated circuit (ASIC), a type of processor designed for particular tasks. While ASICs are generally less flexible than Nvidia’s GPUs, they can be more efficient and cheaper when optimised for specific AI workloads.

The companies described Jalapeño as an “Intelligence Processor” and the first “AI accelerator” in a platform they are developing “to make advanced AI faster, more reliable, and more accessible to more people.”

A physical sample of the chip was delivered to OpenAI today, with initial deployment targeted for the end of 2026 and broader expansion planned in the following years.

Broadcom CEO Hock Tan said the project would begin with limited prototype development before scaling up.

“We will start seeing it really ramp up in ’27 and really going full tilt in first half ’28,” Tan said.

Tan said demand for computing capacity from Broadcom’s major customers remained extremely strong.

“It’s just much more than we can address,” he said, “and this is not just ’26, not ’27, we’re seeing that same and even elevated demand in ’28 as well.”

OpenAI has said developing more of its own hardware is intended to improve efficiency and increase access to advanced AI systems as global competition for computing resources intensifies.

Stay ahead of the stories shaping our world. Subscribe to Impact Newswire for timely, curated insights on global tech, business, and innovation all in one place.

Dive deeper into the future with the Cause Effect 4.0 Podcast, where we explore the ideas, trends, and technologies driving the global AI conversation.

Got a story to share? Pitch it to us at info@impactnews-wire.com and reach the right audience worldwide