OpenAI and Broadcom Launch Jalapeño AI Inference Chip

OpenAI has unveiled its first custom AI chip, a significant milestone in the company's strategy to control more of its technology infrastructure and reduce dependence on external suppliers.

The chip, named Jalapeño, was developed in partnership with Broadcom and will be manufactured by the semiconductor company. OpenAI designed the chip specifically for inference workloads—the compute-intensive process of running AI models to serve responses to users in ChatGPT and other applications.

Why it matters

Custom silicon represents a strategic shift for OpenAI from pure software development to vertical integration across its technology stack. By designing chips tailored to its specific workloads, the company can potentially reduce costs, improve performance, and gain more control over its supply chain at a time when demand for AI computing capacity far exceeds available infrastructure. This move also signals intensifying competition in the AI chip market, where Nvidia has dominated but now faces challenges from custom solutions developed by major AI labs.

Nine-month development cycle

OpenAI completed the chip design in just nine months, an unusually fast timeline for custom silicon development. The company also designed substantial portions of the computer systems where the chips will operate, according to the announcement.

Broadcom will deliver a physical sample of Jalapeño to OpenAI on Wednesday, with initial deployment targeted for the end of 2026. The companies plan to expand the chip's use in subsequent years.

ASIC approach versus GPUs

Jalapeño is an application-specific integrated circuit (ASIC), a chip architecture that differs fundamentally from the graphics processing units (GPUs) that currently dominate AI computing. Industry experts note that ASICs offer less flexibility than Nvidia's GPUs but come at lower cost and can be optimized for specific tasks.

OpenAI and Broadcom are marketing Jalapeño as an "Intelligence Processor" and positioning it as the first "AI accelerator" in a broader platform designed to make advanced AI faster, more reliable, and more accessible.

Diversifying chip suppliers

Since launching ChatGPT and igniting the generative AI boom in 2022, OpenAI has been among the largest purchasers of Nvidia GPUs. However, explosive demand growth has pushed the company to secure alternative silicon sources.

Earlier this year, OpenAI established agreements with Amazon Web Services to use Trainium AI chips, signed deals with AMD, and partnered with Cerebras, which completed its initial public offering in May.

The collaboration with Broadcom became public in October after 18 months of joint development. The partnership aims to eventually deploy enough chip capacity to require 10 gigawatts of power.

Market impact

Broadcom shares rose approximately 2% following the announcement. The chipmaker has emerged as a major beneficiary of the AI boom by enabling hyperscalers and frontier AI labs to develop custom silicon. Broadcom's stock has increased roughly sevenfold since the end of 2022 and is up 10% in 2026.

These details were first reported by CNBC.