OpenAI, Broadcom Unleash ‘Jalapeño’ Intelligence Processor

OpenAI, Broadcom Unleash ‘Jalapeño’ Intelligence Processor

OpenAI, Broadcom Unleash ‘Jalapeño’ Intelligence Processor

https://www.iphoneincanada.ca/2026/06/24/openai-broadcom-unleash-jalapeno-intelligence-processor/

Publish Date: 2026-06-24 14:11:00

Source Domain: www.iphoneincanada.ca

OpenAI and Broadcom have officially unveiled Jalapeño, a custom-built hardware chip designed from scratch to run massive artificial intelligence models.

The announcement marks a massive shift for OpenAI. Historically, the company has relied heavily on third-party graphics processing units to run its software. By moving to custom silicon, OpenAI is looking to control everything from the physical chip level all the way up to the consumer application.

The project came together with surprising speed. The companies managed to take Jalapeño from an initial blank-slate design to a completed manufacturing tape-out, which is the final design stage before factory production, in just nine months. This is widely considered an incredibly fast turnaround for high-performance advanced semiconductors.

Interestingly, OpenAI used its own existing artificial intelligence models to help automate and accelerate parts of the design and optimization process.

Early testing in the laboratory shows promising results. Engineering samples are already running active workloads at their target operational speeds and power levels. In fact, OpenAI revealed that the hardware is already running internal models, including an unreleased system named GPT-5.3-Codex-Spark. The data suggests that Jalapeño will offer performance per watt that is substantially better than the current top-tier options on the market.

To bring this platform to life, OpenAI worked closely with Broadcom for the core silicon design and high-performance networking components, including Broadcom’s Tomahawk networking silicon. They also partnered with Celestica to handle the complex engineering of the circuit boards, rack systems, and overall physical integration.

The ultimate goal of this hardware push is to make artificial intelligence more stable, affordable, and accessible.

Source