Jakarta, INTI - On Wednesday, June 24, 2026, Open AI announced its new AI processor chip, named Jalapeno. This is part of the company's effort to decrease its reliance on Nvidia-made graphics processing units (GPUs).
Developed in partnership with Broadcom, Jalapeno is specifically designed to power current and future generations of Large Language Models (LLMs) running on AI servers.
Jalapeno is not a general-purpose processor. This chip falls into the category of ASIC (Application-Specific Integrated Circuit), meaning it is designed for AI Inference.
AI Inference is the process by which an AI model responds to a user's commands or questions (prompts), such as when ChatGPT types an answer or when Codex writes programming code. Jalapeno is specifically designed to speed up this response (inference) process.
Jalapeno's arrival comes just nine months after OpenAI first announced its partnership with Broadcom. This step is crucial for OpenAI, given that Nvidia GPU supply on the global market is currently very limited due to high demand from various technology companies.
Tech Companies are Developing Their Own Chips
In an interview with Reuters, Broadcom CEO Hock Tan confidently stated that the Jalapeno chip's performance is capable of matching the power of Nvidia's latest-generation Blackwell chips and Google's Tensor Processing Units (TPU).
OpenAI's move also joins the growing list of tech giants racing to design their own silicon chips. Microsoft, Metadata, and Amazon have previously launched custom AI chips to power their servers, although overall performance remains under Nvidia's dominance.
OpenAI emphasized that they are still conducting final performance measurements. However, initial testing results indicate that Jalapeno is capable of delivering a far superior performance-per-watt ratio compared to the most advanced competitors currently on the market.
The Jalapeno chip is targeted for use in OpenAI's operational servers by the end of 2026.
Conclusion
OpenAI has launched a custom AI chip called Jalapeno, developed in partnership with Broadcom, in an effort to reduce reliance on Nvidia-made GPUs. According to Broadcom, the chip's performance rivals Nvidia's Blackwell chips and Google's TPUs, with initial tests showing a superior performance-per-watt ratio compared to competitors in the market. This move places OpenAI joining other technology companies like Microsoft, Meta, and Amazon that are developing their own AI chips to support server infrastructure. The Jalapeno chip is targeted to be deployed in OpenAI's operational servers by the end of 2026.
Read more: Qualcomm Optimistic Its Data Center Business will Generate Rp269 Trillion in Revenue