OpenAI Enters AI Chip Race With Jalapeño Launch
OpenAI has introduced Jalapeño, its first custom-designed AI chip, marking a significant step in the company’s effort to build more of its technology stack in-house. Developed in partnership with Broadcom, the chip focuses on AI inference, the process that allows models such as ChatGPT and Codex to generate responses for users. OpenAI announced the launch on June 24 and described Jalapeño as the first generation of a broader hardware strategy.
The move comes as AI companies seek greater control over the computing infrastructure that powers their products. Until now, OpenAI has relied heavily on Nvidia hardware. However, growing demand for AI services has pushed major technology firms to explore custom silicon that can improve efficiency and reduce long-term costs. Google, Amazon, Meta, and Microsoft have all pursued similar strategies.
Designed Specifically for AI Inference
Unlike general-purpose processors, Jalapeño is an application-specific integrated circuit built exclusively for inference workloads. As a result, the chip targets the tasks that occur every time a user interacts with ChatGPT or other OpenAI services. OpenAI says early testing shows substantially better performance per watt than current leading alternatives.
The company developed the processor in just nine months with Broadcom’s support. Meanwhile, manufacturing is handled by TSMC, while Celestica contributes system-level integration. According to Broadcom Chief Executive Hock Tan, Jalapeño delivers performance comparable to Nvidia’s Blackwell chips and Google’s Tensor Processing Units.
OpenAI President Greg Brockman previously explained the reasoning behind the company’s hardware ambitions.
“We have a deep understanding of the workload. We’ve really been looking for specific workloads that are underserved, [and asking] how can we build something that will be able to accelerate what’s possible?”
A Step Toward Greater Infrastructure Control
Jalapeño forms part of OpenAI’s wider effort to reduce dependence on external suppliers and gain more control over the economics of AI. While Nvidia hardware will continue to play a major role in training advanced models, OpenAI expects its custom chip to handle a growing share of inference workloads. Consequently, the company could lower operating costs while improving the efficiency of its services. Commercially. Instead, it will deploy Jalapeño within its own infrastructure as customer demand grows. The company views the processor as the foundation of a multi-generation compute platform that will support future AI systems and agent-based applications.
OpenAI highlighted the broader vision behind the project in its announcement:
“OpenAI is not only developing frontier models or building products on top of them; it is designing the infrastructure underneath them: chip architecture, kernels, memory systems, networking, scheduling, deployment systems, and product experience. Because OpenAI operates across the stack, each layer can be optimized around the same goal: making its models faster, more reliable, and more affordable for users.”
As AI competition intensifies, custom hardware is becoming a strategic advantage. Therefore, Jalapeño represents more than a new chip. It signals OpenAI’s intention to control a larger portion of the technology that powers its next generation of AI products.
