OpenAI unveils its first AI chip Jalapeño, could make ChatGPT faster and cheaper

OpenAI has unveiled Jalapeno, its first custom AI chip built for inference and already running in its labs. The move could lower ChatGPT costs and sharpen the contest with Nvidia in AI hardware.

In Short

• OpenAI unveils Jalapeño, its first custom AI inference chip
• Chip promises better efficiency and lower AI operating costs
• Move marks OpenAI's entry into the AI hardware race

For years, Nvidia has been the undisputed king of the AI world, supplying the powerful chips that run everything from ChatGPT to Google's AI systems. Now, OpenAI is making its first serious move to challenge that dominance. The company has unveiled its first custom AI chip, called Jalapeo, a processor built specifically to handle one of the most important jobs in artificial intelligence: answering user queries. Think of it this way. Training an AI model is like sending a student to school for years. But once the student graduates, they spend their life answering questions and completing tasks. That second part is known as inference, and that's exactly what Jalapeo has been designed for.

Built specifically for OpenAI's needs

Unlike Nvidia's general-purpose AI chips, Jalapeo was created specifically for OpenAI's needs. The chip was designed in partnership with Broadcom and is already running OpenAI's GPT-5.3-Codex-Spark model inside company labs.

According to OpenAI, early tests suggest the chip delivers significantly better performance per watt than today's leading AI hardware, potentially reducing the enormous costs of running services such as ChatGPT.

AI helped build the chip

The project also highlights how AI is increasingly helping to build itself. OpenAI says engineers completed the chip design in just nine months before sending it to Taiwan's TSMC for manufacturing, with AI tools helping speed up parts of the design process.

Broadcom CEO said the new chip performs on par with Nvidia's latest Blackwell processors and Google's Tensor Processing Units (TPUs). If that proves true at scale, it could mark a significant shift in the AI hardware race.

Following the path of tech giants

The move follows a path already taken by tech giants like Google and Amazon, both of which developed their own AI accelerators to reduce reliance on outside suppliers. OpenAI has reportedly been exploring custom chips since 2023, but Jalapeo is its first public result. Its main rival Anthropic is also planning to build its custom AI chip, sources told Reuters in April.

The chip is only the beginning. OpenAI describes Jalapeo as the first step in a multi-generation hardware strategy that will expand over the coming years. The company plans to start deploying the chip by the end of this year, while Canadian manufacturer Celestica will build the server systems that house it.

Why does this matter to everyday users?

OpenAI says every improvement in speed, cost, and reliability can directly translate into faster ChatGPT responses, more capable coding assistants, cheaper AI services, and fewer outages during periods of high demand.

In other words, the future of AI may not be decided solely by smarter models. It could also depend on who builds the best chips underneath them.

- Ends