TLDR
- OpenAI and Broadcom have unveiled a custom AI chip called Jalapeño, built for AI inference
- The chip promises better performance-per-watt than current industry leaders
- Development took just nine months, claimed to be the fastest ASIC build on record
- Wedbush Securities says this is likely the first of many custom chips from OpenAI
- Broadcom shares dipped 1.9% in premarket trading following the announcement
OpenAI has revealed its first custom AI chip, called Jalapeño, built in partnership with Broadcom. The chip is designed to handle AI inference — the process of applying machine learning to new tasks.
We’ve designed and built our first AI chip: Jalapeño.
Designed from the ground up by OpenAI and brought to production with @Broadcom, Jalapeño is purpose-built for the LLM workloads powering ChatGPT, Codex, the API, and future agentic products.
Chips are foundational to the AI… pic.twitter.com/mHU7DaMMTi
— OpenAI (@OpenAI) June 24, 2026
OpenAI says the chip delivers a performance-per-watt ratio that is “substantially better” than the current industry best. It was built specifically for large language models like GPT, rather than adapted from older AI hardware.
The chip reached the manufacturing tape-out stage, meaning the design has been sent out for initial production. OpenAI and Broadcom say the nine-month development timeline is the fastest ever recorded for a custom application-specific integrated circuit.
What Jalapeño Is Designed To Do
The chip is aimed at making AI services faster, cheaper, and more reliable for everyday users. OpenAI says it could mean quicker responses from ChatGPT, smoother access during busy periods, and more stable pricing.
Power consumption is another target. AI data centers currently use enormous amounts of electricity, drawing attention from governments in several countries. OpenAI says Jalapeño could reduce that energy load.
That said, if OpenAI scales aggressively or launches more demanding models, total power use could still climb even with a more efficient chip.
Celestica helped with board and rack system integration, alongside high-performance networking and production systems. The chip is part of a broader plan to develop 10 gigawatts of custom AI accelerators.
Broadcom plans to deploy racks of these accelerators starting in the second half of 2026, with full rollout targeted by the end of 2029.
What This Means for Nvidia and Broadcom
OpenAI has relied heavily on Nvidia GPUs to power its AI processing. Jalapeño may not replace Nvidia hardware entirely, but it could reduce how much OpenAI depends on it going forward.
Wedbush Securities analyst Matt Bryson noted that successful chip development for compute is usually a multi-generation process. He said broader adoption may require a second, third, or even fourth version of the design.
Bryson described the development as a “probable positive” for Broadcom, while cautioning that early shipments could be modest in scale.
Broadcom shares fell 1.9% in premarket trading on Friday following the news.
OpenAI said its own AI models helped speed up the chip’s development, compressing what is typically a longer design cycle into nine months.
The company and Broadcom described this as the start of a “multi-generation roadmap,” with performance and efficiency expected to improve over time.
Microsoft is among the data center partners expected to deploy Jalapeño at gigawatt scale later in 2026.
🚨 Our JUNE Stock Picks Are Live!
A new month means new opportunities. Our analysts have just released their top stock picks for June, highlighting companies with strong momentum that rank highly on our KO Score algorithm. We’re also now sharing trade ideas for both long-term and short-term investors, giving you more ways to spot potential opportunities in the market.
Sign up to Knockout Stocks today and get 50% off to unlock the full list and see which stocks made the cut.
Use coupon code Special50 for your exclusive discount!







