Last updated: 2026-06-26T17:47:54.591Z

OpenAI & Broadcom Launch Jalapeño — A Custom AI Chip Built for LLM Inference

OpenAI and Broadcom have unveiled Jalapeño, OpenAI's first custom AI chip built specifically for LLM inference. Developed in just nine months, it delivers significantly better performance per watt than current state-of-the-art accelerators. The chip is set for gigawatt-scale deployment with Microsoft and other partners starting in 2026, marking OpenAI's move to control its full stack — from models to silicon.

What Happened

OpenAI and Broadcom unveiled Jalapeño — OpenAI's first custom AI chip (called an "Intelligence Processor"), purpose-built for LLM inference. It was developed from design to production tape-out in just nine months, which is believed to be the fastest ASIC development cycle ever achieved in high-performance semiconductors. Engineering samples are already running ML workloads including GPT-5.3-Codex-Spark at production target frequency.

Why It Matters

Unlike general-purpose AI accelerators adapted from older workloads, Jalapeño was designed from a blank slate specifically for modern LLM inference — optimized around kernels, memory movement, networking, and serving patterns that frontier AI models actually need. Early testing shows it will deliver performance per watt substantially better than the current state-of-the-art. This gives OpenAI end-to-end control of its full stack: from chips to models to products.

Industry Impact

The chip is planned for deployment at gigawatt scale with data center partners including Microsoft, beginning in 2026. OpenAI also used its own AI models to accelerate parts of the chip design process — a signal that AI is now helping design the hardware that runs AI. This sets a precedent for other labs to vertically integrate and reduce dependence on Nvidia.

Key Takeaway

Jalapeño is the first chip in a multi-generation compute platform — meaning this isn't a one-off. The long-term goal is to make AI inference faster, cheaper, and more reliable for everyone — from individual developers to large enterprises — by owning the entire infrastructure stack.