OpenAI reveals its first AI processor: Jalapeño

2026-06-24 · Source: The Verge · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Cloud Computing & IT Infrastructure, AI Hardware · Depth: Novice, quick

Summary

OpenAI has unveiled Jalapeño, its first custom "intelligence processor" chip, developed in partnership with Broadcom. This Application-Specific Integrated Circuit (ASIC) is engineered specifically for AI inference, powering current and future large language models like ChatGPT. The announcement, made on Wednesday, comes nine months after OpenAI initiated its collaboration with Broadcom to reduce its dependence on Nvidia's limited-supply GPUs. Broadcom CEO Hock Tan states that Jalapeño's performance rivals Nvidia's Blackwell chips and Google's Tensor processing units. This move places OpenAI among other major AI companies, including Microsoft, Meta, and Amazon, which have also introduced custom AI chips for their server infrastructure. OpenAI anticipates deploying Jalapeño, described as the "first step in a multi-generation compute platform," by the end of 2026, with early tests indicating substantially better performance per watt.

Key takeaway

For AI Architects evaluating infrastructure for large language models, OpenAI's Jalapeño chip signals a growing trend towards custom ASICs. You should consider how dedicated inference hardware can reduce your reliance on general-purpose GPUs and potentially offer superior performance per watt. Begin exploring custom chip roadmaps from major providers to inform your long-term compute strategy and optimize operational costs.

Key insights

OpenAI's Jalapeño chip, an ASIC for AI inference, aims to reduce GPU reliance and boost LLM performance.

Principles

• Custom ASICs reduce reliance on general-purpose GPUs.
• AI inference chips optimize for specific model execution.
• Performance per watt is a key metric for AI hardware.

In practice

• Deploy custom chips for LLM inference.
• Evaluate ASICs for power efficiency gains.
• Diversify AI hardware suppliers.

Topics

• OpenAI
• Broadcom
• Jalapeño
• AI Inference
• Custom AI Chips
• Large Language Models
• ASICs

Best for: Investor, CTO, VP of Engineering/Data, Tech Journalist, AI Hardware Engineer, AI Architect

Related on AIssential

• Nvidia Plans New Chip to Speed AI Processing, Shake Up Computing Market — Nvidia is launching a new inference-optimized AI processor to accelerate model responses and reshape the AI hardware market.
• Local AI Agents on macOS: Building an Ollama Home Lab — Local AI inference on consumer hardware offers privacy, cost savings, and practical automation capabilities.
• Taalas Specializes to Extremes for Extraordinary Token Speed — Extreme specialization in AI chips by hardwiring models can yield superior performance and cost efficiency.
• Google plans nearly two million new AI chips as it turns to Marvell for custom designs — Google is developing custom AI chips with Marvell to enhance inference capabilities and diversify its supply chain.
• AI Inference on AMD Ryzen™ AI Max Processor — AMD Ryzen™ AI Max+ processors enable local 100B+ parameter LLM inference via Unified Memory Architecture, reducing cloud dependency.
• OpenAI GPT-5.3-Codex-Spark Shows What's Possible With Cerebras — OpenAI's new coding model, GPT-5.3-Codex-Spark, leverages Cerebras hardware, challenging Nvidia's dominance and expanding AI hardware options.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by The Verge.