OpenAI and Broadcom Unveil Custom AI Chip Amid Industry Shift to Infrastructure Focus

Key Intelligence

• ■OpenAI, in collaboration with Broadcom, has unveiled 'Jalapeño,' a custom AI chip specifically designed for efficient Large Language Model (LLM) inference at scale.
• ■This move signals a strategic shift within the AI industry, with major players seeking to develop proprietary hardware and reduce reliance on general-purpose GPUs from vendors like NVIDIA.
• ■The broader AI landscape is entering a new 'arms race' focused on securing critical infrastructure, including specialized chips, massive data centers, stable power supply, and high-performance memory solutions.
• ■Companies specializing in memory chips, such as SK Hynix and Micron, are emerging as key beneficiaries of this infrastructure demand.
• ■Markets are beginning to weigh this next phase of AI trade, with investor attention broadening from LLM software to the foundational computing and physical infrastructure that underpins AI advancement.