technologyScore 65/100Research
OpenAI's first custom inference chip 'Jalapeno' is about to release, part of plan to slash compute cost
“you're about to release Jalapeno, your first custom chip for inference.”
Why it matters
Confirms OpenAI's in-house silicon is nearing deployment, aimed at driving down inference cost — a structural shift in how frontier lab inference demand is served, potentially away from merchant GPU vendors.
Investment implication
If Jalapeno displaces NVIDIA inference volume at OpenAI scale, it validates the custom-silicon trend (bullish for Broadcom, Marvell, and foundry TSMC as design partners; margin risk for NVIDIA's inference franchise). Watch Broadcom as likely OpenAI custom-silicon partner given prior reporting.