Platform WatchJune 24, 2026via AI Business
OpenAI and Broadcom Introduce AI Inference Chip
Why it matters
A joint OpenAI-Broadcom inference chip directly threatens the token-pricing moat that's been core to AI company unit economics. Lower inference costs could reshape margins across the entire model-serving ecosystem.
Key signals
- OpenAI and Broadcom co-developed AI inference chip
- Chip designed to reduce token pricing pressure
- Targets cost concerns from enterprise customers
- Inference optimization focus (not training)
- Published June 24, 2026
The hook
OpenAI and Broadcom just shipped a custom inference chip. Token prices are about to crater.
The chip would give AI model makers the option to offer lower token prices, possibly easing the concerns of businesses worried about higher token costs.