OpenAI, Google, Apple and SpaceX Develop Custom AI Chips

OpenAI unveiled plans for Jalapeño, its first custom inference chip developed with Broadcom, according to TechCrunch reporting published June 26, 2026. The announcement adds OpenAI to a list of companies designing their own silicon for AI workloads. Google, Apple and SpaceX have already introduced or are developing custom chips for similar purposes.

The companies cite greater control over hardware specifications and performance optimization as primary reasons for the shift. TechCrunch noted that custom designs allow tuning for specific model requirements rather than relying on general-purpose processors.

Nvidia has supplied the majority of AI training and inference chips in recent years. The new efforts represent an attempt to limit dependence on any single vendor.

TechCrunch reported that the strategy mirrors earlier moves by Apple, which transitioned its Mac computers from Intel processors to its own silicon. The publication said the goal is incremental diversification rather than an immediate replacement of existing suppliers. No timeline for Jalapeño production or deployment was provided in the coverage.