Home › Tool Registry › Jalapeño
Jalapeño
by OpenAI
Visit Website ↗

Jalapeño is the reported internal codename for OpenAI's custom AI accelerator chip program, being developed with Broadcom and manufactured by TSMC to reduce OpenAI's dependence on Nvidia GPUs. It is an internal silicon initiative, not a publicly announced or purchasable product, and OpenAI has not officially confirmed specifications, timelines, or even the name.

2
Total mentions
2
Articles
See website
Pricing
No
Free tier
← Back to registry
Articles mentioning Jalapeño (2)
01OpenAI_Jalapeno_Chip_Beats_Nvidia_Blackwell_Rubin_In_InferenceAug 25, 2026 02Custom_Jalapeno_Chip_by_OpenAI_Broadcom_Transforms_LLM_InferenceJun 24, 2026
Key Features
Custom AI accelerator silicon reportedly co-designed with Broadcom
Fabrication reportedly planned with TSMC on advanced nodes
Designed primarily for OpenAI's own training/inference workloads at datacenter scale
No public specs, SDK, developer documentation, or release date have been confirmed
Pros & Cons
Pros
Could lower OpenAI's per-token inference cost and reduce reliance on Nvidia's GPU supply and pricing
Tight vertical co-design with OpenAI's own models and serving stack could yield efficiency gains generic GPUs can't match
Cons
Unconfirmed and unlaunched: no official OpenAI announcement, benchmarks, or availability
Custom silicon requires enormous capex and carries major execution/schedule risk, with no external software ecosystem