OpenAI's VP of Hardware on Jalapeño: first custom AI chip taped out in 9 months

thehiphopswami · x · 2026-09-18

In The Data Exchange podcast, OpenAI VP of Hardware Richard Ho details Jalapeño, the company's first custom AI accelerator: a blank-slate architecture bringing memory and compute closer, breaking the throughput-latency trade-off, and targeting lower inference costs. AI tools compressed development so the chip taped out in just nine months. He also covers CUDA's moat, speculative decoding, the HBM4 roadmap, and argues power — not chips — is becoming the data center bottleneck.

Related event: OpenAI's first in-house AI chip Jalapeño taped out in 9 months(5 posts)→

Original post →

More from Infra

Infra channel →