OpenAI announced Jalapeño, an inference-only chip developed with Broadcom and the first custom silicon OpenAI has designed itself. The design targets inference rather than training, and OpenAI said its own models were used in the design work, taking the chip from conception to tape-out in about nine months. Initial deployment was planned for later in 2026, with volume production in 2027. With inference demand becoming the dominant cost, the chip marked a move to serve part of a compute base long dependent on NVIDIA GPUs with silicon of OpenAI's own.
24June 2026
ProductConfirmed
OpenAI and Broadcom unveil Jalapeño, OpenAI's first custom chip
Participants
Also mentioned
Not parties to this event, but named in the text above.
Tags
- hardware
- chips
- infrastructure
Sources
Later developments
25 August 2026 · Development
OpenAI published Jalapeño's first results, reporting 1.5–1.9x more work per watt at peak throughput and 2.1–4.1x faster ultra-low-latency inference on three open models (gpt-oss 120B, DeepSeek R1, Kimi K2.5), presenting details at Hot Chips 2026. The figures are OpenAI's own measurements.