← Founder Notes
Archive

Openai's first custom chip, jalapeno with broadcom, just started shipping and claims ~50% lower…

Yethikrishna ROriginal on Threads

openai's first custom chip, jalapeno with broadcom, just started shipping and claims ~50% lower cost per inference token than current nvidia gpus, with 1.3 gigawatts planned for 2027. the biggest model company just became a chip company.

nvidia's moat was never the die; it was the stack.

Context

OpenAI and Broadcom, in matching releases of 24 June 2026, unveiled Jalapeno, OpenAI's first Intelligence Processor, an inference accelerator developed from design to production in nine months. The release says OpenAI is still measuring final performance, early testing shows performance per watt substantially better than current state-of-the-art, a detailed technical report is due in the coming months, and it will be deployed at gigawatt scale with data center partners over multiple generations. CNBC reports the companies had announced in October racks of OpenAI-designed chips starting late this year, ultimately 10 gigawatts.

How it compares

The codename and Broadcom partnership are company-confirmed. In the text read there is no 50% lower cost per token figure (the release says performance per watt substantially better, which is qualitative), no 1.3 gigawatts in 2027, and no statement that it has just started shipping. Those three specifics are unverified, and no Nvidia comparison was found in company text. Vendor-reported.

Related work

Watch next

  • The promised technical report.

Sources

  1. OpenAI and Broadcom Jalapeno inference chip (OpenAI, 24 Jun 2026)openai.com
  2. Broadcom release, 24 Jun 2026broadcom.com

Provenance

The note above is reproduced unedited from the original post, first published on Threads on 18 September 2026 at 11:50 IST. Sources are the papers and datasets the note draws on.

View the original post
Embed this note
<iframe src="https://founder.myndlabs.tech/notes/embed/openai-s-first-custom-chip-jalapeno-with-broadcom-Ddax03QiPq4" width="480" height="420" style="border:0;max-width:100%" loading="lazy" title="Openai's first custom chip, jalapeno with broadcom, just started shipping and claims ~50% lower…"></iframe>

More notes