Science & Technology

Everyone from OpenAI to SpaceX is building their own chips | Equity Podcast

OpenAI just shared its plans to spice things up with Jalapeño, its custom inference chip built with Broadcom, joining Google, Apple, and SpaceX in a growing list of companies building their way out of single-supplier risk. The goal is less of a clean break and more of a hedge. Custom silicon means more control, hardware…

Published

on

OpenAI just shared its plans to spice things up with Jalapeño, its custom inference chip built with Broadcom, joining Google, Apple, and SpaceX in a growing list of companies building their way out of single-supplier risk. The goal is less of a clean break and more of a hedge. Custom silicon means more control, hardware tuned to specific needs, and the kind of performance gains Apple unlocked when it ditched Intel.

On this episode of TechCrunch’s Equity podcast, hosts Kirsten Korosec, Anthony Ha, and Sean O’Kane dig into what the custom chip trend means for the industry and a few deals of the week.

Subscribe to Equity on YouTube, Apple Podcasts, Overcast, Spotify and all the casts. You also can follow Equity on X and Threads, at @EquityPod.

Chapters:

00:00 Intro

01:28 OpenAI’s Jalapeno chip and the race for custom silicon

07:33 Memory chip crunch and rising costs for consumers

09:12 Nvidia, Grok, and the neocloud pivot

13:18 What’s the point of data centers in space?

18:25 AI agent loops: hype or next big thing?

27:56 Agility Robotics goes public via SPAC

32:49 A24 takes Google DeepMind investment

35:47 Outro

5 Comments

  1. @Cloudshalla

    June 27, 2026 at 11:28 pm

    Loved this video on “Everyone from OpenAI to SpaceX is building their own chips”. If anyone is looking for cloud jobs, CloudShalla has a ton of listings right now. Keep it up!

  2. @---jc7pi

    June 28, 2026 at 8:20 am

    They still have to go to TSMC, so they save some amount of margin but really not that much. And if its just a few % worse, its going to cost them billions. Also, wants you go there, you have to keep doing it every 2 years.

  3. @TheBHAitken

    June 28, 2026 at 11:00 am

    LOL Nvidia/Jensen forgot that AI designs chips….

  4. @davidlt

    June 28, 2026 at 2:49 pm

    Since most model providers have or will have soonish a good enough intelligence we are moving from “more training hardware” to “more inference hardware”. All major hyperscalers have their own chips, even Google split their TPUs into training and inference instead of one. You are selling inference (tokens, and tokens/s) so you want to be more vertically integrated here, optimize it as much as possible to get profitability (hopefully).

  5. @anthenslawyergr

    June 28, 2026 at 4:18 pm

    History keeps repeating its self whoever owns the stack usually owns de margin🤔💯

Leave a Reply

Your email address will not be published. Required fields are marked *

Trending

Exit mobile version