OpenAI custom chip targets the flabby middle
The thesis argued is that OpenAI's custom inference chip, co-developed with Broadcom, is a strategic move to lower inference costs and defend its mid-tier market from open-source disruption.
The argument
The guest argued that cutting inference costs in half allows closed-source providers to offer competitively priced mid-tier models, preventing open-source from hollowing out their market. However, another speaker questioned the efficiency of vertically integrating down to the chip level when hyperscalers are already absorbing the CapEx risk.
The thesis, stress-tested
✓ What validates it
- ✓OpenAI successfully deploys its custom chip in early 2025
- ✓Mid-tier closed-source model pricing drops significantly to match open-source costs
▸ Risks discussed
- ▸High capital risk of custom chip development
- ▸Distraction from core enterprise software execution
- ▸Rapid obsolescence of custom silicon compared to Nvidia's roadmap
Hear it yourself
"they have their shit together. They have an AI story, and the stock has reflected that in the last year. It's one of only two of the MagSafe, and that's up on the last eighteen months. But at the same time, when you look at things like having a viable coding model, having really kind of that next level up from just going from just shipping a model to shipping, you know, interesting products, the truth is Google hasn't done an amazing job and OpenAI and Entropic have. So if you're in the into product shipping, which I think no might well be, then I can see why going to one of those two makes sense."
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE