Zortix
Sign in
NVDAAMDCore thesis · 5/5Save idea

Silicon startups face steep Nvidia barriers

AI hardware startups face an incredibly difficult path to competing with Nvidia due to long design cycles, shifting model architectures, and Nvidia's massive supply chain and software advantages.

The argument

The speakers argued that startups must deliver a 5x hardware efficiency advantage to compete, but by the time their chips launch, the underlying AI models (like the shift from dense transformers to smaller matrix shapes) have often evolved. Furthermore, Nvidia's supply chain dominance and software ecosystem compress any marginal hardware advantages.

The thesis, stress-tested
✓ What validates it
  • A major silicon startup launches a chip that demonstrates a sustained 5x performance-per-watt advantage on active production models
  • AMD's gross margins on GPUs close the gap with Nvidia's 75% gross margins
▸ Risks discussed
  • A major technological leap (e.g., neuromorphic computing) could successfully bypass the current GPU tech tree
  • Hyperscalers or large buyers may aggressively subsidize a second source to break Nvidia's monopoly pricing
Hear it yourself
"Right? I think there's a little bit of that. Right? Cost suddenly matters, and they figured out a way how they can steer that. I think yeah. That I mean, and they talked about how they've been able to dramatically increase their infrastructure capacity. Because I myself was just regularly using o three or 4.5. Right? And now I'm forced to use auto, which sometimes gives me the o three equivalent thinking model, but sometimes gives me just the regular base model Absolutely. Yeah. Which sucks. But, like, I think for the free user, it's actually quite interesting."
04:30 · Verify in source ↗
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE
NOT INVESTMENT ADVICE · A SUMMARY OF WHAT WAS SAID ON THE PODCAST · VERIFY AGAINST THE SOURCE
NVDA: Silicon startups face steep Nvidia barriers · Zortix