Gavin Baker: "which AI company actually survives? Google - already multi-trillion Microsoft, OpenAI, Meta chose open source - the only ones left is xAI and Anthropic - xAI + X will be worth over a trillion" this is him explaining what he actually invests in - and why Taiwan... sells sand to Nvidia for $700 that Nvidia sells for $50,000 His words: what to invest in: "we're investing in things that increase the GPU utilization rate - because if you take a GPU from 30% utilized to 60%, you've doubled the output of that AI factory" what not to Invest in: "the models will basically go to zero - the data center is a commodity - energy might be short for a while but is also a commodity - the only thing in the whole stack with real value is the data moat and reinforcement learning" "the GPU has gotten 50 times faster in the last four years - that's a big part of why we're living through this AI revolution" he doesn't rent intelligence - he owns the thing it runs on bookmark & watch - see how below with local agent ↓show more

$700 sand to $50,000 GPU is the whole AI economy in one line smart take brotha

that's why Nvidia makes such big money and Beker was the early investor in it

@grok based on this what us equities would be possible buys for Gavin with sub $50 billion market cap

mate he doesn't rent intelligence.he owns what it runs on - that's the whole thesis

yeee, and you can do the same with atomic agent

utilization is the trade nobody talks about

@grok which public companies increase GPU utilization rates and give me their forward p/e’s. Which public companies have large data moats and give me their forward p/e’s.

Owning the infrastructure can be more valuable than using it.

first time see such takes about Elon Musks models, worth to analyze savedm bodila thanks!

Happy to give reasons for research and a potential purchase in future hehe

Arm mrvl aaoi crdo alab cbrs

Oh no way, Elon’s buddies think his company is going to do well? I’m shocked.

XAI's Colossus is a 200k GPU cluster. that's 16 exaflops of compute burning cash faster than OpenAI's $5B burn. elon is betting the farm that x (which loses $1B/year) will pay for it. high stakes, high variance

I always learn a lot whenever @GavinSBaker speaks

The way he his hand waving , it seems this guy is dumb and shallow

Bir GPU'nun kullanım oranını %30'dan %60'a çıkaran şey daha fazla GPU almak değil, o GPU'yu daha hızlı besleyen optik sinyal (COHR/LITE), daha serin tutan soğutma altyapısı (VRT/MOD), daha verimli tasarlayan EDA (SNPS/CDNS) ve daha kesintisiz enerji sağlayan SMR'dir (BWXT,OKLO).

@grok what year is this interview from?

@grok which companies fit his criteria?

沙子到显卡这段，差的就是话语权吧

When was this @grok

Buy who has the compute - pretty simple.

Why aren’t transport stocks and airlines being re-rated with higher multiples for the data moats they kick off? Are they just spilling the data to edge storage?

Baker’s bet is that the survivors are the ones who control utilization, reinforcement and data gravity.

Flawed arguments: 1) Distillation can replicate a working model from a frontier one trained with “moated data”, and offer user a much cheaper product, 2) did he check SPCX’s valuation, the supposed “1trn” valuation of x+xAI has already been priced in and more?

is there a u tube to this full talk bruther? love gavin

he's falling into the trap of horsepower car companies chased in the 80s.... next slide

Cool

Isn't this like preferring Intel over Microsoft during Web1? I disagree, I think hardware is flattening and becoming horizontal. This is not to say it won't be profitable for 20yrs. But those looking for a better risk/reward ration, I ain't looking at OpenAI or Anthropic, I am looking at $PLTR, $SNOW and Databricks $RVI (Eleven Labs). These companies represent what I am calling the Robot Data Layer. OpenAI and Anthropic are not doing that as they are answering different questions.

When was this video taken? Antonio was talking about version 12 for FSD which is years ago

- Investor Gavin Baker argues AI models will commoditize to near-zero value, with real margins shifting to infrastructure that raises GPU utilization from low levels like 30% to 60%, alongside data moats and reinforcement learning. - The post pairs this with a demo of open-weight Kimi K3 generating detailed 3D arcade games (Pac-Man, Snake, pinball) rivaling paid Claude Fable 5 outputs, running locally for free on GPUs versus $3.60–$4.50 cloud costs. - Baker identifies Google, xAI, and Anthropic as primary survivors in an open-source-heavy field, predicting xAI combined with X could exceed trillion-dollar valuation by owning the underlying compute stack.

@grok when was this discussion? Fsd 14.3 is out right now.

"data moat" = @NewConstructs

Tận dụng GPU mới là giá trị thật.

Baker’s fund Atreides Management has almost 10% allocated to ALAB. This ties in perfectly to his thesis on connectivity and efficiency demands.

@grok what is he suggesting to invest in