@GavinSBaker / appears on all-in, all-in, bg2, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, all-in, bg2, all-in, tweets, all-in, tweets, all-in, tweets, all-in, bg2, tweets, all-in, all-in, all-in, all-in, tweets, bg2, tweets, all-in, all-in, tweets
The mega bull case for AI infrastructure would be *if* market share shifted away from certain frontier labs with 90%+ inference margins toward cheaper models, whether open-source or closed.
With SpaceX and Meta being vertically integrated and possessing the #3 and #4 models respectively it is more possible than ever.
Pareto dominant for coding by the numbers.
{
"type": "trace.speaker",
"handle": "GavinSBaker",
"name": "Gavin Baker",
[
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
,
],
{
}
}AISO Market Calls classifies public statements from public sources. A classification such as long thesis, short thesis, or explicit call describes the content of the public statement and does not prove that the speaker actually entered, exited, or held a position. Market performance shown is calculated from public market data around the source publication or timestamp. It is not investment advice.
Google never not in a good position, sitting on the most compute and the most valuable data even after losing its chip cost edge, with cloud demand going crazy.
Pareto dominant for coding by the numbers.
everything that cuts against training works for inference. ... I do think distributed inference, distributed inference clouds are coming.
CXMT is going public in China. They're going to— they may be the cure for Apple's ills. They will flood the market with, to some degree, cheap consumer-grade DRAM.
I think Anthropic is worth $3 trillion today.
DRAM is the most important bottleneck. There's a whole segment of people on X who are very focused on bottleneck— bottlenecks. I call them the bottleneck bros. ... The bottleneck that matters is DRAM, and DRAM and HBM DRAM. This is the most important bottleneck simply because memory capacity and bandwidth are foundational to the performance of every AI model.
NVIDIA is the American open source champion. ... They can release GLM 5.2 or better whenever they want.
And I think they're on this, like, you know, they have this crazy belief that, oh, you know, our own internal chips are good enough. They're not.
Yeah, I do think the bottleneck grows, as they are called, are this like bottleneck trade is kind of nearing its end.
And if you have all these points of presence and you can deliver really low latency tokens to high value users, I think what's interesting for all these CDNs is they're getting a massive premium for that latency. It turns out one of the lessons from Cerebris is people are willing to pay for speed.
I am increasingly looking at EV to net PP&E as a valuation metric, just with the idea of being installed atoms on Earth. I think are going to appreciate in value.
So they're doing $50 billion a gigawatt on the Google deal. So how quickly they can add gigawatts really matters. Yeah, right. If you can energize 2, 3, 4 gigawatts in the next year, that's a lot of revenue.
People are hunting for the final ingredient. Maybe it's a— sand that turns into silicon.
Every CDN, yeah, has, has this business. You know, Akamai, they signed— yeah, what was it, a $1.8 billion with Anthropic? And if you have all these points of presence.
It does seem like they monetize gigawatts at a higher rate than anyone else.. And we know from Jensen that they bring on data centers faster than anyone else, per Jensen's words.
we do have an existence proof that once you really get on that Pareto frontier, right, revenue can scale rapidly and it's called Anthropic. And there does seem to be an exhaust— there seems to be a lot of demand for coding
You've had a lot of stocks that forget climbing a mountain or hill... They've gone straight up a cliff... They're tired. They need to rest. And we'll see, do they just rest at the top of that cliff they climbed... Or do they need to go downhill for a bit?
I think going to a city where you can't get in a Waymo or a cyber cab is going to feel barbaric and unsafe... whatever individual municipalities decide... I just don't think it's going to persist.
Those older GPUs, they have a useful life for ten or fifteen years
they were EBIT positive per The Wall Street Journal... returns are there
NVIDIA's AI business is growing faster than broadcoms. Faster than a lot of other companies
every day the Strait of Hormones is closed, think it's relatively good for the reindustrialization of America
You have memory makers that, you know, three to five times PE. You have Nvidia at a really low PE... If the multiples on Nvidia and memory are correct, everything else is probably going to underperform. The AI market is cross-sectionally inefficient right now.
there's a narrative that Nvidia is losing share to the TPU. And Broadcom guided for 143% year-over-year growth in their AI semiconductor revenue
electricity is a base input to every manufacturing or industrial process... And what we make electricity with in America overwhelmingly is natural gas... NG1. It is down this year.
I think the watts shortage will probably begin to alleviate '27, '28. And then I think orbital compute will really solve that.
I don't know anyone like me who's not really bullish on DRAM.
AI is just shifting from all-you-can-eat to pay-by-the-drink... people really like to use AI, and particularly now that one person can have 100 agents working.
No reach data yet.
Drop another platform link to enrich this profile. Drop a link →
