boymanrobshit retweeted
It's clear now more than ever:
Ultimately, this bear market in tech stocks will be remembered as the best buying opportunity of the last decade.
Unless your view is that the AI is worthless and the AI Revolution is over, many key parts of this market are extremely cheap.
Democrats will be more friendly to AI than expected, Nvidia will survive its leadership crisis, and concerns around CapEx and free cash flow will prove to be overblown.
To top it off, current sentiment is as if the market has just gone through another 2008, all while the S&P 500 is posting +32% earnings growth.
The comments on this post will show you just how polarized sentiment has become in the direction "bubble" callers.
Fear creates an investor's biggest opportunity.
Bookmark this post.
This is funny. Kind of the market in a microcosm
On one side it’s awesome. you see the vision, we get the Jetsons eventually. I don’t know who get that reference anymore, but I like it.
But the other side of the coin is all of this money spent on the rollout, and maybe in two years we will have something that can do the dishes. 😂
Seems fairly inconsequential for a trillion dollars. And that’s why the market reaction is fair, but also why the long-term belief in the vision is fair.
JUST IN: Elon Musk predicts Tesla’s Optimus robot will be able to wash, dry, & put away dishes in 12 to 18 months.
4
554
Trying to just lay out some sort of brain dump to make sense of everything
Where I think the market is right:
- to say that credit markets have changed materially is an understatement
- offtake agreements require higher quality where they did not before
- the hyperspawning of new neos is (mostly) paused or over, depending on your LT outlook (although I don’t think the Neos know this or care about this opinion based on conversations at supercomputing this week, lol)
- Chinese cloud offtake will disappear and this is not insignificant
- there is a lot more to come on the China side if NVDA doesn’t cede control and potentially prosecution
- NVDA has pumped more froth into this market than is reasonable in n number of ways
- impact of the above is significant and will will and should cause hyperrevision of semiconductor earnings.
- training is fucking expensive and bloated and it’s going to continue to distort compute ROI as a whole. The labs treat this as existential and will not take their foot off the roadmap
- pure bare metal neos without real engineering or the ability to run token factories probably shouldn’t exist
- many Neos have spent more time negotiating finance agreements than they have spent in a data center
- overall ROI on the rollout so far is negative because of the training bloat and the cost of data center construction
- people outside the industry fucking hate AI, and that is important because of politics
- this current administration is such a disaster and if the Democrats take control, their only strategy is to go exactly the opposite of whatever Trump supported. They don’t have any other intellectual capacity other reactionary strategy. And hint, that’s not good for AI.
Where I think the market is wrong:
- raw token consumption will accelerate by multiples every year
- agents are almost the entirety of the workload and every turn is increasing in consumption. This will continue to be a force multiplier.
- enterprise continues to consume more AI to reduce the offtake % from the labs
- Inference is a GREAT business when you remove the model training. And this is really the key point of the whole thing. This is the business of AI
- to that point Neos that sell tokens are making a lot of money
- The product anthropic is putting forward is immaculate and the roadmap looks like it will eat the world.
- the hyperscale clouds will not miss revenue and AI margins will continue to expand imo
- hypers have strong KYC, and they will have limited impact from the Chinese situation
- the large clouds and SpaceX will not have unsold capacity
- inference clouds will try to proliferate on 15% debt and inference margins do support it
- the good inference clouds will switch from asset light to asset heavy. This one is interesting because those guys have been renting to date.
- data centers are a 20- year investment, so there has to be pain in the first few years with this kind of scale
- frontier AI will not be commoditized
My thoughts:
The market has spoken for now and that actually means a lot more than any fundamental tech argument. I think the reaction is fair tbh because the froth was out of control. However the long-term play of the technology remains unfazed. This shit works and it works REALLY REALLY fucking well
My conclusion from that is what was always inevitable remains inevitable: the largest scale players will win. Most compute wins but it’s a long game and the markets do not like a long game, so more pain.
boymanrobshit retweeted
CEO of Amazon, Andy Jassy, on why their AI spending is justified:
"I look back at the beginning of AWS and we felt like business was growing really fast at the beginning. The first 3 years we had about a $57 million ARR. In the first 3 years of this generative AI wave, just in AI revenue it's over $25 billion so about 440x faster than the way we grew in AWS and we thought we were growing fast."
"Big chunks of capacity are largely reserved already in 27, much of it is reserved already in 28, and a good chunk in 29. We have very clear, very substantial demand signals that are unlike anything we have seen before."
"The economics of the business are such that once you get a few years out and you have all the data centers installed, and you are monetizing with the servers and networking gear you have in place, you really like the free cash flow and return on invested capital."
Via @semafor
$AMZN
This really is a stalemate
Current administration promises never-ending inflation and catastrophic war as the only exit to the current scenario
The other political party wants to shut down the primary source of GDP in the past few years
I love hard problems. I’ve built my career pretty much on that premise, but I do not like unsolvable problems.
Apparently, cash it’s not just a position for pussies
Professor Damodaran at $76-$77 per share for $ANT. I’m interested there. I think we’ll get it 😬
As Anthropic readies to trade, I revisit my Anthropic valuation, updating for both 3rd quarter numbers as well as changes in the AI landscape (midterm elections), and the share count in the prospectus to estimate a per share value of $76-77. bit.ly/4Cvi3or
2
491
boymanrobshit retweeted
The 90th percentile OpenAI researcher now uses around $23,000 worth of tokens per day, up 105% from September.
We're sharing new data on how AI is changing research at OpenAI.
We find researchers' AI use continues to rise exponentially, and agents are completing longer tasks, but human involvement remains central in research. Researchers are also spending more time on safety and alignment, with human effort in these areas growing faster than in any other individual research area.
This report aims to increase transparency around research acceleration and inform the governance of increasingly capable AI. We believe that AI labs should be required to track and publicly report indicators of recursive self-improvement (RSI): the process by which AI accelerates the development of future AI systems.
Our hope is that this expanded report will help sustain and advance shared measurement norms across the industry.
openai.com/index/research-ac…
They offered take or pay contracts to pretty much anyone with inference serving capacity. there’s a whole lot more of these. Slightly different than straight rentals though
China's Moonshot AI, Kimi scrapped a $150M deal to rent Nvidia Blackwell Ultra B300 GPUs via an Indonesian data center, with access expected as early as next year
It even agreed to pay 50% upfront vs the usual 15%, but pulled out after new U.S. restrictions on remote chip access
6
711
It’s actually amazing how good of a business inference is quick example:
Party A buys GPUs
Party A leases GPUs to party B
Party B sells open source tokens
Party B also engages party C in a rev share on some of its capacity
Party C marks up infrastructure further and returns 80% of revenue back to party B
Party A, B and C make money on the same capacity. Double or triple margin stack. Never mind, the GPU was sold at a 70% margin to begin with.
Noncircular end customer consuming tokens is paying the bill.
Kind of both the mockery of Neoclouds and also why they continue to proliferate
boymanrobshit retweeted
$NVDA is now cheaper on multiples than buying a car wash.
Read the two below quotes on $NVDA
“Nvidia is growing 100% year over year. And it's trading at the same multiple as an automobile company growing 2% a year or maybe negative or flat.”
“The fact that the forward PE has come down to around 10, it's trading at a lower multiple than Procter and Gamble and Walmart and Ford and Coca Cola. Like, this is crazy, crazy cheap right now.”
In a sea of uncertainty, there is one thing to hang your hat on: demand for inference is insatiable and accelerating. Every single available Megawatt power is getting squeezed
13
939
Yep, all the backstopz were guaranteed only if deployments are by a certain date, so NVDA can recognize revenue. It was actually quite well put together
Nvidia interim CEO Colette Kress says its guarantee "doesn't exist anymore" if data center buildouts are delayed:
“If... they are delayed or they have a challenge in terms of building that, our guarantee doesn't exist anymore. Is it new? No. Did we change anything? No.”
10
900
Supercomputing 2026 Chicago
Some good conversations with sourcing managers for very large neo clouds today and they don’t feel like slowing down, no matter what X says.
Feels like credit markets are still there for scale players. Anecdotal, based on a my conversations as always.
Very interesting aside is that one such cloud who would not talk to me when Jensen was around is willing to talk now.
6
633
My AI/crypto analogy is this: I started doing hackathon POC stuff as a young product manager with blockchain in like 2015. I was not really viewing it at all from a investment lens because I decided at the time the only thing blockchain/crypto was good for was money laundering and illegal activity.
What I failed to acknowledge is how large of a market illegal activity is.
Certainly not a fully consistent analogy as AI has extreme utility, but clearly its value in illegal cyber attacks is immense right now.
Just interesting, no takeaways
393
Next time people will tell you that politics don’t impact the stock market, remember what’s going on right now. Everything is the fault of this current administration. All they had to do was do nothing.
But this is where the demand comes from ultimately
BREAKING: The top 1% of US businesses by AI spending allocated a record $10,199 per employee to AI in October, according to Ramp data.
Since January 2026, this figure has more than tripled.
The top 10% of firms spent a record $801 per employee last month, up +180% since January.
The median firm's average AI spend per employee rose to a record $16, up +164% since January.
In other words, the top 1% is now spending 625 times more per employee as the typical company.
Meanwhile, AI adoption among US businesses on Ramp rose +1.0 percentage point in October, to a record 56.8%, up from 45.5% in January.
Corporate AI spending is accelerating across the board.
3
365






















