Pinned Tweet
I'm always amused when people are shocked by people doing exactly what they said they’d do. Everything that has happened over the past week was publicly announced months ago.
Stephen Miran, widely viewed as the architect of Trump's trade policy, published a paper "A User’s Guide to Restructuring the Global Trading System" in November of 2024, laying out everything that has happened (as well as many things that are yet to come).
tldr:
1) they’re not insolvent yet
2) numbers improving
3) compute procurement most critical risk
Skimmed Anthropic’s S-1 and it looks surprisingly strong.
Revenue went from $41M in 2023 to $16.3B in the first half of this year. Operating loss is basically gone and adjusted operating income is already positive. They’re getting way more revenue out of the same compute, NRR is ~900%, and they’ve got $132B in cash.
The absolute size of the spend is still the elephant in the room. Non-cancelable compute commitments run deep into the next decade and the raw numbers are massive even if the GAAP losses are mostly mark-to-market noise.
There’s also a circular piece to the revenue. A lot of it still runs through Amazon and Google’s marketplaces same companies that are investors, suppliers, and competitors. It’s unclear how much of the revenue is just ai companies buying ai.
Apparently Dario gets to handpick five people who then handpick the majority of Anthropic’s board, forever.
While this shadow board may seem sketchy, it’s important to remember that, thanks to super voting shares, no one besides the cofounders has any real say in governance anyway.
Skimmed Anthropic’s S-1 and it looks surprisingly strong.
Revenue went from $41M in 2023 to $16.3B in the first half of this year. Operating loss is basically gone and adjusted operating income is already positive. They’re getting way more revenue out of the same compute, NRR is ~900%, and they’ve got $132B in cash.
The absolute size of the spend is still the elephant in the room. Non-cancelable compute commitments run deep into the next decade and the raw numbers are massive even if the GAAP losses are mostly mark-to-market noise.
There’s also a circular piece to the revenue. A lot of it still runs through Amazon and Google’s marketplaces same companies that are investors, suppliers, and competitors. It’s unclear how much of the revenue is just ai companies buying ai.
Sam retweeted
Replying to @protosphinx
The wildest timeline is it OpenAI fails to IPO and gets absorbed into spaceX, and Elon makes it a non profit again
worth checking the comment section here before it gets cleaned up
great place to find out who has absolutely no idea what they’re talking about
1. SpaceX is now a neocloud
I think this was obvious after the Anthropic deal but this tweet makes it official. I’d expect them to deemphasize model development going forward.
Replying to @chamath
Pretty much.
As of the past few weeks, debt markets have almost entirely shut off access to money for any company that doesn’t have substantial revenue independent of AI data centers.
SpaceX closed $40B of debt for GPUs, but we’re investment grade and have Starlink & launch revenue.
That means the 200+ neoclouds are going to see guppy-level death and consolidation. Some will make it through, but not many.
3. SpaceX will likely try to buy other neoclouds or energy companies.
FUD = discounts! Energy companies attempting to pivot into hosting seem like specifically good targets
Lula's victory... may save CoreWeave?
EMEA have had a rough October. Paris is in flames, French sovereign yields are approaching 5%, Indonesia has lost $65B in market value amid threats of an MSCI downgrade w/ a rapidly escalating spat with Singapore. Taiwan, Iran, Yemen, and Ukraine all remain sources of geopolitical uncertainty.
Meanwhile, the American AI buildout is running into a credit crunch. CoreWeave's bonds are yielding around 15% as the markets assess the viability of their business model in a volatile rates environment.
The dollar remains the world's default refuge, less because America is particularly stable than because everything else looks much worse. During the 2008 financial crisis, foreign investors bought $335B in Treasuries over three months. During the 2011 European debt crisis, they bought another $147B.
The more the global markets deteriorate, the more money flows into American bonds, the lower rates fall. A recent Federal Reserve study found a 9 bp impact for every 300b of inflows. This would offer CoreWeave and the other eeoclouds immediate relief.
But the most troubling implication is Washington's incentives: it's pretty easy to argue that it's easier for Trump to destabilize foreign markets than force the federal reserve to cut rates.
Sam retweeted
One of my favorite bits while I was in New York this week was asking transplants if they’ve ever been to House of Prime Rib in the West Village and waiting for them to say yes
Almost every investing blowup can be traced back to confusing luck with skill.
Consistent returns require the humility to recognize when something worked for the wrong reasons and the discipline not to repeat it.
LPs will be choosing between:
-A16Z gaming fund VII (10b, fund VI underperformed S&P)
-“Mykonos Capital” (300m, ex Goldman Sachs Harvard grad whose relevant experience is 6 years as a principal at NEA)
-“Paperclip” (100m, ex Anthropic engineer anchoring with 20m of his own capital made through backing Cursor’s A)
I don’t think this is a hard decision
Replying to @MWCvitkovic
The best superangels will simply found the next generation of top venture funds, anchored with their own capital.
This is how venture has always worked












