New Substack post just dropped!
I share my thoughts on Anthropic, now that we are on the eve of its historic IPO.
TLDR: the IPO will probably go fine.
https://t.co/ZI1ijV41el
It’s great that AI is scrambling political coalitions. Partisanship is a mind killer. Once you succumb to it, you’ve just flushed any hope of having an intellectual life down the toilet.
New Substack post just dropped!
I share my thoughts on Anthropic, now that we are on the eve of its historic IPO.
TLDR: the IPO will probably go fine.
https://t.co/ZI1ijV41el
.@grok about how many shareholder of Nvidia exist overall?
Estimate if you must.
Then, based on this number, tell me what % of Nvidia shareholders is 104,000.
Then, tell me if movement/motions/policies/bills/etc with a 24% approval rating usually succeed.
BREAKING: The petition to reinstate Jensen Huang as CEO of Nvidia, $NVDA, has reached over 200,000 signatures.
The petition has been signed by more than 104,000 shareholders and 11,800 current and former Nvidia employees.
~24% of all employees currently working at Nvidia have signed the petition.
The board has yet to comment on the petition.
@SenZhang9 True. X doesn't have it's own bias at all. Totally isn't flooded with chuds and Tech Right beta males larping as Great Men of History, and manipulating the fuck out of every possible narrative.
It's basically like pure truth straight into my veins.
The whole "Jensen must return to Nvidia" thing feels like one of those chud movements that exists on X but literally nowhere else, like "Pratt for LA Mayor".
"Affair or not, nobody brings more shareholder value than Jensen."
Having three affairs while being the public face of the biggest/most important company in America, two with his own employees, repeatedly lying about it to senior executives, and also playing fast-and-lose with our nation's national security interests w/r/t his super dubious relationship to Megaspeed is good for shareholder how, exactly?
Wouldn't *not* doing these things be better for shareholder value, or is that just me?
Good thing we don't actually have to wonder. The results are in.
"Affair or not, nobody brings more shareholder value than Jensen."
Having three affairs while being the public face of the biggest/most important company in America, two with his own employees, repeatedly lying about it to senior executives, and also playing fast-and-lose with our nation's national security interests w/r/t his super dubious relationship to Megaspeed is good for shareholder how, exactly?
Wouldn't *not* doing these things be better for shareholder value, or is that just me?
Good thing we don't actually have to wonder. The results are in.
@invisible_hand2 So the stock is down 40% because people are being mean to Jensen?
Or is it because he... made choices... ?
People like you will literally jump in front of a train before you'd ever admit that these guys are responsible for the consequences of their actions in any way.
I am shorting the AI bubble by shorting the following names:
$CRWV
$ORCL
$NVDA
My Coreweave and Oracle short positions are very big; Nvidia short position is a bit smaller (bc better company).
But anyway, I think there is a good chance that it's all (or mostly) downhill from here for the AI labs, and the companies financially entangled with them.
While bubbles can go on much longer than one can remain solvent, I think that the catalysts I've been waiting for for many months now (since I published my AI bubble Substack piece in June) have arrived:
1) AI lab revenue growth rate slowing substantially as token prices continue to collapse
2) Coreweave's debt reaching 15% (distressed)
Revenue slowing is of course the key development, as it totally collapses the growth story, which needed to continue to be off-the-charts to justify all this CapEx.
Then, Coreweave getting totally dissed by debt investors (whom we all know deep down are the most rigorous/sophisticated) to me is proof that Real Consequences™️ are starting to arrive, as the first truly major cracks start to form.
Unlike most people on fintwit, I am super open about the positions I take w/r/t putting my money where my mouth is. So by all means, bookmark.
A lot of non-technical people are still underrating the extent to which coding is truly solved.
Even just six months ago, the people on here saying “software is solved” were super annoying, because it wasn’t *quite* true.
Back in, say, June, you absolutely still had to babysit Claude Code/Codex if you were working on something important.
However, since then, models like Opus 5.5 have removed all doubt, even among the final holdouts, that you pretty much don’t even need to look at the code anymore.
The past few months, on the coding front, have been another step change. I’m honestly at the point where I can’t even imagine how new models from here can even *be* any better at coding—at least in a way that actually matters for 99.99% of software use cases.
To me, just knowing this—and seeing how good these models are now at software firsthand (I am still a full-time SWE)—means the CUDA no longer has any moat by definition.
It’s not a matter of “if” CUDA’s moat will be destroyed (it was already destroyed the second that Opus 5.5 was released), it’s just a matter of who is going to actually scale the new solutions.
Why It Could Get Even Worse For Nvidia
It’s not just that hyperscalers are building their own proprietary chips. It’s also that, as @alexbhturnbull has argued, advanced AI is getting better at looking at any piece of hardware and writing optimized machine code for it, in a way that weakens Nvidia’s most important moat, which is its CUDA software ecosystem.
My conclusion:
“CUDA is one of the greatest achievements of human expertise in the history of computing: decades of code built and refined by human hands. If AI models learn to write superior code for any chip, Nvidia will become the latest student of the bitter lesson, and the company whose chips proved that mass compute can beat narrow craftsmanship would become the lesson’s most ironic victim.”
https://t.co/o5nvMyZmbf