Greg

Largest portfolio positions: Private: OpenAI Anthropic LightMatter Databricks Public top 12 positions: $AMZN $IFNNY $MU $VRT $TSM $QQQ (short) $AMD $AMAT $NVDA $BESIY $GOOGL $P

@SantiagoAuFund This is the way…

Company is toast clearly #openAI

i knew Codex was growing fast, but this is unbelievable: 1m → 8m users in just over five months even crazier, it went from 6m → 8m in only three days

@pboockvar Hidden leverage has to be pretty high at this point

Interesting nugget from Wells Fargo earnings call: “Securities based lending has been a key driver of loan growth, with average balances up 31% from a year ago, reflecting our success in increasing the number of financial advisors offering this product to their clients.”

@Jason The ‘Uber’ization of the workforce is a terrible outcome for society. It’s essentially an economic trap for workers. As a society we should really be focused on upskilling these people.

The first AI “stealing” humans' jobs war will be fought over autonomous driving. And it’s getting quite spicy The most popular "settlement" will be cities selling a small number of autonomous medallions for $25,000 each year (~$3 a ride), over many years. Those fees will be earmarked to pay for unemployment and retraining of human drivers China is far ahead of us on autonomy, and they have frozen permits for self-driving over unemployment concerns The CCP doesn't value human life; fuck, they ran over their own children with tanks in Tiananmen Square. They're hardcore murderers who will use torture to control their citizens from using a VPN or reading the wrong book. But the CCP is very sensitive to worker protests. They will delicately smooth those issues over, and they have with limited self-driving permits. America doesn't care about people protesting; people protest losing the McRib here. [ disclosure: I’m a huge fan of self-driving, and I’m invested directly/indirectly in almost all of the providers globally. I saw this slow train coming since the DARPA challenge ] I have unlimited empathy for folks losing their jobs, especially older folks who might not recover from it — I've watched this struggle with family members. It’s brutal for a person who worked for 30 years to be tossed aside by a society they helped build. Can you imagine being a livery driver for 12 hours a day for 20 years and your prize is being replaced by a robot? That's really depressing. As such, I am a fan of creating a soft landing the displaced (and tens of millions will be), while this technology is thoughtfully rolled out to save lives. You can actually be empathetic, reasonable and not dogmatic in a complicated situation like this. And you should be. You can judge a society (and a man) by how they treat the poor, the disenfranchised and the people who are simply down on their luck. Today powerful folks get a kick out of not having empathy, because it plays on the pods and in the ballot boxes. This is a historical blip, I can assure you. Jesus taught us the way, as did many other enlightened individuals like the Founding Fathers. ❤️

@garyseconomics How are you getting interviews on TV. It’s like listening to a five year old explain a topic to their parents.

Working people are paying 50% tax while some billionaires pay nothing

@TheStalwart In other news: NY continues to shoot itself in the foot with dumb policy choices

New York has become the first state to announce a datacenter moratorium. There will be a one-year ban on new development while the state assesses their impact on energy prices and the environment. nytimes.com/2026/07/14/nyr…

Think about it from this perspective. $META has thrown a ton of resources at trying to compete with the top two labs and is coming up short. Their hiring spree is in reverse with key talent leaving to go back to OAI & A\ despite handing out huge pay packages. I’ve been told they have a culture problem and poor leadership in AI. xAI is distilling off of the leading labs which musk admitted in court. They don’t have sufficient demand to keep DC utilization high so now they are selling capacity.

@GS_CapSF @corleonecapital I’m talking about the next 12-months. You are talking about the last 12-months. In terms of access to compute and scale, and pace of improvement - seems pretty obvious to me where the probabilities are shifting. SpaceX also has Cursor, which is a big deal IMO.

OpenAI is worth a lot less today than it was before $META released muse spark 1.1 If Meta is successful, wouldn't be surprised if OpenAI crumbles and never makes it to IPO...

What are you guys talking about?!? Codex users have gone from 1 million to 7 million in 5 months. ARR from $25B to $40B in 6 months & that includes re-writing rev share agreement with $MSFT. They launched on $AMZN AWS & they have the best model right now. It’s actually embarrassing for $META & $SPCX they released their best model and they lag OAI & Anthropic models that were released a while back. We’ll see what $GOOGL does but it’s a two horse race.

@HedgeyeComm Think if they stay frontier on coding, they’ll do well, but they better hope Meta and/or Google doesn’t match Codex on usage, then it’s game over on the B2B side OpenAI having better models than Google for a while now but still losing share is concerning too.

Narrative: OpenAI is on the verge of collapse Reality: Codex growing at >800% annualized

$TSM revenue accelerated to +67.9% yoy in June to an all time record. Trailing 3 months revenue accelerated to +36% from +30% in May. What is the bear thesis here for investors?

@haugejostein China is mercantilist

The two fastest-growing economies in the 21st century have been China and Vietnam. Both follow a model that's shockingly easy to summarise: a socialist market economy focused on export-oriented industrialization.

@omgsidewalks Says a guy typing on a smartphone and posting on a social media platform whose products were brought to market by founders who created enough value to become billionaires

Billionaires don’t create jobs. Stop repeating that myth. Without billionaires, people would still build, design, teach, produce, sell, and buy things. Billionaires don’t create human productivity, they capitalize on it and concentrate the profits.

I think the ultra-bull case is on-prem infrastructure. Who wins the model layer may not matter. Token throughput & TCO get good enough companies decide to deploy ‘small’ clusters while maintaining data sovereignty. Already seeing some leading companies doing this like Eli Lilly which stood up their own GB200/300 cluster.

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. It would increase the ROI on AI spend for end customers by increasing intelligence per dollar, which would drive incremental token demand. Margin dollars would effectively get redistributed from the frontier labs to AI infrastructure providers. The infra winners would be those with the lowest per token cost and the winners at the model layer would be those with the highest token efficiency. There are many reasons Jensen is so focused on open source, but this is likely the most important one as I think he is probably less worried about a monopsony these days. Lower margin % at the model layer = more margin $ at the infra layer all else equal. With SpaceX and Meta being vertically integrated and possessing the #3 and #4 models respectively it is more possible than ever. Note that Grok 4.5 is ahead of Fable for some useful tasks at a much lower cost, so ranking them #3 is conservative. This is not happening yet. Cheap, mostly open source tokens are likely the majority of volume today but the majority of economic value is still accruing to the most intelligent models. Might change though. We will see.

big milestone!

@firesidealpha How does he explain these benchmarks then

Dylan Patel reveals his firm hands OpenAI only the jobs it can run overnight, and keeps everything else on Claude, because OpenAI burns 3-4x the tokens: "This is why Anthropic has been beating OpenAI, because their models are more token efficient than OpenAI." "OpenAI's models, on the edge cases, in terms of leading science, leading math, leading code, oftentimes can do a task that Anthropic models cannot. But they take 3x as long and 4x as many tokens, and therefore it costs a lot more, and the feedback loop of human and AI is not as rapid, and therefore it actually ends up being worse on a customer perception basis." "Because it's one thing to say, hey model, do this task, and then you come back and check if the task is done. It's another thing to say, I have four hours to do this task, and whether it's one call to the model and it does work for four hours, or four calls to the model and it goes back and forth, which one does it better?" "It turns out Anthropic, when you have this human-in-the-loop feedback loop, is actually way faster and better because it's more token efficient." "And so that's the main reason why we still remain a majority Anthropic shop. Some tasks people do use OpenAI, and oftentimes the tasks they let run overnight are the ones they give to OpenAI Codex. But most tasks they keep with Claude Code." _______ Follow @firesidealpha for more highlights of key business and technology conversations.

@howardlindzon That would be devastating

If Open AI disappeared tomorrow would the market even flinch? I know users would not care

OpenAI leading again

Reserves slowly rising after Fed forced to pivot as STIR rates started testing Fed policy bounds. Read about it here: @marginofsafetyblog/note/c-186417022?r=1gxqq&utm_medium=ios&utm_source=notes-share-action" target="_blank" rel="nofollow noopener">substack.com/@marginofsafet…

@edels0n Government support? That’s a just not correct.

OpenAI and Anthropic account for an estimated 60-80% of the AI revenues generated by Amazon, Google, and Microsoft. Without them, this whole thing collapses. And OpenAI just told us they need government support. Uhhh … that's not good? 👇