LBO_Guy (@The_Analyst_Lab) header
LBO_Guy (@The_Analyst_Lab) profile picture

Pushing the limits of Microsoft Excel.

Not Financial Advice
Joined January 2023
8B tokens a week looks like my steady run-rate. One thing I've noticed is that even if I keep moving work from the model into plain code (my Sep 28 setup would probably cost 70% less today), the savings never show up!! Every optimization frees room for new ideas, so I spend the same or more. Probably do 3-4x the work I did on sep 28th but spend exactly the same amount of tokens. All subscriptions maxed, that is the limiting factor for me. If I had an unlimited budget I'd spend much more. We are europoors after all. Anyways, next couple of weeks will probably run higher as there are a few interesting things I wanna do. It really is astounding the amount of work you can do, LOL.
2
6
669
Dwarkesh is probably one of the most intelligent, out-of-the-box thinkers I follow, and he’s now more bullish on superintelligence than ever (until recently he thought it was a decade away). I agree with him. I fully recommend watching the explainer. Truly phenomenal. Would love an interview between @patrick_oshag and @dwarkesh_sp. As AI gets obviously better at everything you can check, from maths and coding to AI research itself, the timelines keep collapsing. It’s truly bizarre to look back at what these models could do in 2024, 2025 and 2026, and it keeps compounding as Dwarkesh puts it "from failing four-digit sums in 2023 to OpenAI claiming a Millennium Prize problem in 88 hours this autumn." Basically the core of the idea is the following: AI research is mostly recombining ideas and running experiments against a score, which is exactly what models already crush. If you automate that, then you get thousands of top-level researchers working at 10x speed, 24/7, squeezing years of progress into one. Dwarkesh puts his median at 2028, although it could be longer or sooner obvsly. Daniel Kokotajlo, one of the few people whose predictions have actually held up, now sees it in 2027. Back in 2021, more than a year before ChatGPT, he called chatbots going mainstream, scaling continuing to work, multimodal models, clumsy early agents that slowly got useful, and the US–China chip war. So pretty accurate guy I'd say. Now METR estimates AI is already accelerating Anthropic’s progress by 1.7x, way ahead of what even he expected. If this is right, the read-across is pretty clear. Spend keeps going like crazy, because whoever gets there first wins everything. Nobody will open-source a frontier model when what you have in-house, and how fast it’s improving, is that valuable. You can already see it at Anthropic: it explains why they want to go public and raise $90-100bn at almost any price. They don’t care about a lower valuation as long as they get the money, because they think they’re close and thus want to do the IPO ASAP. Compute, memory, power and fabs remain the bottleneck. Especially memory and fabs. The only real risk I see is political. Something bad, like a cyberattack on a big bank or a big global company, scares people off and forces a slowdown, or populism keeps rising and people push to pause or slow data-centre development, which would be ridiculously stupid. Anyways, recommend watching the video.
1
2
9
902
Ouch, Claude is out as well.
3
7
807
A few thoughts on Jensen Huang and $NVDA. The long-term view could be even worse. It should be obvious that the board should wait for the investigation to end before voting on whether to oust Jensen. Despite media reports, I don’t believe they’ll oust him before the investigation is finished. It doesn’t make any sense, and they could all get sued for breaching their fiduciary duty. I would also add that unless something big (like deceiving the board) they'll choose to keep him. On the Jensen situation, as @benthompson put it, the real threat lies in the cover-up: whether he deceived his own leadership, or whether it was just personal misconduct (i.e. he was unfaithful). Nvidia’s legal counsel was denying the affair just days before the news broke, and apparently Jensen had secretly hired a crisis PR firm a year earlier. Not a good look. The board may have been misled or kept in the dark, and if that’s true, this really is a big issue. If directors face shareholder lawsuits and personal liability, they will ultimately choose to protect themselves over Jensen. By the same token, this is why I believe it makes no sense to oust him before the investigation is finished or there is conclusive evidence that he misled the board. They’d get sued in that case as well, because they couldn’t show they based their decision on the investigation’s findings or on conclusive evidence. What would be funny is Elon poaching him if he gets ousted. Either way, if he gets ousted, it’s really bad for NVDA and really good for Elon. The “good” part is that this can be solved "fairly quickly” if they decide NOT to oust him, much like the COVID market crash could be “solved” easily by lifting the lockdowns. The share price action is mostly funds rotating out of NVDA, because you can’t hold a big position in the midst of this crisis. However, there is a longer-term issue here, and I believe it is much more important than people are giving it credit for. The real issue, even if Jensen is NOT ousted, is what happens to Jensen’s credibility. A man is only worth as much as his word. Remember, NVDA (per Jensen on the Dwarkesh podcast) has NEVER had a contract with TSMC! “Nvidia and TSMC don’t have a legal contract. There’s always some rough justice. Sometimes I’m right. Sometimes I’m wrong. Sometimes I got a better deal. Sometimes I got a worse deal. But overall, in the whole, the relationship is incredible. I can completely trust them. I can completely depend on them.” Everything was based on trust, trust in Jensen. I don’t know what they’ll think after seeing what he did to his wife of more than 40 years. If he could do that to her, can you really trust him without a contract? If he as well deceived his own board (this would worsen the look on Jensen) could you trust him? Probably NOT. And with a contract? Well, it had better be bulletproof🤣 But that is not how the people at TSMC like to operate. So it’s “no bueno”. And it’s really bad, because NVDA’s moat today is that it has locked up most of TSMC’s CoWoS capacity, as well as memory capacity from SK Hynix, Samsung and Micron. That’s it! Forget all the BS people talk about. That is essentially NVDA’s moat, and why it can capture so much of the value chain’s profits today. CUDA (software) is no longer a moat, and it’s pretty obvious that competitors are getting really close to NVDA’s performance (as Google’s new TPU8 generation shows), and they will only keep closing the gap. Gemini Carbon is ridiculously cheap while sitting pretty much at the frontier. We’ll see how this evolves, but I see a lot of people jumping in to buy the stock, most of them knowing nothing about it. Even my barber knows more, LOL. What its pretty clear from all of this is that the only thing that got impaired and damaged is Jensen's credibility. Will see if ppl care or not (I would). Another pretty obvious thing is how all this has turned out to be. There seems to be a psyop to oust him or impair him in some way (damaging his credibility). Clearly, what I've realized is that Jensen, despite being utterly good at his job, compared to Elon or most of the paypal mafia boys, is an NPC. You cannot get dropped like this, LOL. What an NPC. Elon is on another planet on its own, despite all the interests in getting him dropped, he has a grip on everybody. Much much much smarter than everybody else. No wonder why ppl bet on him.
1
3
7
1,295
MacBookPro with M4 maxed out all the time with claude+codex+grok, LOL. Never seen this, really tough to max out the Macs before, not anymore. Had the macbook restart lots of times these couple of weeks do to it being freezed. Might need to get a new one.
Mas vale que $AAPL no hacia caso a nadie y seguia chetando cada vez mas los iphone desde el A11 bionic porque desde que tengo grok bot el telefono esta ardiendo y la bateria va a tutti. Eso de que nunca ibas a usar ese compute capacity no se Rick, ahora ya si. I'm holding back!
1
567
El podcast que hace Dwarkesh con Jensen es bastante revelador y algo que cuando vi en junio me sorprendio bastante, y mas aun que nadie lo comentara (casi peak del mercado asique nadie queria comentar esto obv). 2 cosas que me lleve del podcast son 1) Dwarkesh es prob con bastante margen el mejor entrevistador que hay en internet ahora mismo simplemente por el conocimiento que tiene, lo inteligente que es y el hecho de que no tiene miedo ha hacer pushback y 2) que a Jensen se la pela todo con tal de vender una GPU mas y si le haces pushback se pone a la defensiva tratando de dejarte en ridiculo, pasa que el que le deja en ridiculo aqui es Dwarkesh.
Dwarkesh: Why would we want to sell China the materials for a serious cyberweapon? It's like selling them nukes with a casing that says 'made by Boeing' and claiming that's good for the US Jensen: Comparing AI to nukes is lunacy. Enriched uranium is a lousy analogy. It's an illogical analogy. What we have to recognize is that AI is a five-layered cake.
2
2
32
6,374
Yo, even utilities ain’t move like this😭👨‍🍳 Is this what “hedge fund” really means? #marketneutral
Falling less than any major US index (let alone memory -10%) due to massive convexity brakes feels like walking on water tbh. Crazy sht how important portfolio mgm is. Yet few use options like they are supposed to. Delta exposure drops in a straight line, bruv.
1
2
831