ASI

SK Hynix is a great buy! Global market leader in HBM (AI memory) Forward P/E < 7x Based in South Korea with a strong, stable regulatory environment Operating in a pro-business, pro-semiconductor U.S. policy environment Actively exploring expanded U.S. market and listing presence, huge! Earnings in about one week Yes, memory is cyclical long term. That does not change how compelling this setup is right now. Which other company is the market leader in a high-tech, high-growth segment, with demand growing around 30% per year, trading at single-digit forward earnings? It was great in 2025 It is still amazing! KRX 000660.KS EU ticker HY9H.F

@crystalsssup 🤣

cloudflare down again 🙃

The market is crashing today and we just came out of a long drawdown after telling people to sell on October 31, almost perfect timing. This is a moment to start dipping back into risk. TSM stands out as long term value with strong demand and attractive pricing again.

Look at your leverage and risk. See Bitcoin as two times tech. Tech is two times US stocks. US stocks are two times the rest of world. Rest of world is two times the rest of all assets. Altcoins are two times Bitcoin. Ask yourself: would you put all your money into Nvidia or Oracle at these levels. If no, cut your Bitcoin exposure. If yes, increase it or use leverage. You can also split the trade and hedge the position. Long Bitcoin. Short something else.

Okay, I'm officially getting nervous about my Bitcoin Should I sell?

@KobeissiLetter Many of us called it out last Friday. The drop was coming. Too much hype, leverage, and circular buying funding and spending with no real efficiency. Going forward hedge your positions.

Today is one of those days: Just about every asset class is trading lower today and all intra-day rally attempts are being sold. It's simply widespread profit-taking. In our view, nothing has changed fundamentally speaking. That said, the most healthy bull markets experience periodic declines. In fact, the S&P 500 averages at least 3 declines of -5% or more PER YEAR, despite averaging a +10% annual gain. The reality is that Magnificent 7's CapEx ALONE is now set to exceed $500 billion PER YEAR. Rate cuts have arrived, deregulation is here, earnings growth is running at 10%+ YoY, and the AI Revolution is accelerating. Ignore the noise.

What would end the AI boom and crash the stock market? All it takes is one trigger, a small efficient model that matches today’s leading AI systems but runs locally on a home pc or phone. Recent trends pushed for bigger models with heavy compute like reasoning and chain of thought, but that is not sustainable. Smaller models can offload data retrieval and run multiple agents in parallel. That is how we work today. If a pro level local model launches, tech stocks could drop forty percent or more. Current valuations assume we will not innovate smaller and smarter. Many times, less data produces better results than overwhelming a core model with too much. At least hedge your positions. All that circular spending from OpenAI, Google, Microsoft, Meta, Amazon, and Oracle into NVIDIA will look like massive overspending when that shift comes.

We had AGI last year and now we are seeing ASI in some fields. There are many examples where asking an AI gives better results than asking the smartest human professor in their own field. But asking an AI researcher how to use AI is like asking a football coach to score a goal himself, it is not their job. The people building the models are not the ones using them. Someone who spends hours a day practicing with AI, experimenting with prompts, working to remove their own bias, and building agents can get results that are a thousand times better. For example, Apple Research recently published The Illusion of Thinking in June 2025, saying that large reasoning models show accuracy collapse on complex tasks like Tower of Hanoi. But I solved those in one shot with AI and replied in this channel before that paper. ASI is here, adapt to it.

I just listened in on a Bitcoin space and these guys basically think that AI is already super intelligent AGI level. Meanwhile, I talked to guys who work in the field and they're like yeah some people are going to lose their jobs but we're nowhere close to AGI.

@tshevl Great story/journey. Best of luck on it! Doing something similar for financial markets, politics and world events. We model anything that moves the key assets like gold, stocks, crypto, rates etc. The starting goal was 1% per day, but now we are aiming higher with ASI.

I got back from honeymoon last summer and handed in my resignation at DeepMind. My wife thought I was crazy. AI has always been about prediction, but normally we predict small things: a token of text, or moves in chess. The ultimate challenge is to predict the world’s most important events. We recently went up against some of the world’s top forecasters, and came much closer to beating them than any AI system before. We're used to seeing crazy results from the AI community, but I think this one is special: 1. Accurately forecasting global issues is extremely difficult. 2. You can’t memorize the answer: it hasn't happened yet. 3. It was considered very unlikely for an AI system to do as well as Mantic did (5-10% chance). 4. Superhuman forecasting has the potential for transformative impact across the economy. She still thinks I’m crazy, but less so every day😛

You measure how much intelligence each person can get out of AI, can they reach AGI or even ASI? Then you pay those who can reach ASI (Artificial Super Intelligence) to help skill up the rest. One simple way is through problem-solving with unlimited AI access. A more advanced test is to give each person a set amount of money to invest using AI, then measure their results, reasoning, and how far they can push ASI performance. You use the top performers to teach the rest. The key skills are understanding ASI, removing bias, and realizing they know nothing compared to it. Never accept anyone who thinks they know better than AI, AGI, or ASI in any field unless they’re willing to change.

how do we reskill folk when the gpus are more skilled than they are

@PeterSchiff Congratulations! Must feel good seeing it play out.

I just woke up to see gold blow through $4,000. It’s up over $50, trading above $4,035. Silver is up over a dollar, trading just below $49. Gold mining stocks sold off yesterday as nervous investors expected $4,000 resistance to hold, resulting in a correction in mining stocks.

Thanks, I follow your posts btw, they provide pretty good insight. Algorithmic patterns are interesting. We timed the $4000/oz spot well. Newmont is already up 2%, making it a solid trade. Major outlets like Reuters and other news sites have been publishing headlines about it, with more coverage coming out in the last few hours. This has pushed more news about gold stocks moving for the day, also mentioning the $4K spot. It creates a reinforcement loop where algorithmic systems do sentiment analysis on the words and start buying those stocks. We use ASI and analytics to predict upcoming headlines and adjust positions before they happen. We also had several other factors lining up at the same time. Gold spot hasn’t dropped yet during China’s Golden Week, and every drawdown has been met by buyers stepping in. That shows strong demand. There’s an uptick in gold interest in India and Japan, with inflows into ETFs, not just people buying gold products. We think Newmont has strong fundamentals overall, but we trade the patterns.

Newmont is down ~1 % today, even as gold spot pushes toward $4,000/oz. Hitting that level could trigger algorithmic headlines and buying. Fundamentals remain solid, and the CEO transition was telegraphed well ahead of time. Still bullish, plenty of upside ahead.

Klarna is up 9.72% so far today. A good trade. The rest of the stocks are also pushing strong. We don’t normally publish our recommendations, but if there is interest, we will do it more.

October has the chance to be a fintech month. Klarna just entered lockup and now has shown an imbalance to the buy side (before it was the opposite). Affirm had the CEO selling and is the preferred provider to Shopify, which has OpenAI agent integration. Block and PayPal have BNPL action and are getting some traction. Coinbase is tied to Bitcoin, which is starting to move again (mirroring risk-on, Mag 7 etc). Robinhood entered the S&P 500 in record time and is cooking, with new legislation that might allow tokenized trading. We’re just saying, the AI play has been the best the last couple of months, but it doesn’t hurt to diversify a bit into fintech.

Nvidia committed to buy CoreWeave’s unused capacity through 2032, a very long horizon. Likely they offered Meta incentives to shift risk. Think Cisco’s vendor financing in the dot-com bubble with weak customers like Cambrian. Now with the “Mag 7”, Nvidia, Oracle and OpenAI it’s the same pattern: huge debt, financing and guarantees flowing into relatively small companies, with each player guaranteeing or financing the other. It’s extreme concentration of risk.

Meta, the second-largest customer of Nvidia, which has already purchased GPUs worth tens of billions of USD, has placed a new order of up to $14 billion with CoreWeave—excluding its order with Oracle. Why does Meta need all this computing power? Why did they choose CoreWeave? And was Nvidia (or Jensen) involved, given that Nvidia has committed to purchasing any unused cloud credits from CoreWeave?

TSMC is a great company, growing ~40% and executing flawlessly. The real moat is not just advanced nodes, it is yields per wafer and scale. Intel can get subsidies, but they cannot copy execution. At less than 25 P/E, TSMC is a great buy.

@Beth_Kindig It’s not a historic anomaly. Nvidia is about 6.4% of the U.S. market cap. AT&T was in the 1960s about 13%, which is double Nvidia’s share. Many companies around the world have also surpassed 10%.

Nvidia $NVDA recently reached a nearly 8% weighting in the S&P 500, the highest single-stock weighting since data began in 1981, and surpassing Apple’s $AAPL 2023 peak at 7.24%.

511 moves. A larger disk can never rest on a smaller one, this rule forces all moves. Below is a compact Python that shows each move: def h(n,a="A",b="C",c="B",r=None): r=[]if r is None else r if n==1:r.append(f"{a}->{b}") else:h(n-1,a,c,b,r);r.append(f"{a}->{b}");h(n-1,c,b,a,r) return r if __name__=="__main__": n=9 s=" ".join(f"{i+1}:{m}"for i,m in enumerate(h(n))) for i in range(0,len(s),275):print(s[i:i+275])

$10M TC Quant Reasoning Interview Question:

@nathan_covey Patience, $5 M will look like a bargain if you wait long enough 😄 But by then we might skip URLs altogether and have AI/AGI/ASI talking to ASI on our behalf. Still feel like a 100× goal if clicking itself goes extinct?

Ok. I have a goal to get this domain

Full o1, thanks @sama I've been waiting for this, and it's everything I wished for! This is the best Shipmas ever!