Weekly Review

You still can't hold storage stocks for the long term in the future. After peaking at the end of June, there are too many trapped positions. The advice to buy storage at the beginning of July is a 'hard mode'—it's difficult to make money.

The three-day crash followed by one day of surge in storage stocks was due to Wall Street's setup involving the 'genius youth.' After his margin call triggered a liquidation, short sellers covered their positions, causing a spike. However, profit-taking led to a pullback. That said, getting married over the weekend is more important.

The AI narrative won't fail or die prematurely, but it doesn't mean you should blindly be bullish. It's like how everyone goes through life after birth, but some don't reach the end due to illness or other reasons.

Cloud computing has been bearish for two months, but this week added $1.5 trillion in market cap, driven by improvements in fundamentals. This reflects the AI monetization capabilities shown in Microsoft and Amazon's earnings reports, as well as Microsoft's trend of gradually controlling Capex spending, which pulled Google and Oracle along with it. I was always bullish on Microsoft and Google when we played them, and I told everyone to hold on. There shouldn't be any losses now, congratulations.

This aligns with one principle: 'The market will definitely gradually shift speculation from shovel sellers to cloud providers, software, large models, and physical AI applications over the next three years.' Value investors who just 'lie flat' holding semiconductors might underperform the broader market even after a year.

Even if TSMC breaks its previous highs, the upside is limited. The 畸形 (distortion) in the AI chain development stems from semiconductor hardware makers having excessively high profits. Computing power/storage scarcity has already been priced in. Yesterday, we also saw Kioxia's excellent earnings report missing expectations... Including bloggers who recommended holding Nvidia tightly in 2025 and those recommending holding storage tightly in June 2026, none seem to sense that this expectation is peaking.

In the past two years, trading Nvidia might not outperform QQQ unless you can accurately find the bottom. Storage is also returning to rationality, with limited space. The internet may see more influencers who haven't realized this leading retail investors astray, which is hard to avoid.

The flaw lies in the irrational rally from April to June. AI infrastructure, including chips, storage, and optical modules, surged significantly, dragging along sentiment and expectations. The stock price drop in June wasn't due to fundamentals, but because sentiment and expectations peaked, capital fled, and combined with the collapse of faith/patience in Musk's two companies, the entire US stock market experienced schizophrenic volatility—a hell-level difficulty.

I built a model where the decision-making for entry points = 50% FOMO emotion (chasing highs/cutting lows) + 30% listening to KOL opinions + 20% reasonable self-trading logic.

I treat this 30% KOL input as a contrarian indicator. The biggest problem with KOLs is the fan economy; they say whatever to please fans, shifting their cognition from objective to subjective. Many crypto KOLs have started trading US stocks, totally messing things up. Currently, three of the four 'stock gods' have fallen: White Hair, ByteDance, and Genius Youth. Only Trump remains.

So, the remaining task is to minimize this 50% portion as much as possible.

Recently, I fed the recently released earnings data into Codex and analyzed the problem from a mathematical perspective. The result showed that overall performance still exceeded expectations. Therefore, earnings data will continue to support S&P index gains in the coming months. The flaw is that the expectation kill isn't complete, with various factions fighting, leading to high volatility without a clear direction. For now, I won't take a heavy position.

This week, due to facing the storage decline, I posted daily updates frequently. Next week, I will only post weekly reviews.