Key points from last night's conversation with 168X: Last time at the end of August (https://(x.com)/qinbafrank/status/2094643546334916871?s=46&t=k6rimWsEbo2D2tXolYcM-A), we discussed the following: With the US midterm elections approaching, market volatility is increasing, and Bitcoin and cryptocurrencies are expected to pull back in October due to tightening liquidity. Here's a summary of the key points from this discussion: 1. The US economy is entering a new infrastructure cycle driven by AI infrastructure. This large-scale infrastructure trend will lead to strong nominal GDP growth. 2. The situation in the Middle East may present a window of opportunity after the midterm elections. If this can be resolved, the pressure on the macro environment will be significantly reduced as oil prices fall. However, if the situation remains unresolved and oil prices remain high with persistently strong inflation, the Fed will not only raise interest rates in December but may also raise them next year. If the initial expectation of two rate hikes turns into three, four, or even more, the entire valuation system will be repriced, and the market will be unable to withstand the pressure. 3. The ARR metric needs to be disregarded. After Anthropic's IPO, we'll need to focus on actual quarterly and annual revenue. Large cloud platforms are already AI operating systems: companies will switch between different models based on results and costs. 4. Large tech companies will truly begin to significantly increase CapEx around Q3 2025. Therefore, much of the revenue growth seen by Microsoft and Google Cloud today corresponds to spending made a year or even earlier. The capacity built up rapidly in the second half of last year is only now entering the production phase, while for the past few quarters, all cloud vendors have been saying the same thing: Demand > Supply, insufficient computing power. If demand continues to grow, and any new capacity deployed can be immediately absorbed, then revenue will naturally continue to accelerate in the coming quarters. Therefore, I believe AI commercialization is not at its peak, but has only just begun to accelerate. 5. We need to see when Cloud Revenue Growth begins to surpass CapEx Growth. We can't always look at revenue alone; the second stage is to look at ROIC. 2026 will be the year of CapEx explosion. Last year it was hundreds of billions of dollars, and this year it's nearly doubled. This CapEx growth can't be 100% every year. What's the truly beautiful inflection point? It's when CapEx growth starts to decline naturally, but AI Revenue Growth continues to accelerate, meaning things start to look profitable. Therefore, in the next few quarters, the most anticipated metrics are: Revenue Growth, CapEx Growth, Depreciation Life, and FCF—how these begin to converge. 6. The AI mainline has shifted: Memory → Optical / Software → Agent / AI for Science/Applications Don't form the fixed mindset of "AI = Semiconductor = Memory," equating AI with semiconductors and semiconductors with storage. The main line of AI shifts at each stage. In the first and second quarters of this year, Memory was the strongest; after August and September, Optical began to strengthen, and at the same time, a batch of software and SaaS that were previously heavily impacted started to emerge. Why? Because with model commoditization and the decrease in token prices, it's not just upstream hardware that benefits. The cheaper the token, the easier the deployment, and the more benefited cloud, software, enterprise integration, and agent companies will become. The next stage may be AI for Science, Bio, and a broader application layer. Therefore, the real task of trend investing is not to cling to the existing winners, but to continuously identify where the next profit pool is migrating to. Semiconductors aren't dead; it's just that the phase of everything surging upwards like in the first half of the year may be over in the short term. 7. The real signal that the AI Thesis should turn bearish isn't a 20% drop in Semiconductors, but rather that the commercialization of AI (cloud vendor revenue growth) begins to slow down, and that we are currently in a phase of acceleration. This is the real major risk in the future market. Another issue is the unresolved Iranian issue, high oil prices, a strengthened upward trend in inflation, and the Fed having to raise interest rates multiple times instead of twice. 8. The era of high yields will no longer be one of universal growth: the market will only reward real cash flow in the future. When bond yields are high, stocks will naturally offer higher risk premiums, which is why the market is becoming increasingly discerning. Google Cloud is doing well, but even with issues with Search and FCF, the market is still selling off. Therefore, when choosing stocks, it's crucial to distinguish between companies that are merely nascent and those whose businesses have truly materialized—whose revenue and cash flow are real, and whose growth potential over the next 2-3 years can be calculated. If large companies experience a valuation drop due to short-term headwinds, I actually think it's a good thing, because it gives you a price to buy back in. High interest rates don't mean all stocks can't rise; rather, the market will become increasingly selective. 9. After the midterm elections, many uncertainties will lessen, such as the fate of Congress. We'll see if the Iran issue can be resolved, and be wary of the drain on liquidity after Anthropic's IPO roadshow begins. Something I said yesterday seems incorrect now: "Yesterday we discussed that if Amazon's earnings are good, the market could surge before Anthropic's roadshow. Amazon's cloud business is doing well this year, but advertising is slowing down, and capital expenditures far exceeded expectations, which also exacerbated market concerns." Perhaps these concerns will need to be addressed after the midterm elections. Thanks @168MrZ