What is the current AI market position compared with the foam of SciDev. Net? First, let's make a conclusion: the current market is not the same as the foam of Scinet in 2000 The thing is like this. Recently, many brothers asked me, is the current US stock market repeating the 2000 NASDAQ foam? I was stunned for a moment. Can this Nima be the same? I can tell you that the difference between the current market and the foam of Scinet in 2000 reflects the difference between the AI industry trend and the Internet industry trend itself. Because the marginal cost of the Internet is close to zero, investors value the expansion of scale more than short-term profits. Causing the PE of popular stocks at that time to be outrageously high. Because everyone doesn't look at PE at all, only at the growth rate of user scale and industry space. But the marginal cost reduction of the AI industry is not significant. So investors value actual orders and AI revenue growth more. The profit growth rate of technology giants is generally higher than the stock price growth. The profit growth rate of hardware companies for basic equipment is generally astonishing. Some software oriented companies are experiencing significant compression in their valuations. As a result, the current PE is much lower than at that time. So I don't recommend making a simple comparison between the current situation and the year 2000. It is easy to be misled. The top of the major industry trend cannot be seen in advance. Don't waste this time. The performance growth rate of the US stock market is very good now. The correct attitude is to accept the foam of technology stocks and accept the future yield pullback. Different from the strategy of A-shares. Why are there so many differences in demand forecasts for optical modules among different companies? Just using this question, let me introduce how the seller makes quantitative predictions. For example, demand forecasting for optical modules. First, let's take a look at the capital expenditures of several major cloud providers in North America and add them all up. Convert capital expenditures into AI server/GPU/ASIC shipments. Multiply by the number of optical modules corresponding to each XPU. Then adjust the prediction results based on this quantity. At least four factors are involved: The actual consumption demand of the terminal, customer procurement lock order demand, supply chain construction plan, and final actual shipment volume. The result is inevitably a huge difference between optimism and pessimism. At present, there are serious supply constraints on computing power. So optimistic predictions are based on the demand port path. Based on the assumption of continuous upward revision of CSP capex in North America, high volume of GPU/ASIC, and accelerated penetration of 1.6T. Relatively cautious predictions often take into account the conservative actual shipment caliber on the supply side. Or under slower 1.6T import rhythms and stronger upstream device constraint assumptions. Even if all links increase in volume, any bottleneck will affect the shipment volume. So, the divergence of 1.6T is much greater than that of 800G. The reason is that the demand calculation for 1.6T is extremely sensitive to architecture upgrades and supply. If we assume that mainstream platforms are accelerating their migration and each XPU is equipped with a significantly higher number of high-speed optical modules, the demand will exponentially increase. Why insist on configuring IGV? This question is very good. Why should I configure 5-10% IGV in the global configuration when the application has fallen to a dog? At that time, it was still at a high level. How should we handle this portion of the position? This is actually the difference between ETF allocation thinking and individual stock investment thinking. IGV is a software industry ETF. The future of software is not about not needing it, but about performance differentiation. Actively embracing new AI leaders to replace old ones. So software has configuration value. It's just that its rhythm is difficult to judge at the beginning. Is the process of differentiation first falling and then rising, or first rising and then falling? If it's a stock investment, it's not wrong to avoid it first. But ETFs are different. During this process, the weight of rising targets increases, while the weight of falling targets decreases. So there must be some turning point when it falls. A round of rise driven by new leaders. Just like the rebound since April, driven by new bull stocks such as PANW and CRWD, their weight continues to rise. This is the difference between ETF allocation thinking and individual stock investment thinking. It values industry trends more than timing to seek benefits and avoid harm. Another reason is that IGV only has 5 points in the combination. Equivalent to a style of insurance. Because the S&P 500 Ghana Index or other technology ETFs in the portfolio sometimes exceed 30 basis points. With the increasing hardware, the proportion is getting higher. So we left five points for the application to balance the style. Who knows when the wind will turn around? After all, the warehouse is only adjusted once a quarter. This also involves the difference between aggressive thinking and balanced thinking. The more balanced your style, the more mediocre you will be in a bull market, but the longer you will live. The more exposed your style, such as a bunch of hardware stocks, the more likely you are to achieve high returns in a bull market, but also the more likely your returns are taken away by a wave. In short, asset management and stock trading are two different things. Every bull market in A-shares is ultimately a plundering of wealth by ordinary people. But the US stock market is different... Don't underestimate every adjustment in the later stages of a bull market? Fundamentally, AI in both the United States and China is accelerating. Is it difficult for A to kill the bear right away here? For an index, both one-way upward and downward trends are rare occurrences. You say 'I won't kill A', which is equivalent to saying that a small probability event won't happen. What's the point of this? Retail investors' losses during bull market adjustment periods will not be less than those in bear markets. More importantly, the elasticity of different stocks varies. You have been holding dividend stocks, although you haven't made much money so far, you may not lose money in the future. But technology stocks have huge elasticity. If the buying position is not good, the six-month return has already been lost in the past two days. If it continues to decline, it will be last year's earnings. In the later stage of a bull market, investors' mentality will become more or less radical and impetuous. You can have confidence in holding stocks strategically for the overall trend. But tactically, we also need to have reverence for every adjustment in the market. Many people's hard-earned money during bull markets is taken away in a later wave of small-scale adjustments. Value investing in the US stock market is the king...