Regarding the AI major section and the US stock market, please sort out the scattered views we have discussed before: 1. Now is a stage of accelerating the commercialization of AI At present, the penetration and popularization of new technologies are accelerating. As early as February, here is https://(((((((x.com))))/qinbafrank/status/2023982690144579970? S=46&t=k6rimWsEbo2D2tXolYcM-A has talked about that: mobile Internet has completed its journey in less than 10 years, and it takes 20 years for AI to complete its journey in less than 10 years; When infrastructure is well-developed, information spreads rapidly, and various commercialization methods are available, this leads to an acceleration of commercialization. https://((((((((((((x.com))))))))))))/qinbafrank/status/2052316328246247804? s=46&t=k6rimWsEbo2D2tXolYcM-A 2. The indicators for measuring the commercialization of AI are actually being migrated Previously, we used the ARR of two cutting-edge models, Anthropic and OpenAI, as the most important indicators to measure the commercialization of AI. The market trend from April to June in the first half of the year was that the ARR of Factory A was accelerating. But in fact, the performance of super large cloud providers after July has become the most representative and comprehensive indicator for measuring the commercialization of AI. The financial reports of Microsoft, Amazon, and Google fully supported the market situation in August and September. The core logic is that both enterprises and individuals are dividing task scenarios, with some high-value complex scenarios given to cutting-edge models, and other low value high-frequency scenarios given to open source models and medium-sized models with a focus on cost-effectiveness. It just seems that the current market has not completely turned around to accept this change. The earliest we talked about this logic was in early July. Last week here, https://((((((x.com)))/qinba frank/status/2110568431192064077? We have talked about s=46&t=k6rimWs Ebo2D2TXolYcM-A once. The most detailed is here https://((((((((x.com)))))/qinbafrank/status/2076839550119444702? S=46&t=k6rimWs Ebo2D2TXolYcM-A has been discussed. 3. The commercialization of AI is accelerating (which is the biggest difference between now and the year 2000), so the probability of a century high peak like the year 2000 is not high, but the fluctuations are becoming more and more frequent. Last week here https://((((((((x.com)))))/qinbafrank/status/2109239495266906227? S=46&t=k6rimWsEbo2D2tXolYcM-A talked about that in 2000, the Internet foam burst, and technology stocks continued to be depressed for a long time. The core is not the foam, but the Internet did not find a business model around 2000. From 2000 to 2006 and 2007, the four major business models of the Internet, namely, games, advertising, e-commerce and value-added services, were really established. The current situation is that the infrastructure is very complete and the information transmission speed is very fast; So the speed of AI commercialization is far faster than ever before. The probability of a major decline at the level of 2000 or 2008 is not high (due to both industry fundamentals and macroeconomic regulation measures). But as in the past five years March 26 Iran War, July deleveraging; The tariff war in the first quarter of 2025, the government shutdown in November of 2025, and the liquidity shortage at the end of the balance sheet reduction The decline of the 24-hour Japanese yen carry trade has impacted, and inflation rebounded in April with a small-scale adjustment; The mid level adjustment brought by the surge of treasury bond at the end of the third quarter of 2003 and the fourth quarter of 2003, and the shock of the thunder of Silicon Valley banks at the beginning of the year; The 22 year inflation report has undergone a large-scale adjustment with crazy interest rate hikes. This kind of adjustment to varying degrees will not be less, and market fluctuations will become increasingly frequent. A year ago, here was https://(((((((((x.com)))))/qinbafrank/status/1984556367093395469? S=46&t=k6rimWs Ebo2D2TXolYcM-A Finally, we discussed that the amplification of fluctuations comes from: 1) The rise of AI leads to the collapse of the middle class, and there is a stronger tendency to gamble in order to make a leap; 2) The more productivity develops, the less people need to do anything and can devote a lot of time to trading. The trend of retail investors is becoming increasingly prominent; 3) In the era of anti globalization, the dominance of strongmen and geopolitical shocks will also amplify fluctuations. The adjustment caused by these macro problems may be experienced every year, but the reasons are different, the transmission mechanism is different, and the manifestation is also different. For us, it's about walking and watching. So it is taboo to see an eternal bull market as soon as it rises or a century's peak as soon as it falls. It is still necessary to judge which level of trend is large, medium, or small based on the industry fundamentals and macro situation from a practical perspective. 4. After coding, new scenarios have emerged for AI: So there must have been something here before: https://((((((x.com))))/qinba frank/status/2091420164109836623? S=46&t=k6rimWs Ebo2D2TXolYcM-A has been discussed in detail. Simply put: 1) Enterprise AI deployment, as well as the data, security, and operational tool requirements generated around deployment; 2) Personal super agents provide greater opportunities for consumer entry restructuring. Take a look at the grok bots, meta muses, and GPT dots that have emerged in the past month; 3) AI drugs and scientific discoveries may lead to a longer lasting reassessment and greater differentiation between companies. 5. But the most crucial thing is that the main sectors of each market trend will undergo migration Last week here https://(((((((x.com)))))/qinbafrank/status/2110238858587004953? S=46&t=k6rimWSEbo2D2TXolYcM-A has detailed discussions: AI is not just about semiconductors, and semiconductors are not just about storage. AI has been iterating and evolving, and each wave of market trends may lead to different sectors: 1) There is no doubt that the storage sector is leading in the first half of the year; 2) But from August to last week, the sectors led by Lumentum were undoubtedly optical interconnect and software. Lumentum also set a new high at the beginning of the month, and the core logic has been discussed before: https://(((((((((x.com)))))/qinbufank/status/2086749504129372432? s=46&t=k6rimWsEbo2D2tXolYcM-A 3) Of course, there are also big technologies and software, with Microsoft and Meta leading the way. Microsoft and Meta logic were previously discussed here https://((((((((((((x.com))))))))))))/qinbafrank/status/2103632905499259058? S=46&t=k6rimWs Ebo2D2TXolYcM-A and https://(((((((((x.com))))))/qinbufark/status/2102058459072197089? We have talked about s=46&t=k6rimWs Ebo2D2TXolYcM-A. The earliest mention of software was in the "August is the best summer tweet" at the end of July: non generic software and SaaS will become the business execution layer of AI. There should be new sectors emerging in the future: Agent、 End side AI, intelligent hardware (AI phones, AI glasses, AI watches, etc.), AI scientific discoveries, and so on may all be the main sectors of the market in the future Thinking about the months of August and September, companies such as Moonlight, Software, Security, Big Tech (Microsoft, Meta), and AIDD were all very strong, while the latter four were completely absent from the market in the first half of the year. This already indicates that funds are starting to shift from the upstream hardware to the middle cloud, software, and even to the downstream application level. The inspiration we receive from this is that in order to invest in AI trends, we actually need to be able to see the changes and migrations in some main sectors. My personal intuition is that as the market trend progresses, the main focus will shift towards the deployment layer, small models, AI vertical OS, and application layer. This is inevitable: On the one hand, the market needs to see more B-end deployments and C-end application outbreaks; On the other hand, after the commercialization of the model, downstream applications will inevitably emerge more and more. Of course, in addition to the AI field, you can also look at sectors that bet on a decline in returns: https://((((((x.com))))/qinbank/status/2109923970518835301? s=46&t=k6rimWsEbo2D2tXolYcM-A Returning to the current core, it is still the financial reports of big technology, this week's Meta, SpaceX, and next week's Microsoft, Google, and Amazon that will provide us with evidence of AI commercialization and the continuous acceleration of token consumption. Please refer to the cited tweet for details This article is sponsored by @ bitget_zh, "Bitget Buying US Stocks: Instant Entry, Smooth Trading