What else can we talk about if we don't discuss AI?
Recently, a friend who has followed the "lying flat index" for many years came to visit me. He said, "Your official account has now turned into an AI account."
He didn't mean that there was anything wrong with running an AI account. It's just that a lot of content that used to be available here is gradually disappearing now.
From the very beginning, this account has been different from other self-media outlets. For most articles, the primary goal is to support my own investment research, and only then will I consider which topics can attract traffic.
This means that for a company without obvious expectation gaps, whose valuation given by the capital market is relatively reasonable, I can hardly judge whether it is overvalued or undervalued, so it is difficult for me to start writing. There are also some companies that are extremely popular but not listed, with no publicly available financial data, such as Dreame. Companies like this will not arouse my desire to write about them at all.
For a long time in the past, this topic selection logic has constantly led me to the AI track: the changes here are fast enough, the market divergence is large enough, and there are naturally many companies and issues worth writing about. Everyone thinks AI is good, and I think so too. But up to now, even I myself have begun to feel that this topic seems to have permeated every corner.
Recently, Unitree Robotics went public and held the bell-ringing ceremony. A photo of Wang Xingxing looking a little serious was taken, and this photo was singled out to make news and even became a trending search. Soon, many entertainment and marketing accounts began to repost it, adding many unique interpretations typical of the entertainment industry, speculating whether he was "unhappy".
This incident itself is absurd. Unitree's robots have appeared in the Spring Festival Gala for two consecutive years, but Wang Xingxing has not entered the entertainment industry because of that. The bell-ringing was originally just a capital market event, but in the end, only the founder's expression was repeatedly consumed. Does a founder have to keep smiling the whole time when his company goes public? AI is no longer just an industry or capital market topic. Even a company's listing can be spread as pan-entertainment news in a very short time.
What's more subtle is that AI has not only changed what I write, but also seems to have begun to change how I write.
Now every time I finish writing an article, I will subconsciously delete the phrase "not... but...". It's not because this sentence was actually written by AI, but because I can no longer look at this phrase with a normal state of mind.
So after much consideration, in this article, I want to greatly reduce the AI content. It's not to be a contrarian, let alone to be bearish on AI. I just want to temporarily shift the camera away, and write down the non-AI parts I saw and thought about in the past six months one by one.
Without talking about AI, of course we still have a lot of things to talk about.
01
Excessive Profit
I want to talk about consumption.
After returning to China after living in the United States for half a year, I really wanted to see what changes have taken place in China's consumption, but Beijing did not bring me any surprises.
The Freshippo store near my home closed, and no new supermarket has opened there; the restaurant that netizens in Jiangsu, Zhejiang and Shanghai recommended to me half a year ago has just opened in Beijing recently. This gap is even more obvious when I scroll through Douyin. New stores, new brands and new consumption scenarios are constantly emerging in Shanghai, Guangzhou and Chengdu, and similar things will not appear in Beijing until several months later. Beijing certainly has no shortage of consumption power, but it does seem to be half a beat slower in keeping up with the most popular consumption trends.
Even so, even if I didn't feel much newness after returning, I still think China's consumption sector has huge potential.
The recently released financial report of East Buy may explain why I still think so. In fiscal year 2026, East Buy's GMV reached 10.2 billion yuan, a year-on-year increase of 17.6%; after excluding the contribution of Hui Tong Hang, the GMV under the comparable caliber increased by 36.4%. Among them, the GMV of self-operated products reached 5.4 billion yuan, a year-on-year increase of 42.1%; the revenue was 5.701 billion yuan, a year-on-year increase of 29.8%. The net profit reached 544 million yuan, a year-on-year increase of 8684.8%.
Last December, after visiting many consumption scenarios in the United States, I wrote the article "The Next Decade-level Consumption Trend: China's Consumption Still Has Huge Potential" based on my real feelings. There is a judgment in that article, which is also the reason why I have always been optimistic about East Buy. I summarized it as "risk premium":
The cost consumers pay for a product includes not only the price of the product itself, but also the time cost invested in buying the right thing, and the risk cost they have to bear once they make a wrong choice. What East Buy does for consumers is to complete the screening, comparison and judgment, saving exactly the latter two parts of the cost. As long as this selection process is reliable enough, consumers are willing to pay a little more for this certainty.
East Buy's operating profit margin in this fiscal year is about 11.6%, and the net profit margin is about 9.5%; the adjusted net profit reached 629 million yuan, a year-on-year increase of 262.2%, and the adjusted net profit margin is about 11%. Such profit margins basically confirm the previous judgment. At least at this stage, the certainty provided by East Buy and the consumer trust it has established in advance are both scarce.
But when compared under as consistent calibers as possible, Costco's operating profit margin is about 3.8%, and the net profit margin is about 2.9%; the operating profit margin of Sam's Club's US business is about 2.6%, and the net profit margin is not disclosed separately. The business structure and revenue recognition methods of the three companies are not exactly the same, but at least in terms of magnitude, East Buy's current profit space is significantly thicker than that of Sam's Club, which it wants to benchmark against, and Costco, which has a similar model to Sam's Club.
Isn't a high profit margin a good thing? In my opinion, it's not good enough, and at the very least, it's unlikely to be sustainable. The work of selecting products for consumers can be done by many people in China. Even idol stars like Huang Zitao can create a sanitary napkin brand and sell it very well.
Consumers in no country get their money for free. Taking a longer-term perspective, East Buy's high profit margin under the current consumption environment will attract more and more companies to replicate its model, and the certainty itself will gradually no longer be scarce. At that time, competition will evolve from who can select good products for consumers to who can sell good products at lower prices.
The low price mentioned here is not the low price achieved through vicious subsidies, but the low price brought by the enterprise that improves the efficiency of the entire supply chain to a sufficiently high level and then gives this efficiency dividend back to consumers. The most solid competitive moat of such enterprises will eventually fall on a relatively low profit margin.
This year's World Cup made many Chinese people more intuitively realize how willing and capable Americans are to spend money. But even in such a high-consumption market, the key reason why Sam's Club and Costco can gain long-term recognition from consumers and large capital is that they can make consumers believe that they can not only spend their money correctly, but also get more value for their money here.
From the end of fiscal year 2020 to the end of fiscal year 2025, Costco's market value has increased by about 172% cumulatively, while its net profit has increased by about 102% in the same period. Low profit margins and relatively slower profit growth have not prevented this company from obtaining higher market value. When an enterprise continuously transfers scale and efficiency benefits to consumers, low profit margin itself can become a very deep moat.
Therefore, the surge in East Buy's performance this time proves that its model of saving consumers' selection costs is indeed valid. The next more important question is whether it can continue to translate its established product selection capabilities, trust and scale into lower prices. As long as it can do this, I believe East Buy will gain more recognition from large capital.
02
Monotony
I also want to talk about cultural entertainment, the mass pop culture industry. Up to August this year, there are not many works that impressed me enough to be worth writing about.
There are no particularly eye-catching games. Perfect World's Where Winds Meet had a stunning opening, and few game companies can truly build a complete and vivid city. Hytale has a strong Japanese-style two-dimensional flavor, and I originally expected it to open up the city little by little, and then create some completely different city styles.
But when it came to version 1.2, the update went back to the fantasy wilderness path, and this direction really didn't attract me much. Its commercial performance is not bad. As of August 18, the total global revenue has exceeded 2 billion yuan, and the mobile game version has repeatedly ranked in the top five of the iOS game bestseller list in major markets at home and abroad. I will continue to follow up on its performance later.
In addition, the nostalgic server of Shengqu Games' MapleStory under Century Huatong is also worth talking about. In the long run, the competition for nostalgic servers ultimately lies in content reserves and update capabilities. The reason why the World of Warcraft nostalgic server was successful is that from the classic version 60 to Mists of Pandaria, there are originally five versions that can be released segment by segment. That is the heritage accumulated by the original 8 years of content, plus a large enough user base, which creates the space for long-term commercial operation.
MapleStory is of course very popular now. A few minutes after the server was launched, all four pre-registration servers were fully occupied, and the long queue situation also became a viral topic. But from August 3 to August 21, Century Huatong's closing price stayed at 13.82 yuan exactly. This boom has not brought a sustained valuation revaluation at least. For me, what is worth observing more is whether this popularity can eventually be converted into long-term revenue.
Kuaishou's self-developed game Lord of Mysteries has just entered public beta, and I haven't played much of it; but I want to follow this topic to talk about China Literature.
Qidian recently strengthened its governance of "non-human automated creation". Works that use AI or other automated means to replace humans to complete core content will be suspended from recommendation, removed from the charts, or even blocked and taken off the shelves depending on the degree of violation. A recent batch of processed works have been removed from the monthly ticket list and bestseller list, and lost the official key recommendation. For online novel authors, this will significantly reduce the opportunities for their works to attract new readers, and their subsequent income will also be directly affected.
What I can't figure out is, in the face of Fanqie's free content offensive, does AI-assisted creation really put Qidian in a disadvantageous competitive position? For readers, is it that they don't recognize AI-assisted online novels, or that they just think the novel itself is really bad? Besides, the platform has not yet made public a clear standard for what degree of AI intervention counts as replacing humans to complete core content. In this case, is it necessary to kick these works out of the core charts first?
Online novels often have 2 to 3 million words, or even 5 to 6 million words. It is inherently very difficult for one person to maintain the characters, plots and foreshadowing for a long time. Sometimes I think, if there was AI assistance when Lord of Mysteries was being written, the plot after Klein was promoted to Sequence 3 Ancient Scholar might not have collapsed so much. Then thinking about the early passing of July Xinfan, the author of Qin Official, and the unfinished Jade Dynasty 2, I hope even more that AI can help excellent authors share some mechanical and energy-consuming work, so that they can worry less, write longer, and live more easily.
What's more, in the field of literary and artistic creation outside online novels, from comics, short dramas, videos and TV series to movies and games, almost all categories are thinking about how to better integrate with AI. At this time, China Literature is instead tightening the space for AI creation. Is it going against the historical trend, or reserving the last piece of private land for the "human touch" in creation?
Finally, let's talk about Pop Mart. Its latest financial report is really not good-looking. In the first half of 2026, the company's revenue increased by 23.8% year on year, and the net profit attributable to owners increased by 10.1% year on year. Both figures were lower than market expectations. Wang Ning also said that 2026 is a "adjustment year", and the full-year revenue growth target of 20% set at the beginning of the year "may not be achieved".
At the same time, the performance of LABUBU is also weakening. In the first half of 2026, the revenue of THE MONSTERS decreased by 7.5% year on year, and the total overseas revenue decreased by 11.1%. You can say this financial report is unsatisfactory, and you can also say that LABUBU is getting further and further away from its peak of popularity. Overall, these unfavorable factors do constitute the reason for the market to be temporarily bearish on Pop Mart and re-evaluate its valuation.
But the same financial report has another side. The revenue of Star Doll increased by 580.6% year on year, and the company already has 6 IPs whose half-year revenue exceeds 1 billion yuan. There is certainly no answer now as to whether Star Doll can replicate the global popularity of LABUBU, but it at least shows one thing: consumers' attention has not left Pop Mart along with LABUBU.
It is no longer difficult to be bearish on LABUBU. What is difficult is to prove that Pop Mart will never be able to produce the next LABUBU again. The ebb of a single character is the normal state that IP companies that use dolls as carriers and characters as core assets have to face sooner or later. Its long-term value depends on its ability to continuously discover new characters, turn them into products, and then amplify their influence through channels.
Therefore, the market can take back the valuation premium that LABUBU brought to Pop Mart, but it cannot assert that the next hit will definitely not come out of here.
03
Patience
I originally wanted to start writing from the photovoltaic industry.
I used to be a practitioner in the energy industry, and I have always had affection for the photovoltaic, wind power and oil and gas industries. In recent years, almost every mid-year and end of year, I will look back to see where this industry has gone. But in the past six months, I wrote the names of Longi and Tongwei into the document several times, thought over and over, and finally closed the document.
Perhaps it is precisely because I am familiar with it that it is harder to write. The current problems in the photovoltaic industry are of course the result of the choices made by all industry participants in the past few years. But now, a company can cut production, reduce capital expenditure, and adjust its technical route. In the face of the continued underperforming performance of Longi and Tongwei and their long-term sluggish market performance, it is hard for me to stand on the sidelines and criticize them.
What's more, I can't come up with any meaningful solutions. Telling them not to engage in cut-throat competition, not to expand production, and to sell products at higher prices is something anyone can say; I even saw some people advising enterprises not to be "traitors" and not to sell equipment to the United States and India, which is completely standing and talking without backache. If an article ends up with nothing but these correct but useless words, it will more or less show the arrogance of someone who has never experienced the hardships of the real world.
The problems currently encountered by this industry can no longer be solved by the power of the industry's enterprises themselves.
After putting aside the photovoltaic topic, I thought of another topic that I have long wanted to write but have been putting off for a long time — gold.
They seem to have nothing to do with each other. Photovoltaics makes me face a sense of powerlessness again and again: the problem is clearly in front of me, but I can't come up with an answer that I believe in. When it comes to gold, the difficulty is different. The logic supporting it has always been there, but it is far more difficult than imagined to hold gold stocks firmly. After all the twists and turns, what remains is still two words: patience.
Whether in published articles or in other semi-public occasions, I have never hidden my optimism about gold, and my attitude has always