Author: Think AI, Aaron
This round of artificial intelligence stocks has been rising for three and a half years, with no signs of stopping yet;
Those who started predicting a US stock market crash and an AI bubble last year have fallen into deep thought.
The Korean index has triggered 19 circuit breakers this year, with the benchmark index surging 4x since last year, and SK Hynix up 260% year-to-date, constantly hitting new highs or on track to do so;
US memory giant Micron has broken through a trillion-dollar market cap, ranking among the top 10 US companies by market value, with a year-to-date gain of over 200% and some stocks rising more than 10x in a year;
In Japan, SoftBank, which previously invested in Alibaba, has seen its stock price soar due to heavy bets on AI, making it Japan’s most valuable company by market cap.
Domestically, Yushu and Changxin have passed IPO reviews, with Changxin’s market cap entering at 1.5-2 trillion yuan, directly ranking among the top 5 in A-shares;
AI companies like Nvidia and Google are always under the fear of rising prices; there have even been reports of a housing price recovery in Shenzhen, where executives from semiconductor and AI companies, boosted by better-than-expected earnings, have started bulk purchasing properties.
But there is another picture: those who bought gold with loans at the beginning of the year are still stuck, domestic consumer stocks continue to decline, and many netizens say they have completely missed this AI bull market, feeling anxious every day—what should they do now?
The minority always makes money
Let’s look at the data first: In 2025, only about 18.9% of A-share retail investors were profitable, while 81.1% suffered losses, with an average loss of approximately 21,000 yuan per person. For small retail investors with positions under 100,000 yuan, the loss rate approached nearly 98.7%.
This occurred even as the A-share benchmark index rose nearly 20% in 2025, the Sci-Tech 50 index surged up to 80% within the year, and overall markets were up last year. Wind data shows that 500 stocks doubled in 2025, with over 120 stocks gaining more than 3x.
When the market corrected from January to April 2026, faith in AI collapsed, and institutions cut their positions at the bottom.
During the downturn, there was a wave of AI stock sell-offs by major shareholders: Zhongji Innolight (optical module) saw its controlling shareholder and executives reduce holdings by 4.914 billion yuan, after which the stock rebounded 35%; Kunlun Tech’s major shareholder Li Qiong planned to reduce 35.86 million shares, causing a 20% drop after the announcement, but the stock rallied 40% within a month—some reducers “sold at the bottom.”
A special report by Securities Times on AI sector share reductions noted that in the 2026 AI computing power sector, premature reductions by institutions led to potential profit losses exceeding 200 billion yuan (calculated based on average post-reduction price gains).
Meanwhile, many institutions that held onto AI stocks firmly bet on the wrong direction.
At the end of 2025, institutions bet on AI vertical applications like AI education and AI healthcare, but these failed to take off. In 2026, these sectors averaged losses of over 20%, while the computing power sector gained 50%+.
Abroad, missing the rally or selling too early is rampant.
Global renowned fund Bridgewater Associates significantly reduced its holdings of Nvidia (by nearly two-thirds), Alphabet (by more than half), Amazon (9.6%), and Microsoft (35%) in Q3 2025. Subsequently, from Q4 2025 to Q1 2026, these stocks averaged gains of over 80%.
There are even plenty of short sellers. Short positions in the US stock market recently hit a new high since 2012.
Data shows that hedge fund total leverage has risen to about 293%, and both the short exposure and Days-to-Cover metrics for the S&P 500 index have set records.
Even Warren Buffett started reducing positions early, with his cash reserves reaching a historic high of $397.38 billion in Q1 2026, completely missing the AI rally and the surge in US tech stocks.
Multiple waves rising simultaneously
But undeniably, AI remains the most certain and revolutionary opportunity in the current market.
AI is not a short-term concept, but a foundational infrastructure revolution like electricity and the internet. Historical experience shows that many who missed the first wave of tech stocks found opportunities in the application layer or the next round of infrastructure upgrades.
China also has unique space for proprietary AI research, application deployment, and strengthening the industrial chain.
When we feel confused and anxious, it’s worth seeing how the best in the industry view things—this may offer us guidance and insight.
Ma Huateng of Tencent, long considered lagging in the AI era, has been catching up rapidly since last year. His speech at the shareholders’ meeting in May sparked heated discussion.
Regarding AI, he said: “A year ago, we thought we were on the boat, but later realized it was leaking. Now we feel we’ve boarded, but can’t sit still—still hoping the boat will speed up a bit.”
On “what to do next,” he pointed out: “You can’t just blindly cross over because others are doing it there and try to take their turf. We’ve done that in the past, but basically failed.”
Indeed, the AI era offers vast opportunities. Combining one’s own strengths and characteristics to find the most suitable opportunity is the best choice.
Now stepping back from the frontlines, Ma Yun, who had deep insights into AI ten years ago, says AI is a once-in-several-centuries industrial revolution opportunity comparable to the invention of electricity. The AI era has just begun, and now is the time to position oneself; there is no such thing as having missed the boat.
From computing power infrastructure and large model platforms to industry AI applications, the entire value chain continues to release long-term investment opportunities.
From this perspective, the entire chain is full of limitless opportunities. Whether in work, investment, or entrepreneurship, as long as you delve deep into the industrial chain and stay immersed in the AI wave, combined with Duan Yongping’s long-termism and prudence principles, you will eventually reap rewards.
Now, large models are smart enough, but only Claude has achieved profitability in the coding agent track, and AI+ is still in progress.
The next structural opportunity will surely be reserved for those who are well-prepared and maintain a stable mindset.
