How do you view the impact of trading crowding on the AI sector?

```

There is an overestimation of market concerns about crowded trading in the AI sector; multiple indicators show the market has not yet reached historical warning levels.

A widely circulated chart in the market recently sparked broad discussion. The concentration of trading volume in the top 5% A-share stocks has risen to 43.9% (20-day moving average, as of April 20), approaching the historical warning line of 45%, causing some investors to have doubts about the sustainability of the AI sector’s group rally.

GF Securities Strategy Team released a special report on April 21, pointing out that judging the end of the group rally based solely on this single indicator is "too arbitrary", and historically, peaks in trading concentration are not synchronized with market tops.

More crucially, growth is the core logic. Against the backdrop of overall tepid demand, AI-related companies represent scarce high-prosperity assets in the A-share market in 2026, and their rise in market value and trading share is fundamentally supported.

Current crowding indicators have not reached extreme levels

Although the trading concentration of the top 5% stocks is at the 91.8th percentile in history since 2003, it is still some distance from the historical highs in February 2021, February 2018 and January 2015.

Using different measurements, the signals are milder: the top 1% stocks’ trading concentration is only at the 83.7th percentile, the top 3% at 89.1%, and the top 10% at 94.2%, all still far from warning levels.

Trading concentration only depicts one dimension of market transactions. Short-term sentiment indicators constructed from the difference between new highs and new lows over 20 days, the proportion of stocks above their moving average, and the probability of gains, have also not indicated overheating after rebounding from the bottom since March.

Historical top signals are not synchronized with peaks in trading concentration

Historically, the warning effectiveness of trading concentration for market tops is limited.

The report reviews eight top signals since 2003: two corresponded to market bottoms (2003, 2008), four occurred mid-bull market (2006, 2015, 2025), and only three effectively warned of subsequent price declines (2007, 2018, 2021).

Even for these three effective warnings, growth slowdown was the more essential reason.

The 2007 subprime crisis impacted fundamentals, in 2015 numerous acquisitions couldn’t deliver promised performance and goodwill impairment risks had long been present, in 2021, growth decelerated for Moutai and other liquor stocks, and household appliances faced cost and real estate pressures. All three market tops were accompanied by declining growth in non-financial return-on-equity for A-shares.

Industry waves change the historical reference for trading concentration

Using the upper limit of A-share historical indicators from the past 20 years as thresholds has little comparative meaning in the current technological revolution.

Referencing U.S. stock experience, each wave of technology and industry innovation systematically boosts the market share and trading volume of hardware and software service sectors.

The same is true for A-shares. Since 2010, the weights of TMT and new energy sectors in circulating market value have steadily increased. Last year, TMT positions in funds surpassed 40%, and electronics positions exceeded 20%, both already above historical norms for A-shares.

Traditional trading concentration thresholds, established during periods dominated by traditional industries, should not be used to assess market structure in the era of technological revolution.

Scarcity of high-prosperity assets supports AI sector group rally

In the profit structure of A-shares in 2026, high-prosperity assets are scarce.

Data shows that current broad-based demand is still rather subdued and the proportion of high-growth listed companies in A-shares is relatively low. Meanwhile, AI progress in 2026 has accelerated, prompting global markets to raise EPS forecasts for 2026. The quantity of tokens generated by the OpenRouter platform has continued to accelerate since February this year, confirming substantial expansion of demand in AI application layers.

Against this backdrop, the rising share of market value and trading volume for high-growth tech companies should be seen as a "natural" result rather than pure sentiment-driven speculation.

Looking ahead, there are two paths to break the scarcity of high-prosperity assets: one is broader economic recovery leading to profit improvement across more industries; the other is an AI bubble burst combined with U.S. economic recession.

Currently, both scenarios are unlikely, so trading concentration in A-shares may remain high, and current crowding is not sufficient reason for an exit.

Risk warnings and disclaimerThe market has risks, investment requires caution. This article does not constitute personal investment advice and does not take into account individual users' specific investment goals, financial situation or needs. Users should consider whether any opinions, views or conclusions in this article are appropriate for their particular circumstances. Investing based on this is at your own risk. ```