The current round of AI-driven industry differentiation is nearing its extreme

On February 16, Huachuang Securities' Yao Pei released a report stating that the current round of AI-driven industry differentiation is approaching an extreme.

Since the 9/24 market, industry differentiation has intensified and is now close to an extreme. Since the 9/24 market, the resonance effect of policy-driven capital concentration and differences in industry fundamentals has significantly increased industry performance differentiation.

As of February 14, the top five industries with the highest rolling six-month gains are computer (79%), media (54%), comprehensive (48%), electronics (46%), and retail (43%), with an average of 54% for the top five industries;

The bottom five industries with the lowest gains are coal (-5%), utilities (-5%), oil and petrochemicals (1%), agriculture, forestry, animal husbandry, and fishery (2%), and food and beverage (7%), with an average of 0% for the bottom five industries.

The average gain of the top five industries minus the average of the bottom five industries has increased from 21.0% at the end of August 2024 to the current (February 14, 2025) 54.0%, placing it at the 90th percentile since 2010, indicating that industry differentiation is nearing its peak.

The main reason for the expansion of industry differentiation in this round is the rapid rise of the financial sector since the 9/24 market and the intensified differentiation between TMT and dividends caused by the AI wave since 2025, coupled with the downward pressure on resource prices.

Huachuang Securities believes that the peak signals are the phase of economic recovery being falsified + trading overheating + changes in industrial policy. Reviewing the four historical phases of industry differentiation expansion and convergence, the convergence of industry differentiation mainly occurs in the following three scenarios:

① Economic recovery expectations are falsified in the short term or significant changes occur in macro policies, such as the core asset market after the pandemic was lifted in 2023; ② Trading heat and valuations continuously rise to historical highs and cannot be realized in performance in the short term, such as TMT in 2014 and AI in 2023; ③ Changes in industrial policy or a downturn in industry prosperity, such as cyclical resource products in 2021.

Based on the current situation, the report believes that the three scenarios leading to the convergence of industry differentiation cannot be validated in the short term in this round of the AI market.

① Since 9/24, macro policies have undergone significant changes, and the continuous introduction and deepening of incremental policies make it impossible for economic recovery expectations to be falsified in the short term.

② In terms of trading heat and valuations, there is still room for the current DeepSeek market to reach its peak. As of February 14, the number of constituent stocks in the DeepSeek index has expanded to 231; the market capitalization of constituent stocks has risen to 6.7% of the total A-share market, with both the expansion speed and scale significantly stronger than the ChatGPT market in 2023. The rapid expansion of constituent stocks not only increases the carrying capacity of trading funds and the rotation of subfields but also accelerates the possibility of fundamental realization.

③ The policy orientation of "technological self-reliance and self-improvement" will not change in the long term. The current Sino-U.S. technological competition has entered a heated stage, and AI, as the core battleground of competition, is expected to continuously improve industry prosperity