Where to Begin

As the AI technology sector undergoes a significant correction, the strategy team at a major investment firm has proposed two battle-tested "golden rules" to help investors navigate the rhythm of the industry wave.

With the disclosure of fund quarterly reports for the second quarter, institutional concentration in AI tech has hit a historic high—TMT holdings exceed 60%, while pan-AI tech is nearly 70%, significantly surpassing the concentration levels seen in the historical "Mao Index" and "Ning Portfolio." Combined with the recent sharp decline in the AI tech sector, discussions on the "collapse of AI tech concentration" are heating up.

In a report released on July 23, analysts stated that the current adjustment in AI tech is more about profit-taking after an overheated phase. The first peak of the "M-top" that typically ends an industry wave "has likely not yet arrived." The report also pointed out that after this decline, the core pricing logic of AI tech will gradually shift from "price increase signals" to "volume increase signals," with the possibility of a shift within the core tracks.

The First M-Top Has Not Been Confirmed; None of the Three Criteria Have Been Triggered

The strategy team previously proposed a systematic framework for identifying the "selling point" of tech stocks in a series of reports, setting three conditions for confirming the first peak of the M-top. First, the trading qualitative signal: the frenzy phase of the first top is characterized by leading stocks driving secondary and tertiary stocks to peak, with leading stocks outperforming second-tier ones. However, in this round, the outperformance of second and third-tier AI tech stocks in the second quarter far exceeded that of leaders, making this signal unclear. Second, the trading quantitative signal: the high-to-low switching index needs to fall from levels well above 60%—historically, when the "Mao Index" and "Ning Portfolio" truly collapsed, this index was above 90%. As of July 17, the index had accelerated its decline to 33%, approaching the bottom of its historical range, suggesting the current high-to-low rotation is nearing its end. Third, the industry logic signal: the peak quarterly profit growth rate of leading companies must be reached—market expectations suggest that the highest single-quarter profit growth for AI leaders has not yet arrived.

The report emphasizes that as long as the industry trend remains intact and there are no clear macro "black swan" events or a collapse in the competitive landscape, the market will return to the main industry trend after a short-term adjustment.

Golden Rule 1: The Four Stages of Tech Industry Investment; Shift to Supply-Demand Gaps by 2026

The first golden rule is a four-stage framework for tech growth industry investment: buy giants (when a blockbuster product appears) → infrastructure (giants begin massive capital expenditure) → key links in the industry chain (the chain forms, completing the 0-to-1 process) → supply-demand gaps (the 1-to-100 process).

The report uses the new energy vehicle industry wave as an example. The giant stage corresponded to the Tesla Model 3 becoming a blockbuster. The infrastructure stage involved charging piles and grid equipment (priced in from 2019). The key chain link stage was lithium batteries (priced in from 2020). The supply-demand gap stage saw price surges in upstream lithium resources and volume increases in downstream auto parts in 2021.

Applying this to the current AI wave, the giant stage corresponds to the emergence of ChatGPT in 2023, benefiting companies like Nvidia and Microsoft. The infrastructure stage involves computing power and optical modules in 2024. The key chain link stage includes AI chips in the second half of 2024. The supply-demand gap stage covers upstream gaps in storage, electricity, and copper, as well as downstream gaps in multimodal models, autonomous driving, and AI large-model software.

The report concludes that by 2026, AI tech investment should migrate to the fourth stage of supply-demand gaps, using the analogy of "guarding the big light and buying the light ring"—similar to holding CATL in 2021 while increasing positions in lithium mines and auto parts, now shifting to second and third-tier companies around AI supply-demand gaps. It also notes that the surge in storage prices from the second half of 2025 to now is highly similar to the lithium price surge in 2021.

Golden Rule 2: The "Big-Small-Big-Small" Pattern; Focus on Portfolio Rotation Between M-Tops

The second golden rule is based on the "big-small-big-small (strong alpha)" pattern across three stages: trend, concentration, and frenzy. The trend phase involves trading large leaders. The concentration phase spreads to small and medium companies. The final frenzy phase sees core leaders peak. After the first M-top, the focus shifts to second and third-tier strong alpha stocks.

The report supports this with three historical cases.

For the consumption upgrade wave (2019-2021), Moutai and Wuliangye led the trend phase. Concentration spread across the entire baijiu industry chain. The frenzy phase began after December 2020, with leader Moutai driving second and third-tier stocks to peak. Between the first and second M-tops, second and third-tier stocks like Jiugui Liquor and Shede Spirits outperformed Moutai.

For the new energy wave (2020-2022), the trend started with CATL in 2020. Concentration spread to the lithium battery chain and auto parts. The frenzy phase began in October 2021, with CATL driving second and third-tier stocks to peak. Between the first and second M-tops, second and third-tier stocks like Defang Nano and Tianqi Lithium, which saw both volume and price increases, outperformed CATL.

For the mobile internet wave (2013-2015), Apple led the trend phase. Concentration spread to the Apple supply chain and ChiNext. After March 2015, East Money and Hundsun Technologies drove second and third-tier stocks to peak. Between the first and second M-tops in June, second and third-tier stocks like Shunwang Tech and Wangsu Tech outperformed core leaders.

All three historical cases consistently show that the investment focus after the first M-top is not on holding leaders but on rotating into second and third-tier stocks with rising volumes and prices.

Second Quarter Report: Institutional Concentration on AI Hardware; "Add Tech, Reduce Cyclicals" Accelerates

The second-quarter report data reveals the current institutional allocation pattern. From a secondary industry perspective, active funds significantly increased positions in semiconductors, components, and communication equipment compared to the first quarter, while clearly reducing positions in industrial metals, chemical pharmaceuticals, agrochemicals, and auto parts. The shift from traditional manufacturing to AI tech is accelerating.

At the primary industry level, the electronics sector's overweight allocation exceeds 20%, standing out. Electronics and communications occupy an absolute leading position in both overweight and increased holdings, forming the core "twin stars" of public fund positions. Historical percentiles show that electronics, communications, machinery, and building materials are at historically high levels.

Within AI tech, institutions are concentrated on increasing positions in AI hardware, integrated circuits, and PCB tracks. At the three-level industry granularity, the most significant increases are in semiconductor equipment and network connections and towers. At the individual stock level, actively increased positions are mainly in optical modules, semiconductors, and electronics. Stocks like New Easycom, Cambricon, and SMIC are at the top of the list for increased positions, consistent with the industry-level direction, confirming that AI computing hardware is the strongest consensus among institutions.

In the resource sector, non-ferrous metals, petrochemicals, and basic chemicals have all been reduced. Within the overseas supply chain, white goods, agrochemicals, and electrical equipment have been reduced. Institutional preference for the overseas supply chain is converging towards the AI industry chain.

The report also notes that, based on fund types, partial equity and flexible allocation funds increased positions overall in the second quarter, while ordinary stock funds remained flat. The current incremental allocation drive mainly comes from flexible products adding to the tech sector. Notably, historical experience shows that the position level at the first top is not the highest; the highest position level corresponds to the second top. This implies that before reaching the second M-top, AI concentration may still strengthen structurally.