When will this AI cycle peak? Cases of six high-prosperity industries over the past two decades show: stock price peaks lead fundamental peaks by about 1-1.5 years, and high-prosperity industries often exhibit an "M-top" pattern.

Core Viewpoints

Core Conclusion: ① Experience from six high-growth industries over the past two decades shows that stock price peaks can occur early during the penetration rate ramp-up and high-growth phase, typically leading the fundamental peak by 1–1.5 years. ② High-growth industries often exhibit an "M-top" pattern, where the second peak can reach 80%–90% of the absolute peak. During this process, the market becomes less responsive to positive news, and divergence on the growth outlook increases. ③ For the current AI rally, the key variable for confirming a top is the capex of tech giants. While aggregate expectations have not yet turned, the weakening second derivative of some companies' capital expenditure already warrants caution.

Industries may see 2-3 year interim price peaks even during rapid growth phases. Since June, previously hot sectors like telecom and electronics have experienced corrections of varying degrees, reigniting investor debate about sector tops. In fact, AI-related sectors—especially upstream hardware—have rallied for an extended period, yet long-term earnings expectations remain high. The core question is: Will stock prices in high-growth industries keep rising until earnings explicitly turn down? Historical examples show that even when an industry is in its early rapid development stage, with penetration rates still rising over the next 2-3 years, stock prices can peak early. For instance, in smartphones and new energy vehicles, the consumer electronics sector reached an interim peak in late 2010 when domestic smartphone penetration was about 25%. Penetration continued to rise to 47% by end of 2012, but the stock index had already fallen approximately 46% from its high.

Six high-growth industry cases over the past two decades: stock price peaks lead fundamental peaks by about 1–1.5 years. Reviewing six high-growth industries since 2009—consumer electronics, communication equipment, semiconductors, photovoltaic equipment, batteries, and energy metals—their absolute stock price peaks led their ROE peaks by 8 to 19 months, with a median of about 12.6 months and an average of about 13.1 months. In other words, "earnings are still being realized" does not mean "the trend hasn't topped." By the time fundamentals officially confirm a peak, stock prices have often already undergone a major correction. Stock prices essentially reflect changes in expectations, not current fundamentals. Therefore, during a phase when growth is still strong, profits are still rising, and orders are still being placed, if forward expectations move from consensus to divergence, stock prices can preemptively reflect a future fundamental downturn.

High-growth industries often form an "M-top," and diminishing stable excess returns from positive earnings news may be an early top signal. Historical reviews show that after the absolute peak and the first decline, because good earnings realization, order inflows, and profit growth are still visible, the sector tends to rebound on the back of earnings confirmations, policy catalysts, or industry triggers, forming a second peak lower than the absolute high. Meanwhile, during the peak phase, positive news does not disappear, but it struggles to generate broad excess returns. The median 20-day excess return after positive earnings announcements during the main uptrend is 3.2%, but it drops to -1.5% in the top window and further to -2.7% in the second-peak stage. Market concerns shift from "whether valuation is expensive" to "whether growth can continue"—an early top signal.

The core observation variable for the top of this AI computing rally: tech giants' capex. Current AI hardware demand is primarily driven by the capital expenditure of overseas cloud hyperscalers. For sectors like optical modules, PCBs, and servers, downstream cloud capex is the aggregate constraint on whether orders can continue to expand, and it is the most transparent publicly available forward earnings tracking indicator. Currently, the 2026 capex guidance of the four hyperscalers still points to high expansion, providing strong medium-term visibility on orders and revenue along the supply chain. No capex downturn signal sufficient to confirm an absolute peak in the computing chain has yet been seen. However, Microsoft's and Meta's Q1 capex second derivative has turned negative, indicating that the expansion slope for some core cloud vendors is beginning to slow. If Amazon and Alphabet also show similar declines, or if future capex guidance fails to sustain high base effects leading to marginal growth slowdown, then the risk of an interim stock price peak warrants attention.

Full Text

Since June, previously hot sectors like telecom and electronics have experienced corrections of varying degrees, reigniting investor debate about sector tops. In fact, the objective contradiction for these sectors is that while AI-related stocks—especially upstream hardware—have rallied for an extended period, their long-term earnings expectations remain high. Will stock prices in such high-growth industries keep rising until earnings explicitly turn down? What basis can be used to judge the top of the trend? This article analyzes these questions. Insights from twenty years of six high-growth sectors: When will this AI rally top?

Previously, in our report "What is a Reliable Indicator of Overheating in a Sector? – Thoughts on the Current AI Rally (20260613)," we analyzed that the hard-tech sectors with high growth in A-shares are already in an overheated range with crowded trading. We pointed out that while trading overheating does not necessarily signal an imminent price top, it often points to increased volatility, reduced odds, and rising tail risks. Since June, the high volatility in telecom and electronics sectors has been validating our judgment.

However, since trading overheating is not a confirmation signal of a top, what exactly marks the top of a mainstream sector? The main theme of a bull market often corresponds to high-growth industries with clear industrial trends and high earnings expectations. It is generally believed that the turning point for such sectors' stock prices will appear at the marginal turning point of fundamentals. We believe this judgment is not wrong but is not entirely accurate. Below, we will review the topping process of historical high-growth sectors.

Industries may see 2-3 year interim price peaks even during rapid growth phases. When studying industrial trends, penetration rate is an important metric to quantify the lifecycle position. According to the innovation diffusion theory, once the penetration rate exceeds 16%, it rises rapidly, corresponding to the steepest S-curve high-growth phase. However, if we compare penetration rates with stock prices, we find that even when the industry is in its early rapid development stage, with penetration still rising over the next 2-3 years, stock prices can peak early. Take smartphones and new energy vehicles as examples. In the smartphone cycle after 2009, the A-share consumer electronics index peaked in late 2010, when domestic smartphone penetration was about 25%. By end of 2012, penetration rose to 47%, but the stock index had already fallen about 46% from its high. Meanwhile, in the new energy cycle since 2020, the A-share new energy vehicle price peaked in November 2021, when the retail penetration rate of new energy passenger cars was about 21%. By end of 2023, penetration further rose to 40%, but the stock index fell by up to 60% during that period.

The cases of smartphones and new energy vehicles show that when penetration rises rapidly from low levels in early industry development, the market concentrates on pricing in industry space expansion, improved earnings for leaders, and valuation system reassessment. This phase offers the highest price elasticity but also tends to form interim peaks after excessive linear extrapolation of expectations.

Six high-growth industry cases over the past two decades: stock price peaks lead fundamental peaks by about 1–1.5 years. Similar to the penetration rate perspective, if we directly compare sector stock prices with fundamentals, we find that "earnings are still being realized" does not mean "the trend hasn't topped." Historically, high-growth industries could see their stock prices peak mid-cycle even when fundamental indicators are still rising. Since revenue or profit growth rates are often affected by base effects and fluctuate greatly, we adopt ROE (TTM) as a fundamental indicator. We reviewed the stock price performance of six high-growth industries—consumer electronics, communication equipment, semiconductors, photovoltaic equipment, batteries, and energy metals—during their boom cycles since 2009. The data shows that their absolute stock price peaks led their respective ROE peaks by 8 to 19 months, with a median of about 12.6 months and an average of about 13.1 months. In other words, by the time fundamentals officially confirm a peak, stock prices have often already undergone a major correction.

High-growth industries often exhibit an "M-top," with the second peak typically reaching 80%–90% of the absolute peak. In our review, we also found that after the absolute peak and the first decline, because good earnings realization, order inflows, and profit growth are still visible, the sector tends to rebound on the back of earnings confirmations, policy catalysts, or industry triggers, forming a second peak lower than the absolute high. In the six cases where price led ROE, the median ratio of second peak to absolute peak was 87.1%, with an average of 86.9%. For long-term capital, the absolute peak is usually hard to confirm in real time, but when prices first pull back significantly, fundamentals remain strong, and the narrative turns optimistic again, the second peak also has allocation significance. At that point, the risk-return profile has changed: upside is capped by the previous high, while downside depends on whether fundamentals are finally confirmed to peak. As the profit growth slope marginally weakens, the adjustment after the second peak tends to be more sustained.

An early top signal for high-growth industries is that positive earnings news no longer delivers stable excess returns. Why do stock prices in high-growth sectors peak ahead of fundamentals? One explanation is that stock prices are leading indicators of fundamentals, reflecting changes in expectations rather than current fundamentals. When the market's long-term expectations for sector growth shift from consensus to divergence, positive news on the current fundamental side finds it increasingly difficult to push prices higher. We analyzed the excess returns of individual stocks in the aforementioned high-growth sectors after positive earnings announcements (quarterly high growth or sustained upward earnings slope). During the main uptrend, the median 20-day excess return was 3.2%, but in the top window it fell to -1.5%, and in the second-peak window it dropped further to -2.7%. The proportion of stocks with positive excess returns also decreased significantly. If we calculate the market-cap-weighted average excess return, the phenomenon of fundamental good news being ignored at the top is more pronounced in large-cap stocks. The market-cap-weighted 20-day excess return during the main uptrend was 6.5%, but in the top window it fell to -3.6%, and in the second-peak window to -2.4%. This indicates that during the high-level phase, it is not that there is no positive news, but that positive news can no longer generate widespread excess returns. Market concerns gradually shift from "whether valuations are expensive" (as argued during crowded trading) to "whether growth can continue"—this is also an early top signal.

The core observation variable for the top of this AI computing rally: tech giants' capex. Current AI hardware demand is primarily driven by the capital expenditure of overseas cloud hyperscalers. For sectors like optical modules, PCBs, and servers, downstream cloud capex is the aggregate constraint on whether orders can continue to expand, and it is the most transparent publicly available forward earnings tracking indicator. As long as Amazon, Microsoft, Alphabet, and Meta continue to raise their capex, the forward earnings of the supply chain are likely supported. But once the capex growth slope turns, as mentioned earlier, stock prices typically preemptively reflect the future fundamental decline. Currently, the combined capex of the four hyperscalers has rebounded from the low in 2023, with year-over-year growth accelerating significantly since 2024. Full-year 2025 capex grew 64% YoY, Q1 2026 grew 80% YoY, and the 2026 guidance midpoint implies total capex of about $710 billion, up 88% YoY from 2025. This indicates that the 2026 capex guidance of the four hyperscalers still points to high expansion, providing strong medium-term visibility on supply chain orders and revenue. From a fundamental confirmation perspective, no capex downturn signal sufficient to declare an absolute peak in the computing chain has yet been seen.

However, at the individual company level, capex may already show a worrisome second-derivative change. In Q1 this year, Microsoft's and Meta's capex second derivative turned negative, indicating that the expansion slope of some core cloud vendors is beginning to slow. If Amazon and Alphabet also show similar slope declines, or if future capex guidance fails to sustain the high base leading to marginal growth slowdown, then the risk of an interim stock price top warrants attention. Based on the lead-lag relationship we reviewed earlier between stock price tops and fundamental tops, assuming AI capex growth sees a marginal turning point around 2028, this could correspond to a stock price top for AI hardware in the second half of 2026 to the first half of 2027. Conversely, if the combined capex of the four companies continues to be revised up, and the quarterly weakness at Microsoft and Meta does not spread, the trend is more likely to continue.

Source: Guosen Securities

Risk Disclaimers

Market has risks; investment needs caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Investing based on this content is at one's own risk.