Shenwan Hongyuan Group: The next rally may still be led by AI industry trends; await the formation of a "second-stage rally"

Shenwan Hongyuan Group released a research report stating that in 2026, A-shares will still operate within the framework of the "two-stage major band rally theory," except that the first-stage rally in 26H1 will be of a higher level, and the correction phase in 26H2 will correspondingly be of a larger scale; however, expectations for a "second-stage rally" should not be abandoned, as it is expected to be completed in 27H2, with investment returns potentially higher than in 2026. It is an objective judgment that there is still room for AI application to open up new tracks, and it is also an objective trend that the AI industry continues to advance; the next major band rally will most likely still be led by AI industry trends. In the long term, A-shares are in a "2018-level correction phase between the two stages of the rally." In the medium term, the one-sided correction absolutely dominated by defense is already in its latter stage, possibly followed by a consolidation phase with active bottom themes, which is an inevitable process in building the "second-stage rally." In the short term, momentum for an oversold rebound is accumulating, the balance of capital supply and demand is being rebuilt, and the spring rally in 2027 around the turn of the year is worth anticipating; an earlier-than-expected oversold rebound is also reasonable. Shenwan Hongyuan Group's main views are as follows: At the current stage, problems at the levels of the asset allocation environment, core industry trends, and stock market liquidity objectively exist, and there are no specific clues for a turnaround yet; only the improvement in short-, medium-, and long-term cost-effectiveness has a clear timetable. At this point, extrapolating caution is easy, while discussing a reversal is difficult. Has the 2026 A-share market broken through historical patterns? The firm believes the market may still be operating within the framework of the "two-stage major band rally theory." It is just that the "first-stage rally" in 26H1 is of a higher level, and the "correction phase between the two stages of the rally" in 26H2 is correspondingly larger in scale. But amid the headwinds, expectations for the "second-stage rally" should not seem to be abandoned. It is an objective judgment that there is still room for AI applications to open up new tracks, and it is also an objective trend that the AI industry continues to advance. When the pricing to correct AI computing power inflation becomes sufficiently full, the outlook for AI industry trends should also gradually shift from guarding against negative factors to anticipating positive factors. Objectively viewing the current problems while thinking about the turnaround in 2027 and seeking breakthroughs at both the aggregate and structural levels may be the questions that need to be answered in the 2027 annual strategy. First, discuss what key issues the correction that began in mid-2026 reflects, and the medium-term turnarounds for these issues. 1. The 2026 rally and correction bands are symmetric in capital supply and demand characteristics: in the rally band, extreme divergence under stock-game conditions, with track-focused funds' positive feedback achieving disproportionate effects with little effort; in the correction band, all types of capital generally reduce positions amid widespread difficulties, and concerns about position reductions by allocation-type funds and absolute-return funds drive negative capital feedback loops. At the current stage, all types of capital have generally reduced positions, because net asset values have fallen back near break-even levels, increasing the proportion of allocation-type funds that concentrate on reducing positions. Conditions for restoring the balance of capital supply and demand in the short to medium term have improved. However, track-focused funds and allocation-type funds still hold high positions in the computing power chain, and long-term clearing has not yet occurred. The trend of ShenZhen New Industries Biomedical Engineering, a certain degree of position adjustment and switching, and policy and institutional adjustments are the conditions for long-term clearing. The 2026 rally and correction bands are symmetric in both fundamental and capital-market logic. In the rally band, on the fundamental side, the logic of computing power inflation continued to spread; on the capital side, track-focused funds' positive feedback achieved disproportionate effects with little effort. 25Q3 saw historically significant incremental game-playing, while 26H1 featured extreme divergence under stock-game conditions. At the same time, there was more offensive capital and less allocation-type capital. The positive feedback of public track-focused funds was the marginal pricing capital, and within the public fund space there was also stock-game competition, with capital flowing out of non-tech tracks and concentrating into high-sharpness tech track-focused funds. In the decline band, the weakening of the computing power inflation logic objectively existed; on the capital side, all types of capital generally reduced positions amid widespread difficulties. Position reductions by allocation-type funds and absolute-return funds constituted marginal capital, driving negative capital feedback loops. Capital that might be forced to reduce positions near break-even levels (corporate annuities, fixed income+, bank wealth management, etc.) concentrated trigger position reductions in October 2026, while other types of capital (margin financing, private funds, active public funds, etc.) simultaneously reduced positions. After all types of capital generally reduced positions, combined with efforts to stabilize the capital market policy, conditions for restoring the balance of capital supply and demand in the short to medium term improved. However, from a long-term perspective, the proportion of capital that has achieved clearing of high-tech positions is still not high. Margin financing and private funds have reduced positions relatively sufficiently; comprehensive position adjustment and switching by public track-focused funds typically waits until a new offensive structure emerges; corporate annuities and fixed income+ have marginally reduced tech positions but remain at historical highs; allocation-type funds' high tech positions require resolution through multiple aspects in the long term. Efforts to stabilize the capital market policy + waiting for ShenZhen New Industries Biomedical Engineering to catalyze a narrowing of floating losses, necessary position adjustment and switching, and adjustments to policy institutions and accounting treatment may be what is needed to achieve long-term clearing. 2. Currently, the overseas asset allocation environment is facing headwinds, but the turnaround may come from Warsh's monetary policy framework reform, and there are opportunities for improvement in the overseas asset allocation environment in 2027. Domestic A-share corporate earnings in 26Q4-27Q1 will be under pressure reflecting cost pricing impacts, with corporate earnings rebounding in 27Q2 and returning to normal central levels, but with limited elasticity. Domestic government transitions are advancing rapidly, and policy implementation is strengthening at a relatively fast pace. External demand is high in the early period and low in the later period, while domestic demand is expected to be low in the early period and high in the later period. The domestic asset allocation environment in 2027 may be generally stable. Currently, the overseas asset allocation environment is facing headwinds. The trend toward global resource protection is intensifying, with inflation stemming more from the supply side. In global economic governance, anti-inflation, fiscal discipline consolidation, resolution of economic structural problems, and maintenance of reasonable economic growth cannot all be accommodated simultaneously and can only be balanced. Major global central banks prioritizing rate hikes to combat inflation may suppress capital markets. Potential improvement in 2027: Warsh's five major topics prepare the groundwork for monetary policy framework reform. The implementation of special research on economic data, inflation frameworks, and AI's impact on productivity and employment will all make monetary policy anchors more aligned with the characteristics of the era and ease expectations of rate hikes to combat inflation. In addition, a Democratic sweep would bring geopolitical and supply-side forces to combat inflation, and subsequent game-playing between the Democratic and Republican parties may also be a source of improvement. Domestically, based on earnings forecasts for all A-share non-financial/non-oil companies, discussing the fundamental outlook, high costs in 26Q3-27Q1 are becoming fully priced in and may suppress profit growth. Starting from 27Q2, profit growth will return to normal central levels, but with limited elasticity. The forecast for the 2027 year-on-year growth rate of net profit attributable to parent companies for all A-share non-financial/non-oil companies is 6.4% (9.3% in 2026), with quarterly year-on-year figures for 26Q4-27Q4 at 3.5%, 0.4%, 4.6%, 12.0%, and 32.2% respectively (due to base effects). Maintain a strategic stalemate, with domestic economic governance prioritizing stability. Pressure from the economic cycle objectively exists, and marginal policy efforts in 2027 are only appropriate. At the same time, domestic government transitions are advancing rapidly, and policy implementation is strengthening at a relatively fast pace. External demand is high in the early period and low in the later period, while domestic demand is expected to be low in the early period and high in the later period. The domestic asset allocation environment in 2027 may be generally stable. 3. AI is "not wrong," but computing power inflation may be "wrong." In the long-term outlook for the AI industry, there is no general computing power inflation. There are opportunities in the stage dominated by computing power inflation, and there are also opportunities in the stage dominated by computing power deflation in the distant future, but the transition period in between is a headwind period for capital markets. The AI industry is spontaneously correcting computing power inflation: on the demand side, a Democratic sweep, power shortages, and high financing costs have strengthened expectations that cloud computing leaders' capital expenditures will decline year by year in 2026-28. On the supply side, rapid capital expenditure growth generally occurs between 26-27, with the peak of capacity formation likely occurring in 27-28, and a timetable for supply and demand to return to balance already exists. Correspondingly, market confidence in the profitability of the computing power chain in 2028 has loosened somewhat. In the short to medium term, the probability of positive and negative factors for AI large models and applications is neutral, while for the computing power chain, negatives outweigh positives. In the long term, positives outweigh negatives for AI large models and applications, while the computing power chain may return to neutral. The probability of a major band rally being rebuilt based on the computing power inflation logic has significantly decreased. Over a longer time horizon, wait for breakthroughs in AI large models and applications to build the "second-stage rally." Next, discuss the positioning of long-, medium-, and short-term bands, and use this to deduce long-, medium-, and short-term market views. 4. Revisiting the "two-stage major band rally theory": in structural dominated by industry trends, there may be two stages of where earnings and valuations rise in tandem, corresponding to two valuation peaks and two institutional allocation peaks. Before mid-2026, computing power chain valuations and allocations were basically rising one-sidedly, which may only be the valuation and allocation peak of the "first-stage rally." The "correction phase between the two stages of the rally" is typically at the quarterly-to-annual level, and the 26H2 correction has not yet broken through historical patterns. Is there still a "second-stage rally"? Or is it a new round of major-level, weakly correlated with the before mid-2026? This question can be transformed into: can AI industry trends still serve as the main line of the next rally? In the view, it is an objective judgment that there is still room for AI applications to open up new tracks, and it is also an objective trend that the AI industry continues to advance; the space for household asset allocation to migrate toward equities has not yet been effectively utilized. The next major band rally will most likely still be led by AI industry trends. Therefore, the "second-stage rally" is worth anticipating. The "correction phase between the two stages of the rally" may last multiple quarters, but will not span multiple years. For this phase, the historical upper limit is 2014, and the lower limit is 2018. The 26H2 correction level is closer to 2018; in 2018, macro shocks ultimately affected the fundamental trends of core assets (after June 2018, mid-level data for food and beverage weakened), while in 26H2 there are disturbances in both macro and industry, with the result being a major band correction in industry mainline assets. The experience worth referencing from 2014 is that before the "second-stage rally" is established, there will be a phase where index gains are limited but themes are active. Before August 2014, internet (from media leading to computer leading), M&A and restructuring (companies in textiles and apparel, commerce and trade, social services, real estate and other industries concentrating on transformation to internet), and reform (free trade zones, state-owned enterprise reform, Belt and Road, regional themes) all staged active thematic. In this process, consensus strengthened, positions concentrated, and the wealth effect fermented. 5. Long-term cost-effectiveness clearing already has a timetable. Computing power chain fundamental expectations are being revised downward, and valuation declines and cost-effectiveness improvement are rapidly matching this. From a reverse DCF perspective, the matching of long-, medium-, and short-term logic and valuation models across various segments of the computing power chain has nearly fallen neatly into place. This is a characteristic of approaching a medium-term bottom. Only a larger-level fundamental downward revision, extending the major band correction, would require all remaining optimistic expectations to be revised downward, and this remains a low-probability scenario. If 2027 is expected to complete the "construction of the second-stage rally," the can fully unfold on the basis of logic and valuation matching neatly. Currently, the computing power chain's return to a long-term high cost-effectiveness zone already has a timetable. For communications static PE valuation, before the 26Q3 report it was at the historical median, after the Q3 report disclosure it fell below the median, and after the annual report disclosure it will be in the historically extremely low valuation zone. From a reverse DCF perspective, leading companies in optical modules, storage, PCB, and gas turbines have basically no additional growth expectations for 29-30 embedded in current pricing. However, directions such as upstream optical communications, semiconductors, PCB upstream materials, and capacitors that rely on self-controllability and logic of entering the industry chain still have relatively high growth expectations for 29-30 embedded in pricing; these directions have upward fundamental expectations (2027 earnings growth higher than 2026), which is the prerequisite for investment opportunities. Currently, the matching of long-, medium-, and short-term logic and valuation models across various segments of the computing power chain has nearly fallen neatly into place, which is a characteristic of approaching a medium-term bottom. Under such circumstances, only a larger-level fundamental shock that pushes the correction level beyond the "two-stage rally framework" would require all remaining optimistic expectations in sub-tracks to be revised downward. This is only a low-probability scenario. If 2027 is expected to complete the "construction of the second-stage rally," the can fully unfold on the basis of logic and valuation matching neatly. 6. Long-, medium-, and short-term band positioning: In the long term, A-shares are in a "2018-level correction phase between the two stages of the rally." In the medium term, the one-sided correction absolutely dominated by defense is already in its latter stage, possibly followed by a consolidation phase with active bottom themes, which is an inevitable process in building the "second-stage rally." In the short term, momentum for an oversold rebound is accumulating, the balance of capital supply and demand is being rebuilt, and the spring rally in 2027 around the turn of the year is worth anticipating; an earlier-than-expected oversold rebound is also reasonable. Core conclusion of market trend judgment: Before March-April 2027, it may be a window for the oversold rebound to unfold, with A-shares overall no longer making new lows. After March-April 2027, this round of market correction may reach the bottom zone, followed by increased structural opportunities and active themes, with new consensus possibly coalescing in this process. In 27H2, anticipate the construction of the "second-stage rally." In 2027, there may be a clear major inflection point from headwinds in the asset allocation environment, industry trends, and capital supply and demand to tailwinds. Correspondingly, investment opportunities in the second half of 2027 may be better than in the first half, and investment returns in 2027 may be higher than in 2026. 7. In 2027, attach importance to thematic investment: On fundamentals, overall highlights will be fewer than in 2026, and AI industry trends will still face disturbances. On the capital side, track-focused funds and allocation-type funds will find it difficult to form synergy on tech leaders, and the relative strength of trading-oriented capital will regain the upper hand. Before the industry trend is rebuilt, seek breakthroughs from style investment; small and micro caps will outperform, and active thematic may be more sustained. Seeking ShenZhen New Industries Biomedical Engineering consensus from thematic clues is also a necessary process in building the "second-stage rally." Future directions with relatively high attention in AI applications, from the AI-ization of personal office work to corporate collaboration, and AI-ization of cross-company and cross-industry links, are directions with relatively high visibility for future profit and cash flow improvement. At the same time, healthcare AI, embodied intelligence, and AI security are directions with relatively high attention from industry and primary markets among long-term AI application directions, and are also directions with smooth thematic logic. Among non-tech directions, cyclical assets with superior supply-demand patterns can still be traded around the intensity of geopolitical disturbances; during phases of intensifying geopolitical conflicts, the oil chain and shipping outperform; during phases of geopolitical easing and fermentation of optimistic pro-cyclical expectations, industrial metals and basic chemicals outperform. At the same time, under Warsh's monetary policy framework reform and the impact of a Democratic sweep, the overseas asset allocation environment easing, gold can outperform in more medium-term scenarios. For domestic demand trading, the market has full awareness of the necessity of policy efforts, but it is difficult to form consensus on specific policy levers. The combination of domestic demand trading with high-quality factors and high-dividend factors (public high-quality investment, allocation-type funds, and marginal pricing by stabilizing capital market forces) may effectively capture excess returns. High-dividend and tech track investment are merely a "seesaw effect." Currently, all types of capital participate to some extent in high-dividend assets, and the combination of high dividends with back-line targets, high quality, and low valuation/low gains is contributing excess returns. In market headwind phases, high dividends outperform; in active thematic phases, high dividends retreat; in the construction of the "second-stage rally," high dividends may undergo a trend adjustment.