Shen Wan Hongyuan: The next round of growth may still be dominated by AI industry trends and waiting for the “second phase of rise” to be built
The Zhitong Finance App learned that Shen Wan Hongyuan released a research report saying that in 2026, A-shares are still operating within the framework of the “two-stage high-band rise theory”. However, the first phase of the 26H1 rise level was higher, and the 26H2 adjustment level was also scaled up accordingly; however, expectations for the “second phase rise” should not be abandoned. 27H2 is expected to complete construction, and the return on investment may be higher than 26 years. It is an objective judgment that the AI application side still has room to open up a new track. The AI industry continues to advance and is also an objective trend. The probability of the next big wave of growth will still be dominated by AI industry trends. In the long run, A-shares are in the “2018 level adjustment phase between two stages of growth.” In the medium term, the unilateral adjustment, which is absolutely dominated by defense, is already in the late stages. This may be followed by a volatile phase where the bottom theme is active. This is a necessary process for constructing a “second phase of growth.” In the short term, momentum for an overfall rebound has accumulated, and the balance between capital supply and demand has been rebuilt. The market can be expected in the spring of '27 at the end of the year, and it is reasonable to interpret the overdraft rebound even earlier.
Shen Wan Hongyuan's main views are as follows:
At this stage, problems in the asset allocation environment, core industry trends, and stock market liquidity levels exist objectively, and there are no specific clues about the turnaround; there is only a clear timeline for improving cost performance in the short, medium and long term. At this point, it's easy to extrapolate carefully, and difficult to reverse discussions. Has the 2026 A-share market broken through the rules of history? The bank believes that the market may still operate within the framework of the “big wave rise two-stage theory”. It's just that the “first stage rise” level of 26H1 is higher, and the “adjustment period between two stages of rise” level of 26H2 has also been expanded accordingly. However, against the trend, it seems that expectations for a “second phase rise” should not be abandoned. The AI application side still has room to open a new track is an objective judgment, and the continuous progress of the AI industry is also an objective trend. When pricing to correct AI computing power inflation tends to be adequate, the outlook for AI industry trends should also gradually shift from looking out for shortcomings and hoping for something favorable. Looking at the current questions objectively, while thinking about the turning point of 2027 and looking for total and structural breakthroughs, this is probably a question that needs to be answered in the 2027 strategy.
First, discuss the market adjustments starting in mid-2026, which reflect the key issues and the mid-term turning point of these issues.
1. The 2026 rise and adjustment period. The characteristics of capital supply and demand are symmetrical: the upward wave, extreme differentiation under the inventory game, and the positive cycle of track-type funds is four or two thousand kg; in the adjustment period, various types of capital are generally struggling to reduce positions. Concerns about allocating capital and absolute profit capital reducing positions drive the negative capital cycle.
At this stage, positions of various types of capital have generally been reduced. Because the net value has fallen close to the break-even line, the allocation ratio of concentrated position reduction has increased. Conditions for restoring balance between short- and medium-term capital supply and demand have improved. However, high positions in the racetrack fund and allocation-type capital computing power chain have not been cleared for a long time. New industry trends, a certain degree of stock exchange, and policy and system adjustments are the conditions for long-term settlement.
The 2026 rise and adjustment bands are symmetric in terms of fundamentals and financial logic. In the upward wave, fundamentals, the logic of computational power inflation continues to spread; in terms of capital, track-type funds are circulating four to two thousand pounds. 25Q3 is a historical-level incremental game, but 26H1 is an extreme differentiation under the stock game. At the same time, there is a lot of offensive capital, and there is little allocation capital. The positive cycle of public circuit funds is marginal pricing capital, and the internal public offering is also an inventory game. Funds flow out of non-technology tracks and focus on high-profile technology track funds. However, in the downward wave, the logic of computing power and inflation is weakening objectively; in terms of capital, various types of capital are generally struggling to reduce their positions. Allocation capital and absolute profit capital reduction form marginal capital and drive a negative capital cycle. Funds that may be forced to reduce positions near the break-even line (corporate annuity, fixed income +, bank financial management, etc.) centrally triggered a reduction in positions in October '26, while other types of capital (two types of finance, private placement, active public placement, etc.) dropped simultaneously. After various types of capital were generally reduced, policies to stabilize the capital market were strengthened, and conditions for restoring the balance between supply and demand for short- and medium-term capital were improved. However, from a long-term perspective, the share of capital that has been cleared from high-tech positions is still low. The downsizing of the two financing and private equity positions is relatively sufficient; public circuit funds usually have to wait until a new offensive structure appears; corporate annuity and fixed income + technology positions have declined marginally, but they are still at historically high levels; high technology positions in the allocation capital need to be resolved in many ways in the long term. Strengthening capital market policies + waiting for new industries to catalyze narrowing losses, and necessary adjustments in position adjustments, policy systems, and accounting processing may be possible to achieve long-term settlement.
2. The current overseas asset allocation environment is headstrong, but the turning point may come from the Walsh monetary policy framework reform. There is an opportunity to improve the overseas asset allocation environment in '27. Domestic A-share companies' profits 26Q4-27Q1 reflect the impact of cost pricing under pressure. The profits of 27Q2 companies rebounded and returned to a normal center, but flexibility was limited. Domestic government changes are progressing rapidly, and the pace of strengthening policy implementation is relatively fast. External demand is high and low, and domestic demand can be expected to be high before and after. The domestic asset allocation environment is likely to be generally stable in '27.
The current overseas asset allocation environment is headstrong. Global resource conservation tendencies are heating up, and inflation stems more from the supply side. In global economic governance, fighting inflation, refining fiscal discipline, managing economic structural problems, and maintaining reasonable economic growth cannot be balanced, but only trade-offs. The world's major central banks give priority to raising interest rates to fight inflation, or put pressure on capital markets.
Potential improvements in 2027: Walsh's top five topics prepare for monetary policy framework reforms. The implementation of economic data, inflation frameworks, and AI research on productivity and employment will all anchor monetary policy more in line with the characteristics of the times and ease expectations of interest rate hikes against inflation. Furthermore, the Democratic Party's sweep brought geographical and supply-side strength to fight inflation. A subsequent game between the Democratic and Republican parties may also be a source of improvement.
Domestically, based on profit forecasts for all A and Africa, discussing the fundamental outlook, the high costs of 26Q3-27Q1 tend to be sufficient, which may suppress profit growth. Meanwhile, profit growth began returning to a normal center in 27Q2, but flexibility was limited. The forecast year-on-year growth rate for all A's net profit in 2027 is 6.4% (9.3% in 2026), and the 26Q4-27Q4 single quarter is 3.5%, 0.4%, 4.6%, 12.0%, and 32.2%, respectively (base reason).
Maintain consistent strategies, and take the lead in stabilizing domestic economic governance. Economic cycle pressure exists objectively, and the marginal strength of the 27-year policy is the proper meaning. At the same time, domestic government changes are progressing rapidly, and the pace of strengthening policy implementation is relatively fast. External demand is high and low, while domestic demand can be expected to be high before and after. The domestic asset allocation environment is likely to be generally stable in '27.
3. AI is “right,” but computing power inflation may be “wrong.” There is no general hashrate inflation in the long-term outlook for the AI industry. There are opportunities in the phase where computing power inflation dominates, and there are opportunities in the phase where forward hashrate deflation dominates, but the transition period in the middle is a headwind period in the capital market. The AI industry's spontaneous correction of computational power inflation: On the demand side, the Democratic Party swept away, electricity shortages, and high financing costs, increasing expectations that the capital expenditure of leading cloud computing companies will decline year by year in 2026-28. On the supply side, high increases in capital expenditure generally occurred between 26-27. The peak of production capacity formation may occur in 27-28, and there is a schedule for supply and demand to return to balance. The corresponding market's confidence in the 2028 computing power chain's profit has loosened.
In the short to medium term, AI models and applications have a negative probability of being favorable, while the computing power chain has more disadvantages than advantages. In the long run, AI models and applications have more advantages than disadvantages, and the computing power chain may return to neutrality. The probability of a large-scale rise based on the logic of computational power inflation has been significantly reduced. Extend the time, wait for AI models and application-side breakthroughs, and build a “second phase of growth.”
Next, we will discuss long-term, medium- and short-term band positioning, and use this to derive long-term, medium- and short-term market views.
4. Let's talk more about the “two-stage rise theory”: in a structural market dominated by industrial trends, there may be two periods where performance and valuation resonate upward, corresponding to two peak valuations and two peak institutional allocation. Before mid-2026, the valuation and allocation of the computing power chain will basically rise unilaterally. This may only be the high point of the “first phase increase” in valuation and allocation. The “adjustment period between two stages of growth” is usually at the quarter-to-annual level, and the 26H2 adjustment has yet to break through historical rules.
Is there still a “second phase increase”? Or is it a new round of large-scale market related to the weak market 26 years ago? The question can be translated into whether the AI industry trend can still be the main line of the next round of growth? In my opinion, the AI application side still has room to open up a new track is an objective judgment, and the continuous progress of the AI industry is also an objective trend; the space for migrating residents' asset allocation to equity has not yet been effectively used. The probability of the next big wave rise will still be dominated by AI industry trends. Well, the “second phase of the rise” is worth looking forward to.
The “adjustment period between two stages of growth” may take several quarters, but it will not take many years. At this stage, the upper limit of historical experience is 2014, and the lower limit is 2018. The 26H2 adjustment level is closer to 2018. In 2018, macroeconomic shocks finally affected the fundamental trend of core assets (after June '18, food and beverage data weakened). 26H2 disrupted both the macro and industry. As a result, the main assets of the industry were adjusted in a large range. An experience worth referring to in 2014 is that before the “second phase of growth” is established, there will be a stage where the index increase is limited, but the theme is active. Prior to August 2014, the Internet (leading from media to computer leadership), mergers and acquisitions (centralized transformation of companies in textile and clothing, commerce, social services, real estate and other industries), and reforms (free trade zones, state-owned enterprise reform, Belt and Road, regional themes) all showed active themes. In this process, consensus was strengthened, positions were focused, and money-making effects fermented.
5. There is a timeline for clearing the long-term cost performance ratio. The fundamentals of the computing power chain are expected to be revised downwards, and declining valuations and improved cost performance are quickly in line with this. Looking at reverse DCF, the various links in the computing power chain, the matching of long, medium, and short-term logic and valuation models are close to being fully settled. This is a characteristic of near the mid-term low. Only when the fundamentals are downgraded at a higher level and extended in a larger band will all the remaining optimistic expectations need to be revised down. This is still a small probability. If the “second phase of upward construction” is expected to be completed in 2027, the market can completely unfold on the basis of logic and valuation matching.
Currently, there is a schedule for the computing power chain to return to a long-term cost-effective region. Communications static PE valuation. It was in the historical median before the three-quarter report in '26, lower than the median after the disclosure of the three-quarter report, and will be in the historically extremely undervalued region after the disclosure of the annual report. Looking at reverse DCF, leading companies in the optical module, storage, PCB, and combustion engine segments have almost no hidden expectations of additional growth in 29-30 years in their current pricing. However, depending on the direction of upstream optical communication, semiconductors, PCB upstream materials, capacitors, etc., which are autonomous and controllable and enter the logic of the industrial chain, the pricing still includes high growth expectations for 29-30 years. The improvement in fundamental expectations in these directions (the performance growth rate in '27 is higher than in '26) is a prerequisite for investment opportunities. Currently, the various links in the computing power chain, the matching of long-term, medium, and short-term logic and valuation models are close to being fully settled. This is a characteristic of nearing the bottom of the medium term. Under such circumstances, only a larger level of fundamental impact pushes the level adjustment to break through the “two-stage upward market framework”, and the remaining optimistic expectations in the segmented circuit need to be completely revised down. This is just a low-probability scenario. If the “second phase of upward construction” is expected to be completed in 2027, the market can completely unfold on the basis of logic and valuation matching.
6. Long-term, medium and short-term band positioning: In the long term, A-shares are in the “2018 level adjustment phase between two stages of rise”. In the medium term, the unilateral adjustment, which is absolutely dominated by defense, is already in the late stages. This may be followed by a volatile phase where the bottom theme is active. This is a necessary process for constructing a “second phase of growth.” In the short term, momentum for an overfall rebound has accumulated, and the balance between capital supply and demand has been rebuilt. The market can be expected in the spring of '27 at the end of the year, and it is reasonable to interpret the overdraft rebound even earlier.
The core conclusion of the general survey: Before March to April '27, it was probably an interpretation of a sharp decline and a rebound; overall, there was no window for A-shares to reach new lows. After March to April '27, the current round of market adjustments may reach the bottom, then structural opportunities will increase, themes will be active, and new consensus may be forged in the process. 27H2 is looking forward to a “second phase rise” construction. In '27, there may be a major inflection point where the asset allocation environment, industry trends, and capital supply and demand resonance headwinds resonate clearly. Corresponsively, investment opportunities in the second half of '27 may be better than in the first half of '27, and return on investment in '27 may be higher than in '26.
7. Focus on thematic investments in 2027: fundamentals. The overall highlights are less than 2026, and the computing power chain industry trend is still disrupted. On the financial side, it is difficult for track-type funds and allocation-type funds to form a synergy in leading technology, and transactional funds have regained their relative advantage. Before the industrial trend market is rebuilt, breakthroughs are found in style investment. Small micromarkets are dominant, and the market with active themes is likely to continue. Finding new industry consensus from subject leads is also a necessary process for constructing a “second phase of growth.”
The future direction with a high level of attention on the AI application side, from the AI transformation of personal offices to the AI of corporate collaboration and cross-company and industry links, is a direction where future profit cash flow improvement is highly visible. At the same time, medical AI, physical intelligence, and AI security are directions that have received relatively high attention from the industry and the primary market in the long-term AI application direction, and are also directions with smooth subject logic.
In the non-technological direction, cyclical assets with an advantage in the supply and demand pattern can still be traded around the intensity of geographical disturbances. In the phase where geographical conflict intensifies, oil chains and shipping dominate; in the phase of geographical easing, procyclical optimism anticipates fermentation, industrial metals and basic chemicals dominate. At the same time, under the influence of Walsh's monetary policy framework reform and the Democratic Party's sweep, the overseas asset allocation environment has eased, and gold can dominate in more medium-term scenarios. With domestic demand transactions, the market is fully aware of the need to strengthen policies, but it is difficult to reach a consensus on specific policy steps. The combination of domestic demand transactions with high quality factors and high dividend factors (public funding of high-quality investment, allocation of capital market power, marginal pricing) may effectively seize excess income.
High dividends and tech circuit investments are just a “seesaw effect.” Currently, all types of capital are participating in high-dividend assets. The combination of high dividends and late-line markings, high quality, undervaluation/low growth is contributing to excess returns. High dividends prevail during the headwind phase, high dividends decline during the active phase of the theme, and a “second phase rise” structure. High dividends may be trending.