Abstract: From 2020 to 2025, the new PV installed capacity in the world will increase from 130GW to 580GW, with an increase of nearly 4 in five years. In 2025, the new installed capacity in China will be 315GW, accounting for more than 54% of the world's total, and the cumulative installed capacity will exceed 1,200GW, officially entering the terawatt era. But at the back of this rapid growth, the whole industry chain is experiencing the most serious overcapacity in history: the four main links (polysilicon, silicon wafers, batteries, this surplus is not the collapse of demand-demand still maintains a growth trend-but the imbalance between supply and demand caused by the triple factors of capacity transfer, policy rhythm mismatch and enterprise prisoner's dilemma under the rapid switching of technological routes. Technology iteration did not bring the permanent elimination of P-type capacity, and a large number of PERC and TOPCon production lines are being upgraded to BC to continue to serve through technological transformation, which aggravated the difficulty of de-industrialization; at the same time, the BC route has rapidly evolved from the differentiation direction of a few enterprises to the homogenization upgrade of most enterprises, further reducing the flexibility of industry clearing. At present, the difficulty of price repair lies not only in the high productivity rigidity of upstream links such as silicon materials, but also in the fact that no link in the whole chain has substantial bargaining power, and the price repair of any link will be absorbed by the excess capacity of the intermediate link. No effective conduction to the terminal . The recovery of the industry will not be a V-shaped rebound, but the coexistence of L-shaped bottoming and structural differentiation. 1. Supply and demand fundamentals of PV industry chain 1. China's newly installed capacity in 2025 is 315 GW, with a year-on-year growth of 13. The annual compound growth rate of global newly installed capacity in 2020-2023 is about 44%, the growth rate in 2024 is about 36%, and the growth rate in 2025 is about 9%. The slowdown in growth is not caused by shrinking demand, but by the high base effect superimposed on the bottleneck of power grid consumption. 
photovoltaic installed capacity relies on five structural driving forces. The first is the consensus on carbon neutrality-194 countries around the world have pledged to triple the installed capacity of renewable energy by 2030, but the pace of national promotion is divided, the United States is swayed by the political cycle, developing countries are constrained by financing constraints and execution, and the overall implementation is relatively limited. Second, energy security -after the conflict between Russia and Ukraine, Europe has accelerated its energy autonomy (Europe has completed most of its implementation, but is facing the bottleneck of absorption), and emerging markets such as the Middle East (Saudi Arabia's 2030 vision plan for photovoltaic installation of 58 GW), India and Africa have just started. The irreversibility of geo-structure determines its high certainty. Third, global re-industrialization -IEA "Electricity 2026" predicts that the average annual growth rate of global electricity demand will reach 3 in 2026-2030. Fourth, AI and data center electricity demand -AI development exceeds expectations is a deterministic fact. IEA Power 2026 points out that data center power demand is one of the fastest growing terminal power consumption categories. The United States, China and Europe are accelerating the layout of AI infrastructure, with large incremental space for medium and long-term power demand, high certainty and clear trend. Fifth, the demand for commercial aerospace and space power-this is a new dimension with a small volume (annual demand at the MW level), but the direction is highly determined, and the premium logic is totally different from that of ground photovoltaic, which is expected to become an important part of photovoltaic demand in the long run.
Consumption bottleneck is the key constraint on the demand side. In 2025, the negative electricity price in Germany will last 573 hours (an increase of 25% over the same period last year). In the first half of 2025, the rate of light curtailment in China was 5.7% -6. Eighteen provinces in China have issued the detailed rules of " fixing capacity by load"-the installed capacity of distributed projects can not exceed the user's own electricity load-Q1 industrial and commercial distribution dropped by 61% in 2026, and the policy impact has been fully demonstrated. Supporting energy storage has changed from optional to necessary. In 2025, China's new energy storage will increase by 62.24 GW/183GWh (according to the National Energy Administration, GW is the power scale, GWh is the capacity scale, and the ratio of the two is about 2. The structural driving force of demand is solid, but the bottleneck of consumption constitutes the soft ceiling for the growth of PV installed capacity.
In 2025, the export volume of China's PV products (including silicon wafers, cells, modules and inverters, according to the General Administration of Customs) was approximately RMB281.7 billion, representing a year-on-year decrease of 4. At the same time, the export destination structure experienced a historic rebalancing: the share of Europe continued to decline. Significant growth in the Middle East and Asia-Pacific-Q1 China's exports to the Middle East surged 470% year-on-year in 2026, and to the Asia-Pacific region (India, Pakistan, etc.) 168% year-on-year, while the four Southeast Asian countries (Vietnam, Thailand, Malaysia, Cambodia) blocked the transit of Chinese enterprises through Southeast Asia through the United States due to the imposition of double anti-tariffs by the United States. China's component exports to Southeast Asia fell by 65%. The adjustment of export tax rebate will further reduce the profit margin: it will be reduced from 13% to 9% in 2025, and will be completely abolished from April 1, 2026, with a marginal impact of about 0.06-0 on loss-making module enterprises. Global sales are moving towards a complex pattern of "production capacity going to sea, barriers increasing day by day" .
1. The nominal production capacity of polysilicon exceeds 3.5 million tons per year (supported by 10000 tons of silicon materials, about 3.85-4. The global annual installed capacity is about 580 GW, and the capacity ratio is 1.30 (the ratio of DC side installed capacity of modules to AC side installed capacity of inverters, industry design practices refer to NB/T 10394-2020, The typical range is 1.1-1.5, 1. The supply-demand ratio is more than 2:1.
From the perspective of output growth trajectory, each link in 2018-2024 has experienced six years of rapid expansion:
Data source: CPIA 2018-2025 version of China's photovoltaic industry development roadmap
2018-2024. Polysilicon production increased from 250000 tons to 1.82 million tons (6.3 times), silicon wafers from 109.2GW to 753 GW (5.9 times), and batteries from 87.2GW to 654 GW (6.5 times). Components increased from 85.7 GW to 588 GW (5). The compound growth rate of the four links in the industrial chain in the past six years exceeded 35% , far exceeding the growth rate of global installed capacity in the same period (about 27% annually). The explosive growth of output is concentrated in 2022-2023: in 2022, the growth rate of all sectors is 57% -64% . In 2023, batteries and components further accelerated to 65% -69% , which is the stage of centralized release of capacity expansion. In 2025, there was a key turning point: the output of polysilicon decreased by 26.4% year-on-year (the first time since 2013), and the output of silicon wafers decreased by 9.
In 2025, the output of each link was obviously differentiated. The output of polysilicon is about 1.34 million tons, with a year-on-year decrease of 26.4%, and the capacity utilization rate is about 38%, which is the first decline since 2013; the output of silicon wafers is about 680GW, with a year-on-year decrease of 9.7%, which is the first negative growth since 2009; the output of batteries is more than 660GW, with a slight increase of 0.9%； Module production exceeds 620 GW. Year-on-year increase 5.
Data source: Solarzoom, Silicon Branch, CPIA
Global component prices are seriously differentiated : about $0.27/W in the United States (comprehensive tariffs exceed 100%), and 0.6-0.67/W in China. The Middle East is about 0.10-0.The price difference is mainly affected by trade barriers, localization requirements and transportation costs. According to the calculation of Solar zoom, in May 2025, the gross profit per watt of components in the whole industry is -0.09 yuan/W, and the gross profit per watt of silicon materials is -0.01 yuan/W; From the annual average, the average price of batteries and components in 2025 is 0.289 yuan/W and
0.1 yuan, respectively. The net profit of Tongwei shares is-9.553 billion yuan, Longji Green Energy-6.420 billion yuan, TCL Central-9.264 billion yuan, Jingke Energy-6.882 billion yuan, Jingao Science and Technology-4.608 billion yuan. Trina Solar-70. The total net loss of the five component giants is about 34.5 billion yuan. In 2026 Q1, the total loss of 22 photovoltaic listed companies exceeded 10.5 billion yuan.
.)
2. Analysis
of the causes of overcapacity 2. Almost all enterprises choose to upgrade at the same time, which leads to the concentrated outbreak of TOPCon capacity. In 2025, the market share of N-type batteries has reached 97% , of which TOPCon accounts for 87.6% , and P-type PERC is only 3.
However, this does not mean that P-type production capacity has been substantially eliminated. As P-type production lines can be upgraded to BC (TBC/ABC/HPBC, etc.) production lines through transformation, many PERC production lines put into operation in 2022 and before have not been permanently withdrawn, but continue to participate in the competition by switching to BC technical routes through technological transformation, which aggravates the difficulty of de-industrialization.
At the same time, N-type substitution has not ended the homogenization competition. The threshold of TOPCon technology is not high, and the effective capacity formed after the influx of a large number of enterprises exceeds 700 GW, including the total capacity under construction exceeding 1,300 GW. The difference of mass production efficiency among enterprises is very small (25.7% -26. The time gap of investment cost aggravates the plight of enterprises: at the beginning of 2023, the investment of TOPCon production line is about 170 million yuan/GW, and by the end of 2025, it has dropped to about
0.2. In 2021-2023, the price of silicon materials has soared to 300000 yuan/ton. The profits of the industrial chain are highly concentrated in the upstream, and the net profit of Tongwei shares in 2022 is as high as 25.7 billion yuan. This excess profit sends a strong signal that expanding production is making money. From 2022 to 2024, the whole industry launched the largest capacity expansion in history, with only four silicon enterprises, Tongwei, Xiexin, Daquan and Xinte, adding more than 1 million tons of new capacity.
However, there is a significant time lag between the policy rhythm and the market cycle. In 2024, the distributed photovoltaic policy shifted ( "load to capacity"), and in 2025, the export tax rebate for modules was reduced from 13% to 9%, which was completely abolished on April 1, 2026. It takes 1-2 years for these policies to produce results, but enterprises still expand production according to the old logic when the policies are introduced. The policy signal has the characteristics of two-way misleading: the upstream sees high profits and desperately expands production, while the downstream sees low component prices and desperately installs machines. The two directions are enlarged at the same time, and eventually reverse at the same time after a certain point in time. The scissors gap between supply and demand reaches its maximum in 2025-2026.
2. When the industry is on the rise, every company is worried that competitors will lose market share if they do not expand production. Tongwei announced in 2023 that it would invest 28 billion yuan to build a project with an annual output of 500000 tons of industrial silicon + 400000 tons of high-purity silicon, and Longji announced that it would invest 45.2 billion yuan to build a 100GW silicon wafer + 50GW battery project. These huge investments are based on the assumption that demand will continue to grow at a high rate, but the result of all enterprises expanding production at the same time is that no one profits. This dilemma is aggravated by the strong specificity of
equipment and the low residual value of exit. The residual value is very low when changing production or selling second-hand equipment (second-hand TOPCon equipment is about 0.5 billion yuan/GW vs new equipment 0. For silicon enterprises, shutdown also means huge restart costs of oven drying, cleaning and commissioning, so even if the price falls below the cash cost, there is still power to maintain part of production. In December 2025, Tongwei, Daquan, Xiexin and other leading silicon enterprises announced production control and signed self-discipline conventions, but the actual effect was limited-the annual output of polysilicon in 2025 was about 1.35 million tons, the inventory increment was about 30000 tons, and the reduction of production did not really digest the inventory.
2. The cause of surplus is the superposition of two forces: one is the homogenization of TOPCon expansion-all enterprises pour into the same technology route in the same window period, forming a high degree of convergence of 700GW effective capacity, and the only competitive dimension is price; Second, after the signal of excess profits generated by the dual-carbon policy is sent out, there is a time lag of 1-2 years between the decline of subsidies and the market-oriented policy shift, and the capacity decisions made by enterprises under the old signal are collectively invalid in the new environment.
The reason why this round of excess is "structural" rather than "cyclical" is that the key is not that P-type capacity is permanently eliminated, but that the speed of technology switching far exceeds the speed of market digestion. The P-type production line can enter the BC route through technical transformation and continue to serve, and the TOPCon production line can also be upgraded to BC. The production capacity has not been withdrawn, but has continued to move along the technical route and continue to participate in market competition. What really causes the excess is the centralized launch of N-type capacity (especially TOPCon), and the BC route has changed from differentiated technology to the "new homogenization" direction of many enterprises in a short period of one year. As of April 2026, 17 enterprises have announced BC product layout, and the differentiated dividend window of BC components is rapidly narrowing. In this context, the recovery of the industry will not be a V-shaped rebound, but the coexistence of L-shaped bottoming and structural differentiation.
3. In-depth analysis
of each link in the industrial chain 3.1 Polysilicon: the steepest cost curve and the most
rigid production capacity China's nominal production capacity of polysilicon is about 3.5 million tons per year. Six giants (Tongwei 820,000 tons, Xiexin 480,000 tons, Xinte 300,000 tons, Daquan 300,000 tons, Dongfang Hope 25. Support 3.85-4 by 10,000 tons of silicon materials.
The cost curve determines the order of clearance. The cost curve of silicon material link is the steepest in the industrial chain. Three-level differentiation is significant:
 decreased from 115 million yuan/thousand tons in 2018 to 0 thousand tons in 2024. The continuous decline of investment threshold is one of the structural reasons for the expansion of silicon production capacity from less than 500000 tons in 2018 to more than 3.5 million tons in 2025. </p><p style=)
Data source: CPIA version of China's PV industry development roadmap
, the current average spot price of N-type re-feeding is 33,500 yuan/ton, which has broken through the variable cost of the third echelon (> 40,000 yuan/ton). Approaching the second echelon can become this line (~ 3). Clearing faces three rigid constraints : First, the construction period is 18-24 months and the process is complex, which determines that the elasticity of capacity supply of silicon materials is the lowest in the industrial chain; Second, the cost of stopping production and restarting is very high (the one-time cost of oven + cleaning + commissioning can reach tens of millions of yuan, and the quality of products is unstable in the first few months), which leads to the tendency of enterprises to carry on rather than withdraw voluntarily; Third, the capital density is the highest (the investment of 10000 tons is about 7-10 billion yuan), and the high depreciation burden makes the exit decision extremely painful. When the loss lasts long enough and the cash flow finally runs out, Clearing is still inevitable-referring to the silicon material trough in 2020 (5.
The ten-year price cycle-the extreme reversal from 300000 yuan to 33500 yuan-is the key to understanding the core logic of the silicon material cycle:
the clearing of each silicon material cycle is mainly driven by the recovery of demand. But this round is different in that the capacity base has jumped from 1 million tons to 3.5 million tons, and the depth of loss is also at a historical level (the spot price is extremely low in 2020). It is the survival test of "enduring continuous losses until the withdrawal of production capacity".
3. In 2025, the annual output was 680GW, representing a year-on-year decrease of 9. N-type silicon wafers accounted for 97%, and the total large-size silicon wafers accounted for 99.
The competition pattern was extremely dispersed. The CPIA Roadmap fully records the evolution of concentration:
 and batteries (clear clearing catalysis-technology iteration), silicon wafers lack clear clearing driving force, and about 70% of enterprises continue to lose money but are difficult to exit, which is the slowest and most painful link in the four main links. 3.3 </p><p style=)
The global battery production capacity is about 1,300GW, and the market share of TOPCon is 87.6%. Mass production efficiency 25.7% -26. TOPCon completed from 15% to 87 in 18 months.
; by 2022, TOPCon has risen to 24.5%, opening a significant gap with PERC (23.2%) for the first time; In 2024, TOPCon reached 25.4%. PERC stagnates at 23. Efficiency gap widens from 0.7 percentage points to 1. </p><p style=)
XBC is the direction of differentiation breakthrough, but now it has rapidly presented new homogenization risks. Longji is fully betting on BC. In 2026, the BC production capacity target is 60GW, the HIBC laboratory efficiency is 28.13%, and the mass production component efficiency is 25. However, the BC cost is still higher than TOPCon, and the scale challenge is great. Longji will lose 64 in 2025. Meanwhile, in 2026, photovoltaic enterprises will lay out BC products one after another. The BC route is rapidly evolving from "differentiation of a few enterprises" to "standardization and upgrading of most enterprises" , and its premium window may be rapidly compressed. The market share of HJT is only 2.6%, the cost of silver consumption is difficult to compete with TOPCon, and the investment in equipment is huge (300 million yuan/GW vs TOPCon). However, the technology of 0BB + silver-clad copper is pushing the silver content down to 40%, and copper electroplating is moving towards mass production. HJT may achieve cost reversal in the next 2-3 years.
3. In 2025, the annual output exceeded 620GW, the module price was 0.6-0.CR5 about 55% (Jinko/Longji/Tianhe/Jingao/Tongwei), the pattern was relatively stable, but the profit was overall loss. The profit margin of
component processing and manufacturing has been reduced to below zero by price war, and the value creation of the industry is moving from single manufacturing to two directions: one is to obtain efficiency premium through the next generation battery technology (BC, perovskite stack), and to exchange technology generation difference for pricing power ; The second is to extend the business from selling components to selling systems, and change the revenue model from one-time equipment sales to continuous service revenue. The case of Sunshine Power has reference significance-its energy storage business revenue in 2025 is 37.287 billion yuan (+ 49.39% year-on-year), overseas revenue accounts for 60.54%, and net profit is realized under the background of collective loss at the manufacturing end. With the accelerated withdrawal of small and medium-sized enterprises, CR5 is expected to rise to more than 70%, but the real way out for component enterprises is not to sell components, but to provide integrated optical storage system solutions.
3. On the whole, this is the only " relative hedge zone " in the photovoltaic industry chain, but there are some differences in each subdivision track.
Inverter : The main battlefield moves from photovoltaic to energy storage. In 2025, the global inverter shipment of Sunshine Power exceeded 143 GW, the energy storage shipment was 28 GWh, and the energy storage business revenue increased by 49.39% year-on-year to 372.2026 Q1 experienced twists and turns: the revenue was 15.561 billion yuan, down 18.26% year-on-year, and the net profit attributable to the parent company was 2.291 billion yuan. The year-on-year decrease of 40.12% was mainly due to the same period in 2025. 7. Excluding the above special factors, the bottom of the energy storage business grew by about 60% , and the gross profit margin of inverters was about 40%, up 3 percentage points year-on-year-the core fundamentals remained stable. Significant differentiation within the industry: Deye's 2026Q1 revenue increased by 73.77% and net profit increased by 68.37%, mainly due to the high degree of overseas (overseas revenue accounted for more than 93%) and the high growth of energy storage battery packs; In 2025, Goodway turned losses into profits after completing the transition dormant period, and the sales volume of energy storage batteries increased by 481.81% year-on-year; Jinlang Technology is in the transition investment period, and its net profit decreased. In 2025, the revenue of energy storage inverters increased by 185.31% year-on-year, and the contribution of energy storage system from zero was 1.82%.
Energy storage batteries : both volume and price are rising, and Chinese enterprises are leading the world. According to BNEF data, the installed capacity of new energy storage in the world will reach 112 GW/307 GWh in 2025, with a year-on-year growth of 48%; it is expected to further increase to 158 GW/459 GWh in 2026, with a year-on-year growth of 41%. Frost Sullivan data show that global energy storage inverter shipments increased from about 10.4G W to about 196.5G W in 2021-2025, with CAGR as high as 108.5% in five years, and is expected to reach about 872 in 2030. In 2025, Ningde's shipments reached 121 GWh, ranking first in the world (market share exceeded 26%), Yiwei Lithium Energy 71.05 GWh (+ 40.84% year-on-year), BYD exceeded 60 GWh, and Ruipu Lanjun's revenue increased 86.2026 Q1 year-on-year. The market share of Ningde Times Power + Energy Storage rebounded to 50. It is worth noting that the rising price of lithium carbonate in 2026 will exert cost pressure on energy storage system manufacturers (such as Sunshine Power Supply), which mainly purchase batteries from outside, but the price transmission in overseas market is relatively smooth, while the transmission in domestic market is limited.
 shows phase in excess cycle For stable profit resilience, Flat 2025 PV glass business revenue 139.At the end of 2025, six departments jointly issued a document to strictly prohibit new flat glass production capacity, new production line approval is more difficult, policy supply constraints have been established, once the demand side stabilizes, PV glass will be one of the first transmission links of supply and demand improvement in the industrial chain. The increase in the proportion of N-type double-glass components (N-type demand for double-sided glass) is a structural advantage, which directly drives the increase in the consumption of glass per GW component. </p><p style=)
The key leading signal is the change 5. Medium and long-term forecast
5. CPIA forecasts that during the 15th Five-Year Plan period (2026-2030), the world's annual average installed capacity will increase by 725-870 GW, reaching 881-1,044 GW in 2030. However, the shift of growth rate is a deterministic event : the compound annual growth rate of about 35% in 2020-2025 is unsustainable, and the first negative growth or zero growth in nearly ten years may occur in 2026 (triple superposition of high base + policy adjustment + consumption bottleneck). It is expected to return to the rising channel after 2027, but the growth rate will slow down significantly. The bottleneck of absorption will continue to strengthen-when the proportion of photovoltaic in the power structure is further increased, the rigidity of energy storage matching will become more prominent, and the integration of photovoltaic and storage will change from trend to standard.
5. The price of silicon materials is 3.5-4. In 2025, more than 100 enterprises have entered liquidation, and more than 30 enterprises are expected to collapse in 2026. The financial statements are still deteriorating, but the stock price may hit the bottom before the fundamentals. The core observation indicator is the operating rate: when the average operating rate of the whole industry rises from less than 40% to more than 50%, it indicates that the production capacity begins to be substantially cleared.
From 2028 to 2030, supply and demand tend to be weakly balanced. In view of the fact that the P-type production line can be upgraded to BC through technological transformation and continue to serve, the actual liquidation depends on the comprehensive game of enterprise cash flow, transformation cost and market price: the tail capacity with exhausted cash flow and unable to technological transformation will be withdrawn; Production lines with technological transformation conditions turn to BC to continue to participate in competition . If the inefficient TOP Con of about 100-165 GW accelerates its withdrawal in the continuous loss, the effective capacity is expected to be reduced to 700-800 GW in 2028. During the same period, the global demand will grow according to the baseline scenario (about 870 GW per year), and the effective supply-demand ratio will drop to about 1.0-1. However, it should be noted that the expansion rate of PV production capacity is very fast (the module production line can be put into operation in 3-6 months), and any window of supply and demand balance may be quickly closed due to a new round of expansion.After
2030-a new round of technology cycle. If the net zero scenario is realized (global annual installed capacity exceeds 1,000 GW +), there may be a structural gap in efficient production capacity. The timing of the mass production of the BC/perovskite technology is the biggest variable: the laboratory efficiency of the perovskite stack has reached 34. If the mass production of the perovskite is carried out ahead of schedule, the existing 1,300 G W TOPCon of production capacity will face technical sinking, which is equivalent to the industry experiencing secondary liquidation. 5.3 Judgment
on each link of the industry chain 
competition pattern provides the basis for the long-term healthy development of the industry. Component CR5 has risen from about 43% in 2022 to about 55%, and is expected to rise further to more than 70% in 2027-2028. Longji BC's capacity planning exceeds 50GW and Aixu ABC exceeds 30GW, which has verified 0.05-0. Sunshine Power's energy storage business revenue in 2025 is 37.287 billion yuan (+ 49.39% year-on-year), and overseas revenue accounts for 60. The focus of industry governance is shifting from price intervention to standard-setting-improving energy consumption standards, strengthening intellectual property protection, improving the system of industrial fund relief and bankruptcy reorganization, and providing institutional guarantee for market-oriented liquidation. The certainty of
medium and long-term demand increment is still solid. The CPIA predicts that during the 15th Five-Year Plan period (2026-2030), the global annual average installed capacity will increase by 725-870 GW, reaching 881-1,044 GW by 2030. The incremental space in emerging markets (Saudi Arabia's 58 GW target in the Middle East, India's 500 GW target, and Africa's electrification gap of more than 600 million people) is far from being realized. If green hydrogen achieves an economic breakthrough in 2028-2030, every 10 million tons of green hydrogen will correspond to 600-800 GW of PV installed demand. Commercial aerospace (SpaceX launched in June 2026, satellite photovoltaic market) sets higher efficiency and reliability requirements for photovoltaic technology, and technology spillover effect will feed back the iteration of ground products.
Generally speaking, the scale of this round of liquidation is unprecedented, the liquidation process will last for 2-3 years, and the probability of V-shaped rebound is low . However, the underlying logic of industry self-repair has not changed-each round of crisis has promoted technological progress and cost reduction while eliminating backward production capacity. After this cycle, photovoltaic enterprises will transform from periodic manufacturing enterprises driven by subsidies to global energy infrastructure system service providers supported by carbon neutral consensus, energy security just need and technology endogenous cost reduction, and the development level of the industry will go up to a higher level. Enterprises that survive and complete technological upgrading and globalization will face a larger and more sustainable market space.
浙公网安备33010802003254号