Raising the roof: Singyes still CL-Buy; Initiate on Longi/Trina Neutral

Transcription

1 Equity Research Raising the roof: Singyes still CL-Buy; Initiate on Longi/Trina Neutral Rooftop solar poised to come of age We see China s rooftop (i.e., distributed generation) solar industry at an inflection point poised to grow rapidly in coming years, driven by growing business model maturity and policy support. Key developments: (1) a better financing environment, such as power generation insurance, launch of yieldco financing platforms and growing interest from capital markets; (2) Crystallization of investment returns through potential introduction of net metering; (3) Alignment of interests among value chain participants with State Grid and property management firms now participating in PV projects investment. RATINGS AND TARGET PRICES Price Potential Company Name 12m TP downside Ticker 8-Aug-14 Ratings (%) Singyes Solar 0750.HK HK$12.7 Buy* HK$ % GCL-Poly 3800.HK HK$2.6 Buy HK$3.3 28% Longi Silicon SS Rmb17.3 Neutral Rmb % Trina Solar TSL US$11.5 Neutral US$12.3 7% Canadian Solar CSIQ US$25.4 Neutral US$26.6 5% Jinko Solar JKS US$25.4 Neutral US$27.4 8% Note: * denotes stock is on Conviction List. Mono c-si cell/wafers to benefit growing market share to 35% We believe rooftop solar pickup will lift demand for high efficiency solar cells/wafers (particularly mono c-si) given their lower unit installation cost and narrower premium over multi products. We expect 22% demand CAGR in for mono wafers, growing to account for 35% of total c-si market share by 2017 (vs. 29% in 2013). Supporting prices, we expect supply to remain tight then, thanks to high industry entry barrier and growing market concentration. Singyes and Longi the best positioned in rooftop solar We remain CL-Buy on Singyes due to its strong exposure to the rooftop EPC (Engineering Procurement Construction) segment. Although limited new orders were secured in 1H14, we expect the potential policy rollout in 2H14 will enable Singyes to resume its order flows. We still like GCL-Poly (Buy) for its leading position in the wafer market. We initiate coverage on Longi (Neutral), the world s largest mono-wafer maker with c.12% global market share in We like its exposure to mono wafers as a beneficiary of robust rooftop demand growth. We project Longi to register 78% net profit CAGR during , mainly driven by capacity expansion and cost reductions, but see much of this is priced in. We initiate on solar component maker Trina with Neutral. We see pressure on its manufacturing business s margin due to high on tariff assigned by the US. Remain Neutral on US-listed panel makers Canadian Solar and Jinko. We revise our E EPS by -21% to 11% for our solar coverage and adjust our EV/GCI vs CROCI/WACC 12-m target prices by -26% to 11% accordingly. RELATED RESEARCH Global Clean Energy Monthly: Spotlight on YieldCo potential and implications for value creation in global Clean Energy, Aug 06, Singyes Solar (0750.HK): Strategic agreement with GCL enhances growth visibility; CL-Buy, Aug 05, MONO NARROWS THE GAP: MULTI AND MONO WAFER NON-SILICON COST COMPARISON (US$/W) Mono Multi Risks: slower global solar installation and delay in policy support in China. Frank He Goldman Sachs does and seeks to do business with +86(21) frank.he@ghsl.cn Beijing Gao Hua Securities Company Limited Shuai Liu +86(21) shuai.liu@ghsl.cn Beijing Gao Hua Securities Company Limited companies covered in its research reports. As a result, investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this report. Investors should consider this report as only a single factor in making their investment decision. For Reg AC certification and other important disclosures, see the Disclosure Appendix, or go to Analysts employed by non- US affiliates are not registered/qualified as research analysts with FINRA in the U.S. The Goldman Sachs Group, Inc. Global Investment Research

2 Contents What s driving China s rooftop solar demand positive on Singyes 3 Stock picks in brief; earnings and target price revisions 3 Getting ready to raise the roof: Distributed Generation comes of age 8 Three elements driving greater adoption of rooftop solar in China 9 Read across from overseas Rooftop PV 9 China is fine-tuning its business model 11 Mono-crystalline: Gaining traction on rooftop solar pickup 17 Favorable demand/supply outlook 17 Mono-crystalline is more cost competitive for distributed projects 18 Mono wafer has more cost reduction potential vs. multi 20 Mono to lead conversion efficiency 21 Mono has high industry entry barrier 22 Dissecting the different types of mono wafers N over P 24 Longi Silicon ( SS, Neutral): Robust earnings growth 25 Trina Solar (TSL, Neutral): Better positioned post US AD/CVD 37 Disclosure Appendix 42 Prices in this report are based on the market close of August 8, Exhibit 1: A-share solar companies generally trade at a premium against HK and US listed peers Valuation comparison of global solar companies Market Price Pricing Target cap P/E (X) P/B (X) EV/EBITDA (X) ROE (%) Net debt/ ROIC ROIC Companies Ticker Rating 8-Aug Currency Price (US$mn) 2014E 2015E 2014E 2015E 2014E 2015E 2014E 2015E equity (%) (%) (%) China offshore-listed Singyes Solar 0750.HK CL-Buy HKD , % 20% 20% 11% 12% GCL-Poly 3800.HK Buy 2.58 HKD , % 16% 156% 4% 6% Canadian Solar CSIQ Neutral USD , % 33% 102% 9% 11% Jinko Solar JKS Neutral USD % 19% 160% 12% 12% Trina TSL Neutral USD N/A N/A 8% 13% N/A N/A N/A Yingli YGE NC 3.36 USD NC 611 N/A N/A N/A -21% 24% N/A N/A N/A JA Solar JASO NC 8.77 USD NC N/A N/A N/A N/A 5% 2% N/A N/A N/A ReneSola SOL NC 2.61 USD NC 265 N/A N/A N/A -11% 23% N/A N/A N/A DAQO DQ NC USD NC 242 N/A N/A N/A N/A 0% 22% N/A N/A N/A Offshore-listed average % 19% 109% 9% 10% China onshore-listed Longi SS Neutral CNY , N/A N/A 8% 15% 1% 8% 14% Zhonghuan SZ NC CNY NC 3, N/A N/A 7% 11% N/A N/A N/A Akcome SZ NC CNY NC N/A N/A 10% 17% N/A N/A N/A Sungrow SZ NC CNY NC 1, N/A N/A 16% 16% N/A N/A N/A Hangtian SS NC 8.32 CNY NC 1, N/A N/A 9% 9% N/A N/A N/A Risen SZ NC 7.94 CNY NC N/A N/A 9% 12% N/A N/A N/A Jing Gong SZ NC 7.25 CNY NC N/A N/A 5% 14% N/A N/A N/A Jingsheng SZ NC CNY NC 1, N/A N/A 6% 10% N/A N/A N/A Almaden SZ NC CNY NC N/A N/A 5% 7% N/A N/A N/A Onshore-listed average N/A N/A 8% 12% 1% 8% 14% US SolarCity SCTY CL-Buy USD ,471 N/A N/A N/A N/A N/A N/A 14% -11% -6% First Solar FSLR Sell USD , % 9% -28% 7% 10% SunPower SPWR Neutral USD , % 18% -8% 8% 11% US average % 13% -7% 1% 5% Asia non-china OCI KS Neutral 175, KRW 150, ,659 N/A N/A N/A % 1% 68% 0% 2% Asia non-china average N/A N/A N/A % 1% 68% 0% 2% EU REC Silicon RECO NC 3.36 NOK NC 1,248 N/A N/A N/A N/A -1% 1% 23% N/A N/A EU average N/A N/A N/A N/A -1% 1% 23% N/A N/A Global average % 9% 39% 5% 8% NC=Not Covered, N/A=Not available Estimates for NC companies are Bloomberg consensus Our target prices are based on a 12-month timeframe Source: Bloomberg, Goldman Sachs Global Investment Research, Gao Hua Securities Research. Goldman Sachs Global Investment Research 2

3 What s driving China s rooftop solar demand positive on Singyes China demand picking up: installation CAGR of 51% According to the target announced by National Development and Reform Committee (NDRC) in May 2014, China is expected to have 70GW accumulated solar power installation by However, by the end of 2013, the actual installation was only 16.5GW, which means China needs to install at least 13GW each year of , with CAGR of 51% vs. our global solar installation estimates of 18% E CAGR. In our view, this will generate significant domestic demand growth for the solar EPC and high-quality solar components target increased: 10GW to go in 2H14, spurring demand The National Energy Administration (NEA) of China on Aug 4 announced 13GW solar installation for 2014 vs. the 10GW target announced at the beginning of this year. With actual completion of only 3.3GW in 1H14, this implies 10GW will need to be finished in 2H14, which we believe will lead to an inflection point in the next few months. Demand moves from utility-scale to distributed solar power By the end of 2013, there was only 5.7GW distributed (rooftop) projects installed in China, taking 35% of the total accumulated installation (including the utility-scale projects) vs. the government s target of 35GW/50% in As such, we expect rooftop solar projects will be the main demand driver of solar installations in coming years. EPC developers and mono product makers are key beneficiaries We expect solar project EPC developers will get more order visibility from the China demand s pick-up, especially from rooftop projects that carry higher ASP and margins than utility-scale projects. Meanwhile, given mono-crystalline products provide higher module conversion efficiency and lower unit installation costs in rooftop projects vs. the traditional multi products, we believe key beneficiaries of the pickup in rooftop solar are those companies exposed to all parts of the value chain predominantly EPC contractors such as Singyes and mono c-si wafer suppliers such as Longi. Mono demand to grow at 22% CAGR during Based on our global solar demand forecast and expectation on mono market share trend, we project mono wafer demand to grow from 10.4GW in 2013 to 18.6GW in 2016, implying a 22% CAGR, vs. 18% CAGR for overall solar demand growth. Stock picks in brief; earnings and target price revisions Singyes (CL-Buy): Singyes has strong presence in both rooftop and utility-scale EPC in Guangdong, Hunan and Gansu. Although limited new EPC orders were secured in 1H14, we expect the potential net metering policy rollout in 2H14 will enable Singyes to resume its order flows. A few rooftop PV investors have already stated their intention to sign rooftop EPC contract with Singyes and just awaiting the policy settlement. Longi (Initiate with Neutral): Longi is the world s largest mono-wafer manufacture with 12% global capacity market share in In 2013, 89% of revenue and 91% of gross profit was generated by its mono products (c-si ingot and wafer). At the end of 2013, Longi has 1.6GW wafer capacity and intends to ramp up to 4GW in The company has a broad customer base in Taiwan, Japan, Korea and US. We estimate 78% net profit CAGR over , driven by capacity expansion and cost reductions. Goldman Sachs Global Investment Research 3

4 GCL-Poly (Buy): GCL remains well positioned to capture the demand growth cycle for multi c-si wafer with its multi and Quasi Mono wafers (which have only marginally lower efficiency than Longi s p-type mono wafer). In addition, the company is also considering adding mono wafer capacity and seeking M&A opportunities in the segment. In 2014, GCL expect quasi mono and N-type mono to account for 10% and 2% of total sales. Canadian Solar, Trina and Jinko (all Neutral): Most downstream cell and module makers (including Canadian Solar, Trina and Jinko, etc) have limited exposure to mono wafer/cell manufacturing. However their existing production facilities can be switched to P-type mono cell and module production, but it would require additional investment and know-how to penetrate into N-type mono cell production. Therefore, we think downstream module players will eventually benefit from the rooftop PV demand cycle, but are not the best positioned at present. Earnings revisions: Singyes up, others down We increase our earnings estimates for Singyes by 9-11% for E, mainly to factor in our more positive view on its solar EPCs after its strategic cooperation with GCL and potential policy support for DG solar. Although we do not expect 1H14 earnings to be encouraging due to sluggish development of rooftop solar, we expect the gradual pickup in rooftop demand to drive a resumption of earnings growth from 2H14. We decrease our earnings estimates for solar component suppliers by between -8% and -21% across E, to factor in the recent price trend in solar components across the value chain (Exhibit 33 and 34). We think the preliminary tariff on anti-dumping and countervailing by the US Department of Commerce will have a broadly negative impact on module manufactures (such as Jinko, Canadian Solar, Trina) from 2H14. Although shipment growth will not be greatly affected due to resilient solar demand in the US, the module margins may face downside risk. For GCL, we lower our earnings estimate mainly to factor in a more conservative view on its captive power plant launch date. Exhibit 2: Poly price declined by 8% since March 2014 Polysilicon price performance Exhibit 3: Wafer/cell/module prices have declined by 10%/19%/8% since March 2014 Wafer, cell and module price performance (US$/kg) 24.0 (US$/W) 0.80 (US$/W) Polysilicon Cell Module Wafer (RHS) Source: PVinsights. Source: PVinsights. Goldman Sachs Global Investment Research 4

5 Exhibit 4: Earnings revisions for covered solar companies Solar companies Currency Current estimated earnings % change from previous YoY growth (%) (mn) 2014E 2015E 2016E 2014E 2015E 2016E 2014E 2015E 2016E Singyes Rmb % 9% 11% 39% 24% 12% GCL-Poly HK$ 1,861 2,837 4,235-8% -10% -10% 380% 52% 49% Longi Rmb N/A N/A N/A 271% 113% 48% Canadian Solar US$ % -18% -21% 396% 39% 35% Jinko US$ % -19% -17% 173% 9% 11% Trina US$ N/A N/A N/A 240% 36% 18% Source: Gao Hua Securities Research. Valuation: Returns based EV/GCI vs CROCI/WACC We continue to use Director s Cut (EV/GCI vs. CROCI/WACC) valuation framework for H- share and US listed solar companies as our primary methodology in valuing solar stocks. In this report, we initiate coverage on Longi and Trina. Given Longi is A-share listed, and thus has a different cost of capital and investor base from offshore peers, we use a 2015E P/B methodology using Longi s historical peak multiple to reflect the company s ROE improving into the mid-teens. We apply a FY15E P/B of 3.0X. This gives a 12-month target price of Rmb20.2, which implies 40x/19x 2014/15 P/E, vs its mid cycle of 30x. We cross check our FY15E P/B valuation with DCF (see company snapshot for details) which generates a value of Rmb We also plot A-share peers under 2015E P/B-ROE for reference, using Wind consensus estimates for peers. Exhibit 5: Longi has traded at an average 120% valratio premium against our covered solar stock since Feb 2014 EV/GCI vs. CROCI/WACC premium against sector average Exhibit 6: Longi has traded at a 38% forward P/B premium against our covered solar stocks on average over the past two years Forward P/B premium of Longi vs. covered solar stocks EV/GCI to CROCI/WACC (X) Feb-14 Mar-14 Apr-14 May-14 Jun-14 Longi Sector average P/B (X) Apr-12 Jul-12 Oct-12 Jan-13 Apr-13 Jul-13 Oct-13 Jan-14 Apr-14 Jul-14 Longi Sector average Goldman Sachs Global Investment Research 5

6 Exhibit 7: EV/GCI vs. CROCI/WACC valuation 2015E Exhibit 8: P/B vs. ROE plot vs A-share peers, for reference 2015E 2015E EV/GCI 2.5x 2.0x 1.5x 1.0x 0.5x y = x R² = First Solar OCI Trina Longi Jinko GCL-Poly Sunpower Canadian Solar Singyes (X) 2015E Price-to-book y = x R² = Almaden Zhonghuan Jingsheng Jing Gong Hangtian Risen Longi Sungrow Akcome x 0.0x 0.5x 1.0x 1.5x 2.0x 2.5x 3.0x 2015E CROCI/WACC Note: The red dots denote our China Clean Energy coverage. Source: Company data, Goldman Sachs Global Investment Research, Gao Hua Securities Research % 5% 10% 15% 20% 2015E Return on Equity Note: We use Wind consensus numbers for the companies except Longi in this chart. Source: Bloomberg, Company data, Wind, Gao Hua Securities Research. Exhibit 9: Singyes ranks in the first quartile of our solar coverage CROCI ranking CROCI E 2015E 2016E Singyes Solar 18% 38% 26% 33% 25% 12% 22% 24% 24% GCL-Poly 345% 17% 37% 12% 9% 17% 16% 16% 17% Longi N/A 14% 117% 13% 7% -8% 1% 13% 13% Trina Solar 11% 28% 38% 10% -7% 0% 5% 8% 9% Canadian Solar -3% 32% 5% 17% -14% 21% 13% 20% 23% Jinko Solar 1% 0% 42% 8% -3% 10% 18% 18% 19% CROCI ranking Singyes Solar GCL-Poly Longi N/A Trina Solar Canadian Solar Jinko Solar First quartile 2 Second quartile 3 Third quartile 4 Forth quartile Goldman Sachs Global Investment Research 6

7 Exhibit 10: EV/GCI vs. CROCI/WACC target price valuation Singyes and GCL offer double digit return potential Period indicator HK$ HK$ US$ US$ US$ Singyes GCL-Poly Canadian Solar Jinko Trina 0750.HK 3800.HK CSIQ JKS TSL CROCI (2011) 32.9% 12.3% 16.5% 8.4% 10.3% CROCI (2012) 24.8% 9.3% -13.9% -3.4% -6.8% CROCI (2013) 11.8% 17.0% 21.4% 9.5% 0.2% CROCI (2014E) 21.5% 16.1% 12.7% 18.1% 5.3% CROCI (2015E) 24.2% 15.6% 20.2% 18.3% 8.0% CROCI (2015E) 24.2% 15.6% 20.2% 18.3% 8.0% WACC 10% 10% 10% 10% 10% CROCI/WACC (X) 2.42x 1.56x 2.02x 1.83x 0.80x Sector Valratio 0.91x 0.91x 0.91x 0.91x 0.91x Valuation premium/discount adjustment Adjusted Valratio (Y) 0.91x 0.91x 0.91x 0.91x 0.91x Target EV/GCI (=X*Y) 2.19x 1.41x 1.83x 1.66x 0.72x GCI (pricing currency mn, 2015E) 5,536 59,024 1,243 1,145 2,453 Target EV (pricing currency mn) 12,139 83,506 2,275 1,902 1,777 Less: Net debt (2015E) (1,037) (30,672) (669) (981) (780) Less: Off-balance sheet debt (2015E) Less: Minority interest (2015E) (0) (1,859) (11) (2) - Implied equity value (pricing currency mn) 11,101 50,975 1, Number of shares (mn) , Director's cut implied equity value HKD HKD 3.30 USD USD USD Previous TP HKD HKD 3.60 USD USD N/A Ratings CL-Buy Buy Neutral Neutral Neutral TP change 11% -8% -26% -17% N/A Current price HKD HKD 2.58 USD USD USD Up/downside 32% 28% 5% 8% 7% Note: Current prices as of Aug 8, Our previous val ratio was Exhibit 11: Our rating and target prices on solar coverage Solar companies Ticker Rating Currency TP %chg Current Potential up/downside Methodology Singyes Solar 0750.HK CL-Buy HK$ % % Directors' Cut GCL-Poly Energy Holdings 3800.HK Buy HK$ % % Directors' Cut Longi Silicon SS Neutral Rmb N/A % P/B vs. ROE Trina Solar TSL Neutral US$ N/A % Directors' Cut Canadian Solar Inc CSIQ Neutral US$ % % Directors' Cut JinkoSolar Holding Co. JKS Neutral US$ % % Directors' Cut Key risks Singyes: Lower-than-expected EPC orders growth; Higher-than-expected project installation cost. GCL-Poly: Slower-than-expected captive power plant and FBR method delivery; Lowerthan-expected polysilicon/wafer ASP. Longi: High-than-expected mono wafer demand and faster-than-expected cost reduction (upside); Competition from new players, lower-than-expected wafer selling price and lack of funding sources (downside). Goldman Sachs Global Investment Research 7

8 Trina: Faster-than-expected downstream solar project development (upside); Lower-thanexpected margin outlook due to the AD/CVD tariff charge in US (downside). Jinko Solar: Faster-than-expected solar yieldco company spin-off (upside); Lower-thanexpected module shipment and margins (downside). Canadian Solar: Faster-than-expected total solution project delivery (upside); Lower-thanexpected module shipment and margins (downside). Getting ready to raise the roof: Distributed Generation comes of age We see China s rooftop (ie, distributed generation) solar industry at an inflection point poised to grow rapidly in coming years, driven by growing business model maturity and policy support. Key developments: (1) a better financing environment, such as power generation insurance, launch of yieldco financing platforms and growing interest from capital markets; (2) Crystallization of investment returns through potential introduction of net metering; (3) Alignment of interests among value chain participants with State Grid and property management firms now participating in PV projects investment. Definition of distributed generation solar Distributed generation (DG) refers to electricity that is produced at or near the point where it is used. Distributed solar energy is normally located on the rooftops of commercial, industrial and residential buildings. Therefore, we refer to rooftop solar as DG in this report. There are quite a few differences in business model between rooftop and utility-scale solar projects in China, as illustrated in Exhibit 12. In addition, the tariff mechanism is also quite different. For rooftop solar, power user receives Rmb0.42/kWh subsidy for self-used power and thermal power grid plus Rmb0.42/kWh subsidy for remaining power sold to the State Grid (on-grid). For utility-scale projects, solar project owner receives Rmb /kWh feed-in- Tariff from State Grid, depending on geographic locations. Exhibit 12: The business model of rooftop solar is generally more complicated than utilityscale project since rooftop power can be used or sold to the grid Comparison of rooftop and utility-scale PV projects Utility scale Rooftop Early preparation Land procurement, grid connection, etc. Rooftop lease, grid connection, secure stage Relatively standardized end users, higher uncertainties Financing Since power generated are sold to State Smaller scale, complicated operation and Grid, financing is easy user demand uncertainties lead to difficult financing environment Construction stage Construction work is quite standardized Generally requires tailor made engineering based on each building's feature Operation stage Limited maintainance works Operating works includes safety checks on rooftop and power sales, etc Electricity fee State Grid settles the basic thermal Individual customers may delay the fee payment collection power tariff monthly while subsidity part Changes in end user may also cause defalt settlement is delayed for a few months risk in fee collection Asset transfer Involves shareholder and State Grid Involves rooftop owner, end user, State Grid EMC, etc Goldman Sachs Global Investment Research 8

9 Three elements driving greater adoption of rooftop solar in China (1) Curtailment of utility-scale projects in some provinces is emerging, such as Gansu, because power network construction is lagging the renewable power construction. As a majority of the electricity generated by rooftop PV is used by the end user directly, it significantly lowers the power network transmission investment and delivery cost, and reduces the power lost over the wires. (2) It requires less government subsidy than utility-scale projects. We find the average subsidy on utility-scale projects (Feed-in-tariff minus thermal power on-grid tariff) is around Rmb0.588/kWh, vs. Rmb0.42/kWh uniformed subsidy on rooftop PV system (Exhibit 13). In other terms, with the same amount of subsidy, rooftop PV can support 40% additional solar installations than utility-scale projects. (3) Certain provinces in China are approaching grid-parity for rooftop PV. Grid parity refers to the unit power generation cost from PV (ie levered cost of electricity or LCOE) than the local retail electricity price. Based on 1,200 insolation hours and Rmb9.0/watt investment cost, we estimate the LCOE for rooftop PV in China is around Rmb0.9/kWh, vs. average power tariff of Rmb0.905/kWh in Zhejiang for commercial users. In other words, even without subsidy, the current power price can cover solar power costs. Exhibit 13: The subsidy on utility-scale PV projects is higher than for rooftop projects Tariff subsidy calculation by province for utility-scale projects by province (Rmb/kwh) Rmb0.42/KWh - Xinjiang II Xinjiang I Guizhou Gansu II Inner Mongolia II Ningxia Shanxi III Shaanxi III Qinghai II Inner Mongolia I Yunnan Hebei III Gansu I Anhui Henan Shanxi II Chongqing Jiangsu Fujian Shandong Shaanxi II Beijing Qinghai I Tianjin Jilin Heilongjiang Liaoning Shanghai Guangxi Zhejiang Hubei Hebei II Jiangxi Hainan Hunan Sichuan Guangdong Source: NDRC, Goldman Sachs Global Investment Research Read across from overseas Rooftop PV According to EPIA (European Photovoltaic Industry Association), rooftop PV installation in 2013 amounted to 23GW, which represents 60% of total global PV installation. In Europe, utility-scale (ground mounted) projects were 34% of total PV installation with rooftop 66% including commercial (27%), industrial (17%) and residential (22%). The residential PV has developed very rapidly in some countries, such as Belgium, where 1 out of 13 households are now equipped with a PV system. Goldman Sachs Global Investment Research 9

10 In the US, residential rooftop, industrial & commercial (C&I) rooftop, and utilityscale projects accounts for 15%, 25% and 60% of PV installations in 2013, according to SEIA. In Japan, rooftop represents 73% of total installation in 2013, based on the data from JPEA. China has ample rooftop to install distributed generation PV system, according to Chinese Renewable Energy Industries Association, the total rooftops available for installation amounts to 300GW in China, 80GW of which is from industrial parks. Exhibit 14: Rooftop accounts for 60% of Global PV installation in 2013 Rooftop s market share by country Exhibit 15: Rooftop accounts for 66% of Europe s PV installation in 2013 Cumulative PV capacity by country in % 70% 60% 50% 40% 30% 20% 10% 0% 27% 40% 66% 73% 60% China US Europe Japan Global average 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% Rooftop as % of total PV installations Source: EPIA, SEIA, JPEA, Goldman Sachs Global Investment Research. Residential Commercial Industrial Ground mounted Source: EPIA, Goldman Sachs Global Investment Research. Exhibit 16: Rooftop accounts for 40% of US s PV installations in 2013 Quarterly PV installation by type in US Exhibit 17: Rooftop accounts for 73% of Japan s solar installation in 2013 Quarterly PV installation by type in Japan Q10 2Q10 3Q10 4Q10 1Q11 2Q11 3Q11 4Q11 1Q12 2Q12 3Q12 4Q12 1Q14 Residential Non residential Utility (MW) 3,000 2,500 2,000 1,500 1, Q12 2Q12 3Q12 4Q12 1Q13 2Q13 3Q13 4Q13 Residential Non-residential Power generation Source: SEIA, Goldman Sachs Global Investment Research. Source: JPEA, Goldman Sachs Global Investment Research. Goldman Sachs Global Investment Research 10

11 China is fine-tuning its business model The bottom line from our analysis of the current situation is that recent developments have lifted our confidence on the industry outlook and lead us think rooftop PV is at the inflection point to register rapid growth in We believe the slow development of rooftop PV ytd in China is due to its immature business model. Although a number of supporting policies have been announced since Aug 2013, execution is generally slow. We believe this is mainly because certain issues are still to be resolved, but that progress is imminent or underway: Credit risk concerns from banks. Conflicts of interest between PV owners and State Grid. Return volatility on PV investment. We note policymakers are fine tuning policy details using feedback from participants during the real operations. Improving financing environment comparison with established markets In China, the financing channel for PV projects is mainly bank loans. Since it is generally difficult for residential or small and medium enterprises (SMEs) to secure loans from banks for PV systems due to the lack of pledge and a credit rating system, the possibility to obtain loans for rooftop PV directly from banks is limited. Although supporting policies related to financing have been announced in 4Q13, execution is lagging. Within the 18 national distributed solar zones assigned by NEA, only 1 has secured a loan from the China Development Bank (CDB) so far. In addition, the current interest rate for loans with over 5 years maturity is 6.55% in China and we understand solar companies actual borrowing costs are around 7%. Although some solar companies have tried finance leasing to lower the initial capital outlay, their realized cost of funding generally is over 8%, which is still high compared to the solar project s average unleveraged IRR of 10%-15% in China. In the US, third-party financing model such as Power Purchase Agreements (PPA) and Solar Leasing are widely adopted for financing and lower the cost of capital, which fuels the development of rooftop solar in both C&I and residential market. In the PPA model, the solar developer builds the PV system and sells power generated to the customer at a fixed rate, typically lower than the local utility. In the lease model, a customer signs a contract with developer and pay for the solar system over a period of years, rather than paying for the power produced. To realize a low cost capital, developers often partner with tax-equity investors due to federal solar incentives to formulate some financing structures such as sale-lease back, partnership flip and inverted lease, etc. Our US solar team estimates the actual cost of capital for tax equity financing structure for Solarcity is less than 3%. Also, Solarcity also issued asset-backed securities (ABS) with 4.8% interest rate in early In Germany, the government encourages the financing of rooftop PV projects. The federally owned KfW Banking group is facilitating renewable energy investment for individual and community groups and it financed over 50% of its PV installations for the country in KfW provides financings to the end renewable users through regular commercial banks. For household user, up to Euro 50,000 can be financed with KfW loans and the cost of capital is around 4%. In Japan, most of the PV projects financings are from bank syndicates for term loans. The government owned Japan Finance Corporation (JFC) is offering up to 270mn yen loans to SMEs for energy conservation facility installations over a 15 year period. Also, the funding cost in Japan is very low, with 10- year Japanese government bond rate currently at 0.544%. Goldman Sachs Global Investment Research 11

12 Exhibit 18: US PPA and finance lease model Lease payment Bill payment (Interest + principal) Customers Solar project Investor operator (Leasee) (leaser) Electricity Sells the system provision and lease back Source: Gao Hua Securities Research. Exhibit 19: Federal bank KfW in Germany indirectly finance the rooftop PV investment for residential and C&I users Business model of KfW bank in loan approval process Customer Files loan application with regular bank before start of investment Enters into the loan agreement and disburses the loan Customer s bank If approved, customer s bank forwards application to KfW Refinances the loan at favorable refinancing interest rates KfW Capital market Promotional mandate: Defined by KfW law State liability: Institution owned by Federal Republic and federal states Backed by Understanding II reached with EU Commission Source: Gao Hua Securities Research. Based on these overseas cases, we find government plays a major role in directly or indirectly providing capex financing abroad. Also assisting in financing and funding cost issues, we note: Capital markets are becoming more supportive of PV investment. A number of listed solar project operators such as United Photovoltaics (686.hk), Shunfeng International (1165.hk), and Jiangsu Akcome ( hk) among others, have raised capital from equity investors in the past 6 months. Those companies all have plans to become Energy Manager Contracting (EMC) providers for rooftop PV (similar to the PPA model in US), offering customers to secure financing and offers expertise in PV system development and operation. We believe companies with strong balance sheet or strong parent support in terms of loan guarantees are likely to be winners in the EMC space. Yieldco spin-off is a potential way to secure low-cost funding. Yieldco investors generally require high earnings payout ratio and sustainable dividend per share growth, which requires Yieldco to have a solid projects pipeline to continuously invest and finance growth. Jinko and Canadian Solar have both indicated an interest in spinning-off a yieldco. Jinko targets to have 800MW PV projects installed by 2014 and plans for the spin-off in Canadian Solar also considers a spin-off once it has 500MW self-operated PV projects. It targets to have 320MW finished at the end of For detailed analysis on Yieldco, please see our 6 Aug 2014 report, Spotlight on Yieldco potential and implications for value creation in global Clean Energy. Goldman Sachs Global Investment Research 12

13 Exhibit 20: Some Chinese solar firms have raised capital in the past 6 months Fund raising activities for major solar companies Company Ticker Source: Company data. Financing method Amount (Rmb mn) Announcement date Financing purpose Longi SS Equity 1,960 Jul-2014 Mono wafer capacity expansion Akcome SZ Equity 973 Jul-2014 Solar project development Hareon SS Equity 3,802 Jun-2014 Solar project development; Wafer/cell capacity expansion Zhongli SZ Banking loans 5,000 Jun-2014 Solar project development Singyes 750.HK Convertible bonds 930 Jul-2014 Solar project development Shunfeng 1165.HK Convertible bonds 1,667 Jun-2014 Solar project development Banking loans 2,000 Jul-2014 United Photovoltaics 686.HK Equity 631 Jan-2014 Solar project development Passing power generation and system quality risk to insurance companies lifts investor/banks confidence in financing PV systems. We believe banks cautious view on PV loans is mainly due to the risks associated with module underperformance in hot climates, panel theft and defunct warranties from manufacturers entering bankruptcy, etc. In China, PV module and component quality insurance is generally offered, but there was no power generation insurance, until June 2014, Anbang Insurance launched the first casualty product that guarantees the power generation for solar, which we believe may help ease some concern on investment risk. Exhibit 21: EMC company plays the key role in the solar distributed project Role of EMC company in rooftop PV value chain Bank Provide the loans Pay back the loans Solar project developer Power sales Power users Investment Revenue Power income Subsidy application Solar project EMC operator Power sales Power income and subsidies Rent the land/rooftop Rental income Land/rooftop owner Government s finance department Subsidies Grid company Source: Gao Hua Securities Research. Investment return crystallization through potential introduction of a net metering mechanism. Based on the current subsidy and tariff structure in China, the benefit for selfused (off-grid) usage equals to the saving on retail power tariff plus the Rmb0.42/kWh subsidy, which ranges from Rmb /kWh in different province. By applying 30% discount to the retail power tariff, which is normally the actual selling price offered by EMC company to end user, the range becomes Rmb /kWh. On the other hand, the benefit for on-grid sale amounts to thermal power on-grid tariff plus Rmb0.42/kWh subsidy, which results in a range of Rmb /kWh. That said, the return on a rooftop PV system Goldman Sachs Global Investment Research 13

14 depends largely on the utilization split between on-grid and off-grid. There are various issues that could affect the end power usage for a typical power user, such as seasonality and business operations, etc. Our sensitivity analysis shows for a typical commercial rooftop project in Jiangsu under the EMC model, the unlevered IRR is 9.0% with 90% offgrid consumption, vs. 6.4% for 30% off-grid utilization. In the US, Net Metering system is widely used for rooftop PV, which is a billing mechanism that credits PV system owners for the electricity they add to the grid at the retail electricity price in most states. This essentially makes whoever actually owns the PV irrelevant to the power usage between on-grid and off-grid. To ensure the returns for EMC operator, the NEA is also currently unifying the benefits between on-grid and off-grid. We believe the potential rollout of net metering mechanism in China would be a big positive for the rooftop PV development. Exhibit 22: Rooftop solar project net metering system in US Rooftop solar project One-way metering system Total solar power generation Project owner (Power user) Net metering system Surplus is sold to the grid Grid power Shortage is purchased from the grid Source: Gao Hua Securities Research. Goldman Sachs Global Investment Research 14

15 Exhibit 23: We highlight customer categories in each province that EMC companies have an incentive to sell power to end users Comparison of end power tariff and on-grid thermal power tariff blue indicates a financial incentive (Rmb/kWh) 30% discout on end-user power tariffs On grid thermal Residential Commercial Industrial power tariff Beijing Tianjin North Hebei South Hebei Shanxi Shandong West Inner Mongolia Shanghai Zhejiang Jiangsu Anhui Fujian Hubei Hunan Henan Jiangxi Sichuan Chongqing Shaanxi Gansu Qinghai Ningxia Xinjiang East Inner Mongolia Heilongjiang Jilin Liaoning Guangdong Guangxi Yunnan Guizhou Hainan Average Source: Gao Hua Securities Research. Interest alignment among value chain participants. The development of rooftop PV system involves collaboration between various participants (Exhibit 21), including power network operator (State Grid), building owner, PV system supplier, developer and operators, government, financing entities (banks and investors) as well as insurance companies, etc. Aligning interests among the various parties is critical to ensure the development of rooftop PV in China. As majority of the power generated by rooftop PV is consumed by end-users directly, distributed PV essentially lowers power purchases from the State Grid and leads to a conflict of interest for the State Grid to support the development of rooftop PV in China. In essence, solar power is challenging the business model of conventional utilities. On the other hand, in order to receive the Rmb0.42/kWh subsidy, the policy states rooftop PV system needs to be connected to the grid even that 100% power is self-consumed. In addition, electricity fee collection is not an easy task and rooftop PV operators lack Goldman Sachs Global Investment Research 15

16 measures to prevent customers defaulting or payment delays. Also, as more PV installations are developed and integrated into the grid, utility companies may face grid stability issues as most of the power generated from PV system is in the daytime. The US also face conflict of interests between utilities and solar power. Utilities are required by state government to buy the power that a rooftop system produces but does not use. However, In Oklahoma State, legislature permits utilities to charge user fees on customers with renewable energy installations, partly to cover the infrastructure cost that utilities provides to PV system users. Possible ways to reduce conflict of interest We believe to introduce State Grid as a shareholder on the rooftop PV system could essentially solve conflict of interest issues and electricity fee collection risk. For example, Jiangsu Linyang Electronics partners with State Grid in jointly developing the rooftop PV in Jiangsu. State Grid has energy conservation department that focuses on clean energy investment. Each year, the department has an investment/development target in renewable energy. Moreover, partnering with property management companies in electricity fee collection is also a feasible approach that is being adopted by some rooftop PV operators nowadays. Goldman Sachs Global Investment Research 16

17 Mono-crystalline: Gaining traction on rooftop solar pickup We think China s accelerated distributed solar installation will benefit the mono-crystalline solar products most in the next few years, mainly because of the better cost-efficiency of mono vs. multi products where space is a consideration. Favorable demand/supply outlook Mono sets to gain market share in total solar demand Multi c-si has gained market share in the past few years due to efficiency improvement and cost reduction. (Exhibit 28) While average multi c-si ingot sizes are likely to continue to grow, the conversion efficiency gain may start to slow as crystallization improvement are widely realized. However, we believe the rooftop solar demand pickup, ongoing efficiency gains and cost reduction will enable mono c-si to regain market share from 2H2014 onward. Solarbuzz expects mono c-si will capture 35% market share in silicon wafer sector, vs. 29% in We discuss these dynamics below. Exhibit 24: We expect mono wafer to gain market share against multi in the coming four years Mono vs. Multi Silicon wafer market share 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0% 39% 36% 61% 64% 29% 29% 30% 31% 33% 35% 71% 71% 70% 69% 67% 65% E 2015E 2016E 2017E Multi Mono Exhibit 25: c-si is still the main stream PV technology 2014 Solar PV module production by technology c-si p-type Mono Advanced, 10% CdTe (First Solar), 4% c-si n-type Mono, 6% c-si p-type Mono Standard, 14% CIS/CIGS Sputter, 2% c-si p-type Multi Advanced, 27% a-si Glass/Glass, 2% c-si p-type Multi standard, 35% Source: Solarbuzz, Gao Hua Securities Research. Source: Solarbuzz, Gao Hua Securities Research. Mono demand to grow at 22% CAGR during Based on our global solar demand forecast and expectation on mono market share trend, we project mono wafer demand to grow from 10.4GW in 2013 to 18.6GW in 2016, implying a 22% CAGR, vs. 18% CAGR for overall solar demand growth. Mono supply to remain tight during We see growing supply concentration for mono wafer supply. We estimate China controls 67% global mono wafer capacity in 2013 and for this to increase to 74% in Based on each company s capacity expansion plan (including Longi, Zhonghuan, Cometc), we expect mono wafer supply will remain tight during and we expect demand/capacity ratio will stay at the 100% level in the next 3 years, unless we see multi-wafer suppliers such as GCL-Poly entering the market. But given that banks remain cautious on financings solar component makers, we think the tight credit environment and heavy capex required from switching multi to mono capacity will also constrain new supply in mono wafer sector. Therefore, we think the demand/supply dynamics will be favorable for mono wafer, thus benefiting the pricing trend for mono wafer suppliers. Goldman Sachs Global Investment Research 17

18 Exhibit 26: We expect mono wafer supply will remain tight during , which bodes well for the pricing outlook Mono wafer demand/supply model Mono-wafer capacity (MW) E 2015E 2016E China Longi 600 1,000 1,500 2,000 3,000 3,500 4,000 Zhonghuan 500 1,000 1,500 1,800 2,300 2,800 2,800 Comtech ,400 1,700 1,700 Solargiga Energy Jing Long Others ,000 1,500 3,500 China's total capacity 2,800 4,600 5,600 6,800 9,200 11,000 13,500 Yoy 64% 22% 21% 35% 20% 23% Other regions' capacity 2,327 3,368 3,232 3,352 3,472 4,192 4,812 Yoy 45% -4% 4% 4% 21% 15% Global total capacity 5,127 7,968 8,832 10,152 12,672 15,192 18,312 Yoy 55% 11% 15% 25% 20% 21% Global mono wafer demand 8,208 10,050 8,722 10,365 12,746 15,238 18,614 Yoy 22% -13% 19% 23% 20% 22% Demand/capacity ratio 160% 126% 99% 102% 101% 100% 102% Mono-crystalline is more cost competitive for distributed projects We believe the balance of system (BOS) contribution to the total installation cost and unit price premium against multi c-si as the key factors in gauging the cost competitiveness of mono PV systems. Our analysis shows mono is already more cost-competitive in the rooftop solar segment. Mono has a lower BOS cost than multi, within the same power output. Solar project installation cost contains two parts, module and BOS. BOS cost mainly includes the cost of land, labor, inverter, framework, etc. BOS has become a more important cost factor for solar installation in the past few years. As the module price has decreased c.57% in the past three years and the BOS cost was relatively steady, the BOS proportion of total installation cost increased from 40% in 2011 to 50%-60%. As mono systems have greater efficiency than multi systems it requires less land, supporting structure, cables and labor than multi PV system to generate the same amount of power output. The price premium of mono against multi has declined recently. As illustrated in Exhibit 28, the per watt price premium of mono module against multi module has declined from 22% in early 2013 to 10%, making mono c-si a cost competitive choice for PV system developers. We believe it is partly due to the know-how improvement such as efficiency improvement, the introduction of diamond wire sawing and increasing ingot pulling speed, etc. In addition, we believe mono has higher potential to cut cost than multi module (see below). Goldman Sachs Global Investment Research 18

19 Exhibit 27: BOS cost over total installation cost has increased in the past few years BOS and module cost breakdown for solar installation Exhibit 28: Price premium of mono module has declined from 22% in early 2013 to the current 10% Unit selling price difference between mono and multi module 100% 90% (US$/w) 25% 20% 80% 70% 60% 68% 64% 60% 49% 44% % 10% 50% % 40% 30% 20% 10% 0% 51% 56% 32% 36% 40% BOS cost Module cost Jan-13 Feb-13 Mar-13 Apr-13 May-13 Jun-13 Jul-13 Aug-13 Sep-13 Oct-13 Nov-13 Dec-13 Jan-14 Feb-14 Mar-14 Apr-14 May-14 Jun-14 Jul-14 Multi c-si module (LHS) Mono c-si module (LHS) 0% -5% -10% Premium (RHS) Source: PV News, Gao Hua Securities Research. China s solar installation is mainly driven by utility-scale projects located in the western provinces ytd, where site area constraints and land cost are not major concerns for project developer (Exhibit 15). With the growing adoption of rooftop solar in eastern areas, however, where limited installation space would lift the BOS cost a percentage of total investment cost, mono modules has advantages over multi modules in terms of unit investment cost, IRR and LCOE. Exhibit 29: Our sensitivity test indicates that mono c-si module offers lower installation cost if (1) the unit selling price premium is below 15% and (2) BOS represent over 50% of the total installation cost Multi vs. Mono module installation cost difference ASP premium of mono vs. multi (%) Source: Gao Hua Securities Research. BOS cost as percentage of total installation cost (%) 45% 50% 55% 60% 65% 70% 75% 0% % % % (0.02) % (0.41) (0.35) (0.29) (0.22) (0.16) (0.10) (0.04) 30% (0.61) (0.55) (0.48) (0.42) (0.36) (0.30) (0.23) 35% (0.80) (0.74) (0.68) (0.62) (0.55) (0.49) (0.43) We provide case studies to illustrate the unit installation cost comparison between multi and mono c-si modules in utility-scale and rooftop project respectively. We find the cost difference between multi and mono module is quite limited in utility-scale projects but that mono module is more cost competitive in rooftop projects (Exhibit 30 and 31). Goldman Sachs Global Investment Research 19

20 Exhibit 30: The difference of unit project installation cost between multi and mono module is limited for utilityscale project Utility-scale solar project installation cost comparison Exhibit 31: while mono products have lower unit installation cost in the distributed projects Distributed solar project installation cost comparison Utility-scale project Mono Multi Project capacity (MW) Land area (sqm) 10,000,000 10,000,000 Land cost (Rmb/sqm) Power per panel (W) Module price (Rmb/W) Module cost (Rmb mn) BOS cost (Rmb mn) Land cost (Rmb mn) Inverter cost (Rmb mn) Mounting structure cost (Rmb mn) Labor cost (Rmb mn) Other BOS cost (Rmb mn) Total installation cost (Rmb mn) Unit installation cost (Rmb/W) Source: Gao Hua Securities Research. Distributed project Mono Multi Rooftop area (sqm) Amount of solar panels Power per panel (W) Project capacity (KW) Module price (Rmb/W) Module cost (Rmb) 147, ,100 BOS cost (Rmb) 174, ,900 - Land cost (Rmb) 72,360 72,360 - Inverter cost (Rmb) 32,562 32,562 - Mounting structure cost (Rmb) 32,304 36,180 - Labor cost (Rmb) 24,228 27,135 - Other BOS cost (Rmb) 12,663 12,663 Total installation cost (Rmb) 321, ,000 Unit installation cost (Rmb/W) Source: Gao Hua Securities Research. Exhibit 32: Distributed projects using mono modules could have high IRR Solar projects equity IRR comparison Exhibit 33: Mono modules can also bring down the LCOE for distributed projects Solar project LCOE comparison 20.0% 18.0% 16.0% 14.0% 12.0% 10.0% 8.0% 6.0% 4.0% 2.0% 0.0% 15.6% Utility-scale projects (Mono) Note: Key assumptions: 17.1% Utility-scale projects (Multi) Source: Gao Hua Securities Research. 13.5% Distributed projects (Mono) 12.5% Distributed projects (Multi) (Rmb/kWh) Utility-scale projects (Mono) Note: Key assumptions: 0.67 Utility-scale projects (Multi) Source: Gao Hua Securities Research Distributed projects (Mono) 0.75 Distributed projects (Multi) Mono wafer has more cost reduction potential vs. multi The production cost gap between mono and multi wafer has narrowed significantly. Wafer production cost includes silicon cost and non-silicon cost. Mono and multi wafer have similar silicon cost, with the main difference coming from the non-silicon part. Ingot cost and cutting cost take up the greater part of the non-silicon cost. By adopting the new technologies in ingot furnace (e.g. larger ingot) and cutting (e.g. diamond wire), Longi s mono wafer s non-silicon cost has declined 42% from 2011 to The cost gap between mono and multi wafer has narrowed from $0.12/W in 2009 to $0.02/W in 2013 (Exhibit 34). Goldman Sachs Global Investment Research 20

21 There s still room for more. As mono wafers have better distributed crystalline structure, its cutting yield rate is higher than multi (95% vs. 90%). Therefore mono wafers have lower per watt cutting cost. The main difference comes from ingot cost: mono wafer s current ingot cost is around US$15/kg (US$0.06/W) vs. multi wafer s US$5/kg (US$0.03/W). By making larger ingots, we believe the cost of mono ingots will near the cost of multi ingots in the next 2-3 years. On the other hand multi ingot cost has limited downside potential from current US$5/kg (Exhibit 35). Also, due to the crystal distribution difference, diamond wire sawing technology is more suitable to apply in mono wafer cutting than in multi, which could further allow mono wafer to gain cost advantage in the cutting segment. Exhibit 34: The production cost gap between mono and multi wafer has narrowed significantly Industry average wafer non-silicon cost Exhibit 35: There s still large room for mono wafer s cost reduction by reducing ingot cost Mono wafer non-silicon cost breakdown (US$/W) Mono Multi 0.12 (US$/W) E 2016E Multi (2014E) Ingot cost Cutting cost Mono to lead conversion efficiency As mono wafers offer higher conversion efficiency, it better fits the development strategy of high efficiency cell/module. High efficiency solar panel development is the prevailing trend for major PV system developers nowadays. Solarcity acquired high-efficiency cell/module maker Silevo in June SolarWorld acquired Bosch s high performance cell plant in Germany in Nov According to the solar companies announced technology roadmap, we think mono will continue to maintain its leadership efficiency in the foreseeable future. This is evidenced by SunPower s guidance of 23% of mono module efficiency in 2015, which is 5.5% higher than the highest guided multi module efficiency (Exhibit 36). Meanwhile, our sensitivity analysis suggests that a 2% of mono module conversion efficiency could bring down the installation cost by 6% (Exhibit 37). Therefore, we believe mono s leading position in conversion efficiency will help it to gain more market share in the next few years. Goldman Sachs Global Investment Research 21

22 Exhibit 36: Mono technology will strengthen its leading position in conversion efficiency vs. multi in the next few years Module conversion efficiency roadmap by companies Reported module efficiency Company guidance Technology E 2015E 2016E SPWR Mono 19.0% 20.0% 21.5% 22.0% 22.8% 23.0% N/A CSIQ Multi 15.5% 16.0% 16.5% 16.7% 17.0% 17.5% 18.0% Mono 17.0% 17.3% 17.5% 17.7% 18.0% 18.5% 19.0% YGE* Multi 14.8% 15.0% 15.3% 15.5% 16.0% 16.4% 16.8% Mono 16.5% 17.0% 17.3% 17.7% 18.1% 18.5% 18.9% TSL* Multi N/A N/A 15.4% 15.8% 16.2% N/A N/A Mono N/A N/A 17.0% 17.3% 17.5% N/A N/A JKS* Multi 14.8% 15.3% 15.6% 16.0% 16.5% N/A N/A Mono 16.0% 16.5% 17.0% 17.4% 18.0% N/A N/A JASO* Multi 14.2% 14.8% 15.3% 15.9% 16.2% N/A N/A Mono 15.6% 16.2% 16.8% 17.4% 18.0% N/A N/A SOL* Multi 14.5% 14.9% 15.3% 15.7% 16.0% N/A N/A Mono 15.5% 16.0% 16.5% 17.0% 17.5% N/A N/A Note: * Companies only provide conversion efficiency at the cell level; we assume a 2% efficiency loss at the module level. Exhibit 37: Our sensitivity analysis suggests 2% of mono module conversion efficiency change could bring 6% installation cost reduction for the solar project Sensitivity analysis on mono module conversion efficiency Solar project installation cost change (%) Note: Key assumptions: Mono module conversion efficiency (%) 15% 16% 17% 18% 20% 22% 24% 11% 7% 3% 0% -6% -10% -14% Source: Gao Hua Securities Research. Mono has high industry entry barrier Technical know-how. As the production process for solar-grade mono c-si ingot is generally similar to the semiconductor grade, mono c-si wafer makers normally have a background in the semiconductor industry, such as Zhonghuan, Longi, Comtec. This sets a high entry barrier for non-semiconductor players to enter the market and become a qualified supplier for downstream high efficiency cell and module makers. For mono cell production, the high-efficiency mono cell technology is controlled by a few companies such as Sunpower and Panasonic, etc. Capex. We estimate capex for mono c-si wafers at around US$0.18/W, which is comparable to cell capex and is much higher than module. Therefore, apart from the existing manufacturers capacity expansion plan, we have not seen new players planning to enter the mono wafer market. Goldman Sachs Global Investment Research 22

23 Exhibit 38: Currently wafer has higher unit capex than cell/module Solar value chain and capex Solar value chain Polysilicon Wafer Cell Module Capex US$ 0.21/W US$ 0.18/W US$ 0.15/W US$ 0.05/W Source: Gao Hua Securities Research. Exhibit 39: Mono wafer manufacturing process Mono-crystalline silicon process: CZ/MCZ methods Polysilicon Melting Seeding Pulling Mono-crystalline silicon ingot Mono wafer process Mono wafer Polishing Etching Lapping Slicing Source: Gao Hua Securities Research. Scale. Due to wafer s high capex, fixed cost such as depreciation take up a larger part of wafer s production cost compared to cell/module (Exhibit 40). Therefore, the larger the wafer capacity, the lower the unit fixed cost will be, which would be another obstacle for the new joiners. According to our estimates, Longi (mono wafer) and GCL-Poly (multi wafer) s wafer non-silicon cost has decreased along with their capacity expansion (Exhibit 41). Goldman Sachs Global Investment Research 23