Production and Cost, 2009 Forecast
Table of Contents
1 PV Through 2012: The Anatomy of a Shakeout
1.1 Introduction: Will 2009 Be an Inflection Point for PV?
1.2 Report Methodology and Scope
1.3 Key Findings
1.4 Report Structure
2 PV Technologies
2.1 An Introduction to PV
2.2 A short History of PV
2.3 The PV Value Chain
2.4 PV Technology Options
- 2.4.1 Crystalline Silicon (c-Si)
- 2.4.2 Thin Films
- 2.4.3 Multi-junction and Concentrating PV
- 2.4.4 Emerging Materials
- 2.5.1 Commodity Materials
- 2.5.2 Polysilicon
- 2.5.3 CdTe Feedstocks
- 2.5.4 CIGs Feedstocks
- 2.5.5 Amorphous Silicon Feedstock
2.7 Why and When Does Efficiency Matter?
3 Manufatcuring Production and Capacity
3.1 Production vs. Capacity
3.2 Actual Versus Producible Supply
3.3 Estimation Methodology for Capacity and producible production
3.4 PV Industry in 2008
3.5 Worldwide Projected Capacity
3.6 Worldwide Producible Production
- 3.7.1 Cells and Module
- 3.7.2 Wafers
3.9 Production by Technology
3.10 Polysilicon Supply as a Limiting Factor for PV Production
4 Manufacturing Costs and Prices
4.1 Modeling Costs
4.2 Module Cost Structure
4.3 Feedstock Price
4.4 Conversion Non-Cash Costs
4.5 Conversion Cash Costs
4.6 Fully Loaded Module Manufacturing Costs and Prices
4.7 Costs by Technology/Location, Across Time (2008–2015)
- 4.7.1 Crystalline Silicon Technologies
- 4.7.2 Thin-Film Technologies
4.9 Costs and Prices by Time (2008–2015)
- 4.9.1 A Note on Prices
- 4.9.2 Price and Cost Estimates, 2008
- 4.9.3 Price and Cost Estimates, 2010
- 4.9.4 Price and Cost Estimates, 2012
- 4.9.5 Price and Cost Estimates, 2015
5.1 Adjusting Our Technology/Location-Based Costs for Scale
5.2 Global PV Module Supply Stacks, 2008–2012
5.3 Normalizing Supply Stacks for Efficiency Differences
5.4 Normalized Supply Stacks, 2008–2012
6 Concluding Thoughts
7 Company Profiles
Adema (E-ton), Aleo solar, Ascent solar, Avancis, Bangkok solar, Baoding Tianwei Yingli, Best Solar Hi-Tech, Bp Solar,
Calyxo GmbH (Q-Cells), Canadian Solar, CEEG Shanghai SST, Centrosolar, China Sunergy, Conergy, Daystar Technologies, Delsolar, E-TON Solar, EPV Solar, Ersol Solar, Evergreen Solar, First Solar, Flexcell (VHF Technologies SA), Fuji Electric,
G24 Innovations, Gintech, Global Solar, Gloria Solar, Glory Silicon Energy, Green Energy Technology,
Heliovolt, Honda Soltec, Hyundai Heavy Industries, Isofoton, JA Solar, Jiangsu Shunda, Jinglong Solar, Kaneka Silicon PV, Kyocera, Kyungdong Photovoltaic Energy Corp. (KPE), LDK,
M. Setek, Masdar, MEMC, Miasolé, Microsol International, Mitsubishi Electric, Mitsubishi Heavy Industries, Moser Baer, Motech, Nanosolar, Neo solar, Neosemitech, Nexolon, Ningbo Solar Electric, Pevafersa, Photovoltech, Photowatt, Primestar Solar (GE), PV Crystalox, Q-Cells, Qs Solar, REC, Renesola,
Sanyo, Scheuten Solar, Schott Solar, Shanghai Comtec Solar Technology,
Sharp Showa Shell Sekiyu, Signet Solar, Sino-American Silicon, Smart Applications, Solaicx,
Solar Semiconductor, Solar-Fabrik, SolarDay, Solarfun, Solargiga Energy, Solaria Energia, Solarworld, Solibro GmbH, Solland Solon solyndra, Sovello (Everq - GMBH), Sulfurcell, Sumco, Sun Well Solar (CMC), Sunlight Group, Sunpower, Suntech Power, Sunways, Symphony Energy,
T-solar, TATA, BP Solar, Titan Energy Systems, Trina Solar, Tynsolar, Wacker Schott Solar, Webel Wuerth Solar GmbH, Yangguang solar
PV THROUGH 2012: THE ANATOMY OF A SHAKEOUT
E.1 Will 2009 Be an Inflection Point for PV?
The story of pv’s meteoric rise in recent years is by now well known. Having languished through the 1980s, it was resurrected during the mid-1990s as rising energy prices, growing concerns about the impacts of climate change and the depletion of fossil fuels drove governments around the world to promote the adoption of solar generation on a large scale. The results have been there for all to see. Fueled by aggressive policies and generous subsidies in Germany, Japan, the u.s. (especially California) and spain, worldwide demand for photovoltaics has grown at a remarkable pace over the last decade: global installations have ballooned from a mere 125 MW in 1999 to 4.5 GW in 2008, a CAGR of 47 percent for the last 10 years (see Figure E-1).
Figure E1As policy-led demand outstripped the ability of manufacturers to keep up, module prices reversed a decades-long trend of declining prices and increased from 2004 to 2007 by as much as a third. simultaneously, manufacturing costs continued to fall steadily, driven by continuous echnological innovation and process improvements, while profit margins increased dramatically for manufacturers. Not surprisingly, manufacturers both old and new announced gigawatts in capacity expansion plans and vC capital flowed to new technologies in order to cash in on this rapidly growing and lucrative market. Although some industry watchers and participants expressed concern over the potential for a resulting oversupply, those concerns were dwarfed by the lure of profit potential and easy access to investment capital. Most assumed any supply-demand imbalance would be relatively modest, and would be quickly absorbed by further growth in the marketplace. With attractive project returns, fossil fuel prices on the rise, and supportive policy, the market believed that strong secular pv demand growth was here to stay.
But that was then, and this is now. over the last six months, the global pv market has witnessed a perfect storm of headwinds emerge: a global financial crisis has unfolded; the world’s largest economy has officially entered into a recession; equity markets have shrunk and credit markets have tightened severely; spain, the fastest growing pv market in 2008, has lost its solar appetite, and other markets could follow suit. The industry has been thrust into a period of uncertainty: Amid the news of slowing demand, idling lines, and cancellations of capacity additions, concerns over modest oversupply have escalated into fears of a full-blown shakeout, and a growing number are convinced that 2009 will finally see pv transition from a period of secular growth to a cyclical downturn.
Our work at GTM Research and The prometheus Institute, therefore, was cognizant of the singular and unprecedented array of questions and challenges facing the industry. Accordingly, this report and its sister publication—2009 Global pv Demand Analysis and Forecast—are far more than an annual update on the state of the market. Although technologies, supply, costs and manufacturers are discussed in great depth as always, we have radically rethought both the modeling and the meaning of our analysis to account for these changing market conditions. our ultimate goal is to comprehensively lay out the causes and implications of recent dynamic shifts within the pv industry over the next few years. specifically, we seek to identify which technologies and companies will be in a position of strength to weather the storms, and those perhaps most likely to be at risk from a variety of global shakeout scenarios.
E.2 Report Methodology and Scope
In driving towards answers to the questions above, accurate estimates of supply, demand and prices are crucial. Most “traditional” analyses of these key variables, however, are unlikely to be robust enough for the job. Essentially, in the past, “supply-constrained” world, market analyses often consisted of little more than applying growth rates consistent with recent trends to historical data, assuming that past trends serve as precedent for the future.
The flaws intrinsic to such an approach are all too apparent, now more than ever, when market
dynamics look to be headed towards an inflection point: They do not take into account suppliers’
response to changing market conditions and cost improvements (on the supply side), as well as interest rates and shifting government incentives (on the demand side). They cannot adequately deal with the issues of over-capacity that a “demand-constrained” world necessarily entails. Moreover, they are not sufficiently granular to provide insight into how specific technology options or individual companies will fare in the event of a downturn in the business cycle. shifting times, therefore, call for more rigorous and accurate measures.
Our approach to providing the necessary insight was to construct module supply and demand curves for different periods from the ground up—company by company, technology by technology, year by year—and estimate clearing prices based on reconciling these, creating global supply stacks that demonstrate relative ability to meet necessary price and performance characteristics that the market desires.
This report focuses on the improved methodology and conclusions on the pv wafer, cell and module supply side of the problem, including current and future costs, production capacities and potential production volumes aspects. Its final goal is to build global module supply stacks from 2008 through 2012 with all technologies—including those with different performance and value-propositions to the customer—being represented within a single global framework. Each of the steps required to do so was dealt in a separate section in the report. In order, these are:
- Estimating capacity and producible supply by company
- Estimating company-specific manufacturing costs based on technology, location and manufacturing scale
- Attaching costs to producible supply to construct supply curves
- Adjusting those supply curves to account for differences in efficiency and value to make supply curves representative of real customer allocation decisions
While demand limitations, and the resulting conclusions, will be dealt with in 2009 Global pv Demand Analysis and Forecast, the polysilicon forecasts independently developed in our prior report, polysilicon: supply, Demand and Implications for the pv Industry can be incorporated to show how feedstock availability will act as a gating factor for actual production, at least through 2010. Figure E-2 displays global producible pv module output by technology, factoring in polysilicon constraints. This chart represents the true supply potential of the global pv industry after feedstock limitations are included.
Figure E2
E.2.1 Estimation Methodology for Capacity and Producible Production
This section details the methodology used to project company-specific capacity and producible
production volumes through 2012 (wafers, cells, and modules). The following are the key features of the estimation process:
- As a starting point, primary data collection was undertaken, using a combination of surveys,publicly available data (press releases, announcements, investor presentations, and filings),and communications with company representatives.
- At the conclusion of the primary data collection, a partially complete set of capacity data was obtained. The “holes” originated from two sources. In some cases, primary data was available for only one or two out years, with gaps between the most recent historical year (2008) and then. Here, data for the years falling in between was estimated using linear interpolation.
- Secondly, our forecast period was out to 2012, and few companies are willing or able to project out that far with any accuracy. In many cases, decisions about manufacturing scale in 2012 will not be made for a few years, and will incorporate much more relevant data about market conditions, costs, and capital availability at that time. Here, we applied a capacity “ramp multiplier” to project capacity beyond the last available year’s data. This multiplier is meant to incorporate factors such as the state of development, projected demand for the technology offering, and limits to growth the company might encounter.
- For companies participating in multiple components of the value chain (wafers, cells, modules), often data was provided for only one of these (most commonly cells). To estimate manufacturing capacity for other components in such cases, we used appropriate derate (i.e., conversion) factors where applicable to account for yield and efficiency losses (5percent for wafer-to-cell and 12.5 percent for cell-to-module).
- The last step was derating the announced capacity data to obtain final estimates. As many capacity announcements were made prior to evidence of deteriorating market conditions as a result of the credit crunch and a global macroeconomic slowdown, it was necessary to temper the announced capacities with a bit of realism. Thus, the capacity estimates made per the methodology above were derated to account for probable delays and cancellations as a result of suppliers’ response to changing market conditions and possible difficulties obtaining financing. As most capacity expansion for the next two years is already in progress, derates were focused on years 2010 and beyond.
- Once this was done, we estimated producible volumes (or production) from capacity. In doing so, we assumed: (1) complete utilization of the previous year’s ending capacity; (2) linear ramp of new capacity additions over the course of the year; (3) a linearly increasing production run rate from the added capacity; and (4) some effect of downtime and imperfect yields. Figure E-3 displays the various steps in the modeling process graphically.
E.2.2 Modeling Costs
The key variables that determine manufacturing cost for pv are technology, location, and manufacturing scale. To account for variations in technology and location, seven different pv archetypes or “buckets” were created, and each component of the module cost structure was estimated independently for a standardized 250-MW manufacturing line to facilitate apple-to-apple comparisons. scale adjustments were then made on a company-specific basis and will be examined in a later section. The key considerations in creating this system of classification were twofold:
- Exhaustiveness: It should be possible to classify major pv manufacturers under this system as belonging to a particular archetype without difficulty.
- Exclusivity: The archetypes thus created should be sufficiently different so that no ambiguity exists in classifying a manufacturer.
- A global vertically integrated (polysilicon to module) multicrystalline silicon manufacturer
- A European multicrystalline manufacturer, integrated from wafer to module
- An Asian multicrystalline wafer-to-module manufacturer
- A high-efficiency monocrystalline (or “super mono”) wafer-to-module manufacturer
- A CdTe-based module manufacturer
- A CIGs-based module manufacturer
- An amorphous silicon-based module manufacturer
Figure E-4 summarizes the cost archetypes, along with examples of major manufacturers coming under each bucket.
Figure E4
E.2.3 Module Cost Structure
In order to understand the relative cost structures of manufacturing modules for the various existing pv technologies, it is important to break down and examine the major components.
Where crystalline silicon-based technologies are concerned, an intuitive framework for module cost structure is to break it out by costs incurred at various stages of the value chain—i.e., feedstock, wafer, cell and module costs. However, since thin-film modules are generally manufactured in a continuous process, a more useful framework that provides for an apple-to-apple comparison is detailed below:
Figure E5
- Feedstock price is simply the price of the material comprising the absorber layer, which would be polysilicon for the crystalline silicon technologies, and the relevant thin film for thin-film technologies.
- Conversion non-cash cost is the capital expenditure for the equipment, depreciated over its useful life.
- Conversion cash costs include all other costs incurred excepting feedstock and capital costs.
E.2.4 Adjusting Our Technology/Location-based Costs for Scale
As mentioned earlier, the three most important determinants of pv manufacturing cost are technology, location, and scale. By modeling costs by technology and location for a standardized line size (which allowed for apple-to-apple comparisons), our cost analysis thus far has only taken the first two into consideration, which means it is leaving out a key factor – the benefits of economies of scale. All else equal, a company that has a manufacturing capacity of 1 GW should have a lower cost per watt than a company with a capacity of 100 MW, as the fixed costs are spread over higher volumes.
To connect our cost estimates to company-specific producible volumes, it is thus necessary to adjust our technology/location-based costs for manufacturing scale. To do so, we took the variable component of our “raw” cost estimate for a given year as is, and scaled the fixed cost component by projected manufacturing capacity for that year relative to the standardized line capacity of 250 MW. We then attached these company-specific scale-adjusted costs to our projected producible volumes to obtain the supply stacks for different years.
Shyam MEHTA, GTM Research
Travis BRADFORD, The Prometheus Institute
About the Authors
Shyam Mehta
Shyam Mehta is a senior Analyst at Greentech Media, focusing on global solar markets. Before joining Greentech Media, shyam was a Financial Analyst at Goldman sachs Global Investment Research where he covered equities in the alternative energy sector, primarily solar companies. prior to Goldman, shyam was a Research Analyst at the Brattle Group, an economic consulting firm, where his work focused on problems within the electricity industry. shyam received his Bachelor’s in Mathematics from UC Berkeley.
Travis Bradford
Travis founded the prometheus Institute in 2003 prior to founding the prometheus Institute, Travis was a partner at steel partners II, LP, a hedge fund based in New york investing in publicly traded and privately owned businesses. In this capacity, Travis served as a board member and active management participant in businesses ranging from industrial filters to fertilizer distributors.
Travis is the author of solar Revolution: The Economic Transformation of the Global Energy Industry (MIT, 2006). He has worked for the Federal Reserve Bank, has lectured at top universities including Columbia university, Duke university and New york university on finance and entrepreneurship, and is co-author of a paper in the Journal of Applied Corporate Finance entitled Private Equity: Sources and Uses. He is also a partner at Atlas Capital, a hedge fund based in Cambridge, MA.
Research Assistance
Roger NAUTH, GTM Research
David J. LEEDS, GTM Research
