Prices for silicon based solar photovoltaic products declined dramatically in 2011. From peak levels at the end of 2010, average selling prices ("ASPs") for components among the four major crystalline solar verticals dropped by as much as 70%. Polysilicon ASPs declined from over $80/kg to under $30/kg while silicon wafer ASPs went from as much as $1.00/watt to as low as $0.30/watt throughout the course of last year. Solar cell pricing also experienced severe depreciation from $1.50/watt to ranges around $0.50/watt. Finally, solar modules which are the end product in the supply chain witnessed ASP declines of approximately 50% to levels just under $1.00/watt by the end of 2011. While individual component pricing declines may seem inline with industry trends, the pricing spreads between corresponding verticals reveal more detailed information about the current state of the industry.
Examining the pricing spreads between adjacent verticals is important because it reveals the gross margin level for each vertical. As the industry has shown in the past, no individual vertical has been able to maintain a disproportionate level of profitability for an extended period of time relative to other main verticals. Due to the rapid changes of supply and demand dynamics for each of the four major crystalline verticals, any single vertical has only experienced gross margin advantages or disadvantages for brief periods often spanning only two or three quarters. Thus understanding which verticals have been experiencing distorted margin levels can help predict where overall pricing may trend in the intermediate future.
The furthest upstream vertical is polysilicon production which in recent years has been the most under supplied segment of the value chain. Due to its technological and capital intensive nature, polysilicon production came from only a handful of global manufacturers. Supply shortages not only caused incumbents to expand capacity but also led to the introduction of new producers mostly based in Asia. According to GCL-Poly (3800.hk), the largest polysilicon producer in the world at the end of 2011, global supply will approach 280,000 metric tons this year which is enough for over 40GW of silicon based solar cell production. With global installations around 27GW in 2011 and expectations of 25-30GW in 2012, once severe shortages in polysilicon have turned into oversupply. From as high as $400/kg in 2008 to $80/kg levels at the end of 2010, spot market polysilicon pricing fell further during 2011 to end below $30/kg.
While most large scale polysilicon producers do not disclose exact production costs, a range of $20-25/kg has been indicated in the past. GCL-Poly reported a blended production cost slightly above $20/kg in 2011 although through higher scale and horizontal integration, costs have further been reduced to below $20/kg. It is also important to note that because of its high capital costs, a significant portion of polysilicon's fully loaded costs is comprised of non-cash depreciation. Thus while selling at below fully loaded costs is not sustainable in the longer term as it discourages reinvestment, polysilicon suppliers in the short term could sell at negative corporate margins because the cash cost of production can be as much as 20-30% lower. While newer Chinese entrants such as Daqo New Energy (NYSE:DQ), Renesola (NYSE:SOL), and LDK Solar (NYSE:LDK) reported US GAAP fully loaded production costs around $30/kg, each company would likely maintain production at full capacity despite market pricing around $25/kg because cash cost of production is closer to $20/kg.
Although the polysilicon vertical appears to be oversupplied for another couple of years, or until global demand reaches 40-45GW, current spot market pricing around $25-28/kg should be sustainable in the near term. Break even gross margin at the fully loaded level should discourage new expansion, but still allow efficient manufacturers to maintain operations as long as market pricing is sufficiently above the cash cost of production. Based on the vertical's existing capacity, it would appear pricing may remain under pressure at current levels until global demand surpasses 45GW or in a less likely scenario sufficient appreciation in downstream pricing.
The ASP for a solar module has stayed steady at around $0.90/watt in recent months as reported by PVinsights.com[pvinsights.com]. With its adjacent solar cell ASP of approximately $0.50/watt, the gross margin spread for a single vertical module manufacturer is about $0.40/watt. Large scale module producers such as LDK Solar reported in its Q3 earnings conference call a solar cell to module conversion cost of $0.30/watt. Although integrated module suppliers like Suntech Power (NYSE:STP), Trina Solar (NYSE:TSL), and Yingli Green Energy (NYSE:YGE) have not recently broken down single vertical module conversion costs, past indications mirror LDK's metric with variations of $0.05/watt in either direction. Thus assuming an average $0.30/watt module processing cost for a cost efficient manufacturer combined with an additional $0.04-0.05/watt non-production related costs such as warranty and freight expenses, a competitive single vertical module supplier could generate roughly $0.05/watt gross profit at the recent spot market pricing spread between solar cells and modules. In short, the module vertical appears to be operating at historical and more importantly sustainable levels.
Operating metrics at the single vertical cell level may be more distorted in contrast. Large scale cell manufacturers like JA Solar (NASDAQ:JASO) have reported cell processing costs under $0.20/watt in the past. Although less efficient suppliers such as China Sunergy (NASDAQ:CSUN) historically reported cell processing in the range of 0.22-0.28/watt, most large scale single vertical cell producers disclosed conversion costs around $0.18-0.20/watt. With a pricing spread of $0.20/watt between silicon wafers and solar cells, the majority of single vertical solar cell manufacturers appear to be operating at no gross margin levels. Still predominately a solar cell manufacturer, JASO's fourth quarter gross margin of 0.5% verify recent spot market quotes and pricing spreads. On average, it appears the majority of the single vertical solar cell companies are not profitable enough at the gross profit level to sustain corporate profitability.
An obvious explanation why the single vertical solar cell industry is in its current state is over capacity. The boom during 2010 led many cell manufacturers to expand capacity dramatically. An example how cell manufacturers may have misjudged the industry's supply and demand curve is seen in JASO's actual 2011 performance vs. the expectations the company gave at the end of 2010. While JA Solar's annual shipments did grow 15.8% on an annual basis to 1.69GW in 2011, it was far below the company's original guidance of "at least" 2.2GW. JASO's optimistic view at the end of 2010 led the company to target cell capacity expansion targets towards 3GW which is almost double continued downwardly revised annual shipment guidance of 1.6GW offered in its third quarter 2011 earnings report. As the world's largest silicon solar cell company, JASO misjudged demand by as much as 50% or greater.
Large scale single vertical solar cell suppliers in Taiwan fared much worse than JASO as many reported annual revenue declines by as much as 80% in extreme cases. The situation for single vertical peers in China appeared similar based on conference call remarks by Suntech Power and Trina Solar, but is less verifiable vs. Taiwanese counterparts because many smaller solar firms in China are not public companies. Overly optimistic projections led to the over expansion of cell capacity which not only generated a supply glut but also weakened the financial strength of its suppliers. Corporate losses and capacity shut downs followed weeding out less competitive players in this segment. Inventory liquidations by defunct manufacturers further compounded pricing pressure. The distorted pricing spread between silicon wafers and solar cells suggests consolidation at the solar cell vertical is still ongoing and has not been fully played out.
Based on recent spot market pricing, the silicon wafer vertical is perhaps the most dysfunctional. Before this vertical can be examined in greater detail, it is important to break down its pricing aspects such that ASPs can be compared on an apples to apples basis. Nearly all solar companies which discuss wafer pricing on a per watt basis. However wafer transactions between companies are always quoted on a per piece basis. Because every company makes different efficiency assumptions, per watt pricing statements may differ even if companies transacted at the same per piece price.
For example, LDK Solar and Renesola quoted third quarter 2011 wafer ASPs of approximately $0.50/watt and $0.54/watt respectively. At first glance, it appears Renesola sold its wafers at an 8% pricing premium to LDK. However both companies use different wattage assumptions per wafer. Renesola assumes each wafer generates 3.90 watts while LDK assumes 4.11 watts. Translating both company's per piece ASP results in $2.11/piece and $2.06/piece for Renesola and LDK respectively. While Renesola still enjoyed slightly over 2% ASP premium vs. LDK, it wasn't as large as the 8% gap represented at the per watt level.
Since LDK Solar has not yet reported its fourth quarter 2011 earnings, small assumptions have to be made in order to compare metrics with Renesola's already reported Q4 earnings. In the fourth quarter, Renesola stated it was able to reduce wafer processing costs from $0.23/watt to $0.20/watt. In LDK's third quarter, wafer processing costs were stated at $0.20/watt with a target of $0.18/watt. For this example, LDK's fourth quarter wafer processing costs are assumed to be reduced slightly to $0.19/watt, slightly above its targeted range and slightly below levels in the prior quarter. Because LDK and Renesola are the world's second and third largest wafer suppliers respectively, both should be among the production cost leaders due to scale and thus represent a good baseline for determining the industry's wafer production costs.
With the remaining wafer cost component being polysilicon, a constant polysilicon procurement cost assumption must also be made. Using recent polysilicon spot market prices of $26-27/kg, a wafer's polysilicon per watt cost is roughly $0.14/watt. Thus with wafer conversion costs detailed above, LDK and Renesola's real time wafer unit cost is about $0.33/watt and $0.34/watt respectively which translates to $1.36/piece and $1.33/piece respectively. Yet, Q4 2011 spot market wafer pricing ranged around $1.10-1.20/piece. GCL-Poly, the world's largest silicon wafer supplier, implied wafer ASPs around $0.32/watt in its Q4 2011 earnings report which adjusted for higher efficiency assumptions equated to about $1.36/piece. Unlike the solar cell vertical where the industry's most competitive manufacturers could at least generate neutral gross margin, silicon wafer suppliers would operate at negative margins if company pricing levels mirrored spot market rates. Because depreciation only represents $0.02-0.04/watt, spot market wafer ASPs have also been below most of the industry's cash cost of production.
The industry's three largest silicon wafer producers, GCL-Poly, LDK Solar, and Renesola accounted for combined capacity as high as 14GW at the end of 2011 vs. overall industry installations of 27GW last year. Yet none of the three sold its silicon wafer production at spot market pricing which in the case of LDK and Renesola represented below cost levels. This implies the remaining portion of silicon wafers sold were most likely below cost liquidations. Portions of this inventory liquidation may originate from less competitive wafer manufacturers shutting down operations, such as once incumbent industry leader REC. Most likely the bulk of these liquidations came from downstream solar cell manufacturers that expanded and built up inventory for business expansion which never materialized last year.
Whatever the cause, spot market silicon wafer pricing has been so distorted larger and more integrated companies like Canadian Solar (NASDAQ:CSIQ) and Hanwha Solarone (NASDAQ:HSOL) stated in recent earnings conference calls to idling internal wafer capacity because procuring wafers at spot market pricing was lower than internal manufacturing costs. Other US listed Chinese solar peers such as JA Solar and Daqo New Energy also wrote down wafer capacity assets in the fourth quarter of 2011. Obviously pricing at levels below even the industry's most competitive companies cannot be sustained indefinitely. As soon as inventory liquidations level off, market pricing will likely return to at least the cost levels of the industry's lowest cost manufacturers.
The million dollar question is when will pricing spreads return to sustainable levels? This is difficult to predict because the industry's recent consolidation is still ongoing and may continue for as long as required to weed out less competitive companies. Until this cycle has mostly been resolved, industry margins at every vertical will likely remain at razor thin non-reinvestment levels - high enough for the most efficient companies to generate adequate operating cash flow, but low enough to slowly choke off higher cost structured and/or weaker balance sheet peers.
In the near term, the most distorted pricing among the industry's major verticals should also be the ones which experiences pricing normalization sooner. Either pricing for silicon wafers must increase to the production cost level of its most cost effective producers, or manufacturing processes improved to reduce costs towards spot market pricing today. Whichever the event, near term earnings for large scale wafer suppliers such as Renesola and LDK Solar may be furthest from normalized levels and thus could see the greatest magnitude of margin increase from recent levels once the industry's pricing spreads between individual verticals returns to a more proportionate ratio.