Canadian Solar (CSIQ) has reported its results for the full year, and for the fourth quarter. The company sold more than 1.3GW modules on the year and brought in $1.8B in revenue, a 63% increase in MW sales. Financially, Canadian had a loss of $90M or $2.11 per share in 2011, while the overall gross margin was 9.6%. The company guided 2012 module sales into the range of 1.8 to 2GW. In pursuit of margins, Canadian plans to move its business into project development and full solution service, which could allow this business to grow to 25% of revenue in 2012, and 40% in 2013. In my last article I had reported very impressive operational processing costs based on Canadian's PowerPoint presentation shown at the Jefferies investor conference. The company exhibited $0.76 per watt in processing costs, based on procurement of the wafer from GCL Poly at a price of $0.31 per watt. On the conference call, CEO Shawn (Xiaohua) Qu had revealed the road map for the industry and the company, a combination of the vision and possibilities that exceeded even most optimistic views.
The majority of the companies to date have spoken of non-polysilicon costs dropping to below $0.60 by the end of 2012. Based on our discussion with Trina's IR (TSL), 70% of long-term contracts for 2012 are in the area of $25 to $30 per watt. With our own optimistic prediction of 5.5g of poly content, this would produce $0.15 per watt in cost, enabling module costs to slip under $0.73 per watt.
Canadian, however, pulled out all the stops and made a big splash by speaking of $0.55 to 0.60 per watt by the end of the 2012, including all costs. Management has placed its view on wafer cost below $0.25 per watt, and processing cost, which combines cell making and assembly of the module, at $0.30 to $0.35 per watt.
The best announced processing cost to date, for cell, is around $0.18 per watt, while that cost for module is in the area of $.22 per watt. Moving the cell processing cost down to $0.15 per watt is not unfeasible. CEO Qu spoke of the elimination/reduction of silver paste, which is not a new idea, since efforts have already been made by others. Schott had released cells made with copper metallization, made in processes already available commercially. Silver paste represents anywhere from 50% to 60% of metallization costs; by double printing that cost can be reduced by 30% using Applied Materials' cell processing system. Canadian assembles modules at $0.25 today, so moving it down to $.20, with others being at $.22, is not impossible. E.L.P.S, the most efficient of Canadian's modules, is currently assembled manually; if full automation would take place, CEO Qu sees savings of $0.05 per watt. Contrary to popular belief, the cheap labor is not the reason behind the low costs in Chinese manufacturing. Based on the above, wafer-to-module costs at $0.35 per watt seem reasonable.
While others process polysilicon ingots into wafers at $0.18 per watt, moving us to familiar territory from Trina's and Yingli's (YGE) conference calls, Canadian will simply procure the wafer from the supplier for $0.25 or less per watt by the end of the year. Today, liquidation sale wafers sell for around $0.30 per watt. Moving down those five cents is very difficult for anyone in the polysilicon business, considering that such a selling price leaves almost everyone but one company, GCL, without profit. GCL had mapped its wafer processing costs to around $0.13 per watt already, while producing poly at $20 per kg today. This is the best cost on the market, but not enough to profit with Canadian's ASP expectations. In order to achieve a 10% gross margin on wafer sold at $0.25, poly must be produced at $0.095 per watt ($0.025 GM on $0.25 is $0.13 processing, plus $0.095 poly) with 5.5 grams per watt. That means poly cost cannot exceed $17 per kg. Such a contractual ASP would give a tremendous advantage to all of GCL's clients among other players, not just Canadian, but would leave Yingli and Jinko in a corner, since they do not have contracts with GCL. GCL will be announcing its Q4 results by the middle of the month, and we will be looking for confirmation or denial of this in operational dynamic.
While GCL Poly is the world's cost exception in production of poly, the improvements in processing are only as far away as the next capital expense. Diamond wire offers a thinner cut, resulting in smaller kerf and fewer grams per watt. Another productivity improvement is achieved by cutting through large load ingots with a 100% to 200% speed increase versus the steel wire. CEO Qu spoke about 150 microns' thickness versus the standard 180 to 200 microns. Applied Materials Baccini Pegaso Cell System line can handle wafers at 120 microns thick, with 0.15% breakage for the entire process. Finally, an increase of cell conversion, in one way while applying selective emitter technology, allows for direct reduction of the module cost at a ratio of $0.10 per 1% increase.
Canadian is increasing average conversion of its cell production lines from 17% of multi, 18.5% mono and quasi at 18% by 0.5 % for each, expecting to drop module cost by an additional 5 cents. Canadian's ELPS mono cell-based module, built on MWT and all back contact architecture, offers 16.47% conversion at the high end of the product line. Up to 500MW high conversion capacity will be available in 2012, with 250MW in expected sales, bringing 8 cents to 10 cents premium per watt.
Lastly, Canadian delivered a second punch line in efficiency and cost reduction thesis. CEO Qu gave an estimate on BOS at $1 to $1.50 per watt, including $0.75 for module ASP by the end of 2012. In Germany installations on utility-scale projects are currently conducted in the area of $2.25 to $2.50 per watt, while in the U.S. it's more like $3.25 to $3.50. It is widely known that costs of installation need to follow the dramatic reduction in cost of modules, particularly in the retail sector.
In summary, Canadian's solar road map may seem exceptional, but only by time and not in the technical capability. While I think that solar module installation at $1.50 is more than a year away, it is possible to see pricing like that in unique circumstances on certain projects. I communicated with GCL IR in regards to wafer pricing last night. By the end of Q3, GCL cost per watt was $0.28. I was assured that GCL will provide its clients with a competitive market price; GCL did not see $0.25 as problem. The timeline, however, was not specified.