At Solar 2006, the annual meeting of the American Solar Energy Society there was much talk of the shortage of polysilicon wafers, which are used to make the dominant type (crystalline) of solar Photovoltaic panels. The other major use of polysilicon is for computer chips, which has been the dominant use until recently.
I sat in on the following panels where the industry was discussed:
The CEO spotlight, where Goran Bye, the CEO of REC Silicon; as well as the “Market Status and Trends” panel, where several of the panelists discussed the polysilicon supply. In particular, Hilary Flynn presented a paper by herself and Travis Bradford entitled “An assessment of global silicon production capacity and implications for the PV industry.”
According to Flynn and Bradford, the silicon wafer industry is highly consolidated, with the following major players: Hemlock, Wacker, REC Silicon, Tokuyama, MEMC Semiconductor, Mitsubishi, and Sumitomo having about 99% of the market in 2005. Demand currently far exceeds supply, with more polysilicon being used in 2005 than was produced, with the excess being a drawdown of inventories, recycling, or an artefact of inaccuracies in the sampling method, or a combination of those factors.
From my own reading, there is much anecdotal evidence of the polysilicon supply shortage, with PV manufacturers scrambling to tie up contradts for supplies sufficient for their projected production, and even some failing to do so, with MEMC even reneging on an agreement with Evergreen Solar to supply them.
However, the silicon processing is extremely capital intensive, with long lead times, and all the major manufacturers are announcing large planned expansions to investment, and several new players are also entering the industry. There is a long lead time between when a plant is announced, and when it come on line, so the silicon market is likely to remain out of balance, with extremely high prices and profits for silicon processers through both 2006 and 2007, assuming 30% growth in PV production.
It seems unlikely to me that PV manufacturers will be able to continue to make up the polysilicon shortfall in 2006 and 2007 from inventory (although no one seems to know what inventories are, except that they’re small, hence supply of PV growth will be constrained below 30%.
Hence I expect all polysilicon manufacturers to be very profitable through 2007, with prices beginning to subside (and perhaps crash) in 2008-9. A crash of polysilicon prices would be facilitated by overbuilding of producers, combined with less than anticipated demand from PV manufacturers. This might be aggravated if one of the other PV technologies (CIGS, CdTe, or Amorphous silicon) were to grab market share from crystaline silicon due to price breaktroughs and constraint in the supply for polysilicon. Both CIGS and CdTe have the potential to be cost competitive with crystalline silicon (“PV value chain supply and demand challenges” Booz Allen Hamilton, presented at the conference), but will probably be constrained by the limited supply of Indium (CIGS) and Tellurium (CdTe) both of which are very rare. As an aside, this might produce a large opportunity for investment in mining companies with large proven reserves of Tellurium or Indium– if one of these technolgies make much more rapid strides than crystalline PV.
There are two technologies in use by polysilicon manufacturers. The most common is the Siemens process on which the majority of production facilities are based. This process is extremely energy intensive, about 10x more so than the other technology, fluidized bed, which is currently used by MEMC and REC Silicon plans to use in it’s new capacity. Most other manufacturers seem to be sticking with the Siemens process, most likely due to patent issues. For this reason, my favorite silicon manufacturer is MEMC, since their less expensive process is likely to make them better able to weather a crash in the price of processed silicon in 2008 or 2009. In the next year or two, I think most players in the industry will probably continue to benefit in terms of profits, although much of this may already be reflected in their stock prices.