Analysts have taken a more critical look at Applied Materials’ (AMAT) success with its SunFab line of turnkey module production facilities. New orders for the company’s production line have plummeted, and existing deals have been scaled back as clients moderate expansion plans in the face of massive module oversupply. As a result, rumors began circulating earlier this month that the company was contemplating an exit from its thin-film silicon (TF-Si) business – though the company has only made veiled comments to date about its plans.
The market had been quite optimistic about the AMAT solution when silicon was scarce but, as increasing polysilicon supplies have pushed contract prices to $75/kg, AMAT’s x-Si module prices have fallen to $1.95/W. If average polysilicon prices to device manufacturers reach $45/kg, modules prices could drop to $1.50/W or lower, which could enable cost of goods sold (COGS) as low as $1.05/W. Despite the obvious and significant threat to thin-film silicon’s viability, the current desire to bury the technology is as much of an overreaction as the initial hype was. There is significant room for TF-Si in the long-term technology mix − maybe just not AMAT’s variant.
Our outlook has been more measured. In the report Solar State of the Market Q1 2008: The End of the Beginning (client registration required), we projected that TF-Si would lose market share as polysilicon prices collapsed in 2009, and then rebound considerably in the future as economies of scale kicked in. Indeed, TF-Si stalwarts Sharp and efficiency leader Kaneka see this reality and are continuing to invest strongly in the technology, with Sharp supporting a 1 GW facility in Sakai, which has on-site silane and glass manufacturing. The facility recently started production with 180 MW of initial capacity using equipment from Tokyo Electron – with the status of the remaining 820 MW to be determined. Further, AMAT’s arch-nemesis Oerlikon has achieved commercial efficiencies over 10% on modules from its TF-Si tool − giving it a significant cost edge over AMAT, which has been consistently unable to break 9.5%, though it is currently struggling with its own financial issues.
Since we estimate SunFab’s TF-Si panel manufacturing costs at $1.40/W, including depreciation, $45/kg polysilicon would wipe out smaller-scale AMAT clients. In fact, Sunfilm, one of AMAT’s first SunFab customers recently fell by the wayside. We have heard of similar troubles with other AMAT clients and expect announcements of bankruptcy or production shutdowns soon. A few have quietly happened already. AMAT’s SunFab, with its lower efficiencies, unwieldy and damage-prone large module size, and expensive capex, is among the most vulnerable, but TF-Si will still grow on the backs of Sharp, Kaneka, and others to a 2.4 GW market in 2015.+
Even so, don’t count AMAT out: The company has a strong technical pedigree. Plus, while it’s overreached with SunFab, it has room to correct the flaws with its approach, as equipment upgrades will be a continual need for TF-Si, even as many module makers fail (see section 4.2 of the report Solar State of the Market Q3 2008 - client registration required). What’s more, with its experience riding out up-and-down cycles in the display industry, it’s not likely to get spooked by the current thin-film silicon panic. And even if its TF-Si business does fail or gets radically scaled back, it still wields considerable weight in the x-Si value chain and is likely working quietly on CIGS, OPV, and other technologies in the background. It has lost a battle, but certainly not the war.
