|
Arch: Can UMG Bounce Back?
Upgraded Metallurgic Grade (UMG) silicon was a good investment story when polysilicon was trading at sky-high prices. Unlike polysilicon which is produced using chemical processes (e.g., the Siemens process), UMG silicon is produced using a physical process. The purity of UMG is 99.999% (5N), while that of polysilicon is 99.9999% (6N), but the cost of UMG is significantly lower. In Q2 2008, when the polysilicon price was $380/kg, the price for UMG was only $70-80/kg.
Canadian Solar (Nasdaq:CSIQ) was the first Chinese solar company to adopt UMG material. The company launched UMG solar modules, eModule, in June 2008. The efficiency of UMG modules is generally 1-1.5 percentage points (pp) lower than multi-crystalline modules, so there is a 5-10% discount on price per watt; non-silicon cost (process cost) is also slightly higher. But the material cost was so low that CSIQ announced that the gross margin of UMG products exceeded 40% in Q3 2008!
Trina Solar (NYSE:TSL) was the second to join the UMG party, launching MeSolar modules in November 2008. By then, however, the polysilicon price had already dropped, and MeSolar failed to attract investors' attention.
The polysilicon price dropped to $120/kg in February and keeps falling. Some analysts predict that by the end of 2009, the price will fall below $80. As a result, many believe the UMG story is dead.
On the other hand, the price of UMG is also dropping steadily, currently at around $40-45/kg. Furthermore, efficiency still has the potential to rise, and the process cost is likely to drop as the technology matures. At least, my friends in JACO Solarsi, the biggest UMG silicon producer in China, believe so.
JACO said the price of UMG silicon is $45/kg now, still 65% lower than polysilicon, and, when the polysilicon price drops to $80 in H2 2009 or 2010, UMG will be at $30. Cell efficiency of UMG-made cells is 1 pp lower than multi-crystalline cells, but the gap could be narrowed to 0.7%-0.8% in the near future. The process cost is also expected to drop gradually and be in line with that of multi-crystalline modules in the long run. As a result, JACO estimates that the cost of UMG-made modules is 13%-14% lower than multi-crystalline modules now, and will be at least 8.8% lower in the long run.
While I do not think the process cost of UMG products will come down that much, I believe the efficiency of UMG cells/modules will rise as the purity of UMG silicon improves and some cell/module makers, such as CSIQ and Jetion (LSE:JHL), spend heavily on R&D for UMG cells.
I also believe that the cost and price of UMG modules will always be 5-10% lower than polysilicon modules -- not a huge difference, but big enough for products that are becoming increasingly similar.
So, will investors buy the UMG story again?
That doesn't seem to be the case for CSIQ, whose stock price has dropped 85.5% in the last six months. It also doesn't seem too likely for TSL, who hasn't mentioned its UMG products since launch.
UMG does look somewhat promising for two other companies, though: JACO, who received a $10 million direct investment from HSBC in November 2008, and Jetion, who is planning to launch UMG cells/modules.
2009 will be an important year for UMG, but, to find out who will win and who will lose, we just have to wait and see.
|
Arch Pei works with our sector research team with a focus on China's solar industry.
