You know and I know it: thin-film PV modules have to be cheaper than crystalline silicon, and it's the efficiency, stupid. With regard to efficiency, thin-film module prices require a downwards adjustment relative to c-Si to overcome the efficiency-induced BOS cost differential; the exact penalty depends on the BOS price (in dollars or euros per square meter), as well as the efficiencies of the technology in question. The figure below displays the effect of efficiency and BOS price on module prices for CIGS and CdTe as a percentage of a 14% c-Si module price. As shown, an 11% CdTe module would have to sell at around a 14% discount to drive the same installed system cost, while in the case of an 11% CIGS module, the discount is around 8%. As a real-life example, integrators and installers indicated in June 2009 that First Solar would have to sell for a discount of around EUR 0.20-0.25/Wp (around $0.28-0.35/Wp) to be competitive with c-Si modules at the time, which amounts to a discount of EUR 0.06-0.08/Wp ($0.09-0.11/Wp) per point of efficiency difference.
Source: juwi Solar
Less well-known, and offsetting the impact of BOS and efficiency, however, is the variable of performance. Due to superior temperature coefficient and spectral sensitivity, thin-film technologies generally have better kWh-per-kW energy yield than c-Si technologies, which gives them an advantage in driving a lower delivered cost of electricity. Until recently, this factor was not really incorporated in module pricing, as reliable data on performance were not available for thin film systems; moreover, changing the focus from $/Wp to $/kWh requires a level of sophistication that many distributors and developers heretofore have lacked.
Source: National Renewable Energy Laboratory (NREL), GTM Research
In early 2010, however, juwi Solar, a large German developer, looked at operating data from its many CdTe installations and attempted to incorporate these into an analysis of rational pricing. These results are presented in the figure below. As can be seen, when kWh/kW performance is taken into account along with efficiency/BOS impacts, thin-film module prices require much less downward adjustment relative to c-Si. An 11% CdTe module here needs only a 6% discount to be competitive with 14% c-Si (as opposed to 14% discount when only efficiency/BOS is considered).
Source: juwi Solar
The graph below presents the same data, this time with module prices shown in $/Wp, and adding amorphous silicon, whose pricing curve is slightly below that of CdTe. As shown, assuming a Chinese multi c-Si module price of $1.75/Wp (representative of Q2 2010 pricing), CdTe would sell for $1.65/Wp -- a discount of only $0.10/Wp, compared to the $0.28-0.35/Wp that would be required if only efficiency is being taken into account. Similarly, a CIGS at 12% would be priced at $1.61/Wp. In the case of amorphous Si, a 9% tandem-junction module would be at $1.39/Wp, while a 7% single-junction module would only fetch $1.15/Wp -- undoubtedly below or close to cost for most manufacturers, which suggests that single-junction a-Si is not a competitive option against Asian c-Si.
Source: juwi Solar, GTM Research
Aside from a rare few, however, most thin film manufacturers still have to offer meaningful discounts over and above that required by the technological differences described above. This is necessary to offset concerns of bankability and product reliability, as many thin film companies are smaller and less established, and field data is not yet available for their modules. In fact, as the graph below makes clear, even First Solar was not immune to this effect in the not-too-distant past. The graph shows four series:
1. Actual historical prices for Chinese multi c-Si (yellow)
2. Actual historical prices for First Solar (blue)
3. First Solar ASP based on BOS/efficiency, but not performance (red)
4. First Solar ASP based BOS/efficiency, and performance advantage (green)
First Solar was selling its modules at a hefty discount all throughout 2007 and 2008, far below the expected ASP just based on BOS/efficiency differences, ignoring performance (the red line). This was done to drive market traction and get the company's modules in the field, and the high utilization drove further cost reductions through scale. By early 2009, First Solar was an established and bankable producer, and when the market turned and c-Si prices dropped by 25%, First Solar saw ASP declines of only 5%, sharply reducing the discount offered. Although protection through long-term contracts was also responsible for softening the company’s ASP drops, prices would certainly have been re-negotiated if they were not competitive. Indeed, this was done indirectly in Q2 and Q3 2009, as aggressive pricing by Chinese firms forced First Solar to establish a rebate program which retroactively set its module ASP in sync with c-Si prices. At the same time, the spread between actual and “expected” prices continued to drop, and by Q3 2009, ASPs were in line with the efficiency-based required price.
Since then, First Solar ASPs have remained stable even as c-Si prices continued declining, to the point where in Q1 and Q2 2010, ASPs converged to the expected performance-based selling price -- a discount of only 4% to 7%, compared to nearly 40% back in 2008. This signifies both the pricing power enjoyed by First Solar in a market needing bankable product, as well as the fact that its customers have woken up to the additional value offered by CdTe’s superior energy yield, and are willing to pay accordingly. Going forward, as the thin-film PV market continues maturing, one would expect this to be the case for other firms and technologies, as well.
This is an excerpt from GTM Research's recently published global PV supply chain report, PV Technology, Production and Cost Outlook: 2010 - 2015. The report addresses technical characteristics, production/capacity volumes, facility-specific manufacturing costs, supply-demand dynamics, and competitive positioning across all relevant PV technologies and nearly 200 wafer, cell, and module firms. For more on the report, go here.