Process and product innovation as well as continuous improvement have always been at the forefront of a company’s competitive strategy. This holds – with different degree of intensities – for commodity and speciality product producers alike. But what happens if the speed of innovation is suddenly being accelerated by external factors? The transformative impact of such a regime change can currently be investigated in the polysilicon market, which has turned into a much more dynamic place in conjunction with the rise of the crystalline silicon photovoltaic industry.
Over the past decade incumbent polysilicon producers with a long-term track record have been blessed with phenomenal demand growth that turned into a double-edged sword as resulting capacity additions finally lost track with market requirements at the beginning of 2011. A prolonged and devastating overcapacity-induced price decrease followed with an average annual price decline of 30% over the past four years, leading to a total price decline of 75% (Figure 1).
The oligopolistic polysilicon market structure changed as new manufacturers, particularly from Asia, entered the playing field. Consequently, the number of high-volume and high-quality producers increased with some new entrants even catching up to the former elite of “top tier” manufacturers. This resulted in a greater than 10-fold increase in top-tier manufacturing capacity to an absolute amount of 320,000 MT/year in the period from 2005-15e (Figure 2), or more than 46GW of c-Si module equivalents (taking account of material needs from the semiconductor industry). Theoretically, this capacity alone could be fully utilised if PV end markets develop as expected by most market observers in 2015.
Under such circumstances lower tier “reserve” capacities satisfy marginal demand in cases where facilities have been in continuous operation or idled for future usage on short response. This reserve capacity could amount to approximately 148,000 MT over the next two years, or 23.9GW of c-Si module equivalents. The exact figure and timing depends on project progress of sites that are currently under development. Roughly 50% of this lower tier reserve capacity is operational and therefore imminent shortages of polysilicon, other than “higher tier” market tightness, is not likely to occur over the next 18 months.
Polysilicon reinvestment price
Nevertheless, in markets where higher tier suppliers operate close to full utilisation, price volatility typically increases, sending various signals to active and inactive producers. Here, the reinvestment price level is of particular interest, meaning a sustainable price that is sufficient to satisfy broader industry capital return requirements on expansion projects.
Despite notable improvements in energy consumption, in particular within the TCS formation and deposition steps, as well as design and process optimisation efforts with corresponding decreases in the capital cost components, the average market prices of +/- US$20/kg over the past 30 months seem far too low to justify a broad based polysilicon expansion programme – at least for mainstream direct and hydrochlorination Siemens reactor based processes.
In order to derive the relationship between unlevered project IRRs and sustainable reinvestment ASPs, certain assumptions have to be made to apply to a steady-state-free cash flow to firm discounted cash flow model. These postulates relate to critical input factors such as cash and capital costs. Polysilicon cash-costs are highly dependent on location specific factors, with electricity and metallurgical grade silicon being the biggest cost components. Best-in-class top-quartile cash costs from large-scale industry leaders range from US$15-20/kg.
On the capital investment front, a benchmarking of 50 greenfield, brownfield and existing site expansions over the past eight years gives a pretty good macro view on capital budgets for polysilicon plants. The figures are based on communicated budgets and do not include unexpected or unaccounted cost. Figure 3 shows that the majority of polysilicon manufacturing investments were budgeted at or above US$80 per kg nameplate capacity, which will be used as a mid-point benchmark for the determination of reinvestment price levels.
For modelling purposes, the allocation of investment between equipment and other fixed capital structures like building, facilities & auxiliaries have been set to 40/60 and the depreciation periods are 15 and 30 years, respectively. In addition, a 7.5% SG&A surcharge and 30% corporate tax rate are used for the determination of free cash flows. This prepares the ground for the deduction of the unlevered project IRR and ASP relationships for the polysilicon industry.
Figure 4 reveals this relationship for constant cash-costs of US$15 per kg and various unit invests of US$60/80/100 per kg. As shown, at a minimum investor hurdle of 10% (x-axis) the sustainable reinvestment price is in the mid-twenties for the US$60 per kg capital budget line and approaches thirty for the 100 USD per kg capital budget line. The same approach for the different capex scenarios at a cash-cost of US$20 per kg is revealed in Figure 5. Here the polysilicon reinvestment price range varies between the upper twenty to mid-thirty dollar levels.
Figure 6 plots out the relation for the two extremes, with the continuous line representing different project IRRs and sustainable ASP combinations for the low points for cash cost and unit invest (DS 15.0; 60), while the dotted line draws the same relation for the “high” input factor combination (DS 20.0; 100). In any case, it can be observed that, even if top-quartile producers accept a 10% unlevered IRR hurdle, polysilicon prices would need to be significantly higher than current levels (by a magnitude of 25% to 50%!) in order to jump-start further expansion rounds.
Putting the hardly digestible finance mumbo jumbo aside, it’s time to give some thought to the implications for the PV downstream. Increases in raw material prices can likely not be passed on to the final consumer, at least not to the full extent, as the industry needs to be cautious about the economics of utility-scale power plants, which are a significant pillar of global demand.
Technological progress on the cell level with advanced design concepts that fit as upgrades into the existing capital stock usually exhibit a very strong influence on costs per watt-peak and therefore stand ready to compensate a margin squeeze from raw material price increases. However, further progress is needed and in this instance the focus should be on efficient material usage, as pure play wafer manufacturers stand to lose most. Here, kerf recycling, transition to thinner wafers and casting concepts are of principal interest.
Polysilicon manufacturers have certainly made a great leap forward in bringing capital and operating expenditures down in response to subdued ASPs. However, the scenario analysis shows that the downstream should not take comfort in ultra-low polysilicon prices, as overall industry utilisation rates (including second tier producers) are likely to lead to rising feedstock prices over the next 18 to 24 months.