SolarCity, the largest residential solar PV installer in the US, has started down a path to become the largest domestic producer of solar cells and modules after construction of its ambitious 1GW Buffalo Riverbend Manufacturing Facility started in September last year.
Not only is the plan highly ambitious but the construction, tool install, qualification and ramp of the facility are also aggressive, especially when factoring in the use of high-efficiency cell technology from a start-up, Silevo.
This has raised a few eyebrows within the PV industry and investor community as the risks are high and the dream to become an integrated producer could end up crippling the company.
Status of PV manufacturing in the US
Without cataloguing the large number of US PV manufacturers, whether crystalline silicon-based, thin-film (CdTe, CIGS and a-Si) or CPV, that have existed the sector, gone bankrupt, closed down or been acquired, the situation has been dire for several years
Although some blamed low-cost Chinese module producers for their demise, the reality for some was simply that their technology and products were not good enough to compete in markets that few existing competitors actually competed in. Markets such as BIPV, BAPV and the commercial flat rooftop markets spring to mind.
Those that did directly compete with Chinese rivals often failed to scale production fast enough to benefit from lower production costs or simply did not scale at all, eventually succumbing to market forces.
However, market forces also included the dumping of Chinese modules in the US and European markets, according to anti-dumping investigations in the US and EU, which accelerated the rate and number of company failures.
In an era of anti-dumping duties the US PV manufacturing sector is looking to bounce back, supported by significant downstream growth in the country. PV Tech’s in-house tracking of global manufacturing capacity expansion announcements highlighted last year that a significant rebound in US-located PV manufacturing was underway.
We reported back in November, 2014 that 10 companies in the US had announced production plans that totalled close to 2GW of new capacity, with almost half of the new capacity expected to come online in 2015. Of course 1GW of that capacity came from SolarCity, highlighting the importance the Buffalo facility has on a US-centric recovery in PV manufacturing.
It is interesting to look back at SolarCity’s rationale for entering the highly competitive PV manufacturing sector.
SolarCity chairman Elon Musk, its co-founder and chief technology officer, Peter Rive, and co-founder and chief executive Lyndon Rive wrote a blog at the time of the Silevo acquisition about this. Simply put, they believed the growth in PV installations in the US would grow significantly for years to come and that being an integrated supplier would drive overall costs down significantly.
No problem with that assumption; the Chinese producers that led the integrated manufacturing phase did indeed drive costs down and since have gone on to become downstream project developers. SolarCity, as PV Tech previously noted, has simply been the first downstream player to attempt to reverse-engineer that successful business model.
SolarCity’s management also highlighted that with future growth expected in rooftop PV installations, the high-efficiency modules needed to provide the lowest installation costs would not be available in the quantities it would require. Therefore future growth of the company determined the need to become an integrated manufacturer.
With downstream market demand slowly eating into the previously chronic overcapacity in the module manufacturing sector, helped by the shutdown of uncompetitive capacity, the move by SolarCity to be able to source product from its own facilities would seem justified.
It was well reported at the end of 2013 that ‘effective’ (saleable) module manufacturing capacity stood at around 45GW, while 2013 end market demand turned out to be around 37GW. General consensus puts 2014 end market demand at around (a conservative) 45GW, which was backed up by many tier-one suppliers being capacity constrained at the end of the year and many claiming to be sold out through the first quarter of 2015.
However, almost six months after SolarCity announced its 1GW fab, it is clear that although module supply and demand dynamics have indeed improved it’s a very fragile balance, as PV Tech’s recent 2014 full-year analysis of capacity expansion announcements indicate.
Simply put, effective capacity expansion announcements made in 2014 have kept slightly ahead of end market demand, demonstrating that many existing tier-one suppliers are not afraid to add capacity.
Within the top 10 module producers Trina Solar, JinkoSolar, JA Solar, Hanwha Q CELLS and First Solar have all announced measured capacity expansions for 2015.
We can add Canadian Solar to that list as it announced expansion plans for 2015 in the fourth quarter of 2014.
Over 10GW of c-Si module assembly capacity expansions were announced in 2014 and around 2GW of thin-film expansions. Of course not all these announcements can be classified as effective capacity (start-ups included) and not all of the capacity announcements will even start volume production in 2015.
However, there is enough effective capacity already announced in 2014 (including January and February 2015) that has already come on stream or will come on stream in 2015 to meet general-consensus end-market demand of around 53GW (still conservative) for the year.
Indeed on the thin-film front, PV Tech recently reported that First Solar may have an operating capacity of around 1.8GW but upgrades to its 30 operating lines and warehoused equipment totalling 1.3GW give it around 4GW of potential capacity. Furthermore and very importantly, a number of large tier-one module manufacturers, including which include Yingli Green, Sharp and Kyocera, have yet to announce planned capacity expansions for 2015.
Only recently, CEO of (now the largest renewable energy developer) SunEdison, Ahmad Chatila was asked by financial analysts at its Capital Markets Day event concerns over module availability for its aggressive downstream project development plans that would see the company double installations from 1GW in 2014 to around 2GW in 2015. Chatila dismissed the concerns, noting that although a fabless company there were plenty of companies able to support its downstream ambitions and dismissed the notion that it needed to own and operate manufacturing plant in the future.
The bottom line on effective capacity now and in the next few years, regardless of demand, is that many companies are willing to add new capacity and build market share. Importantly, the upstream polysilicon sector is not capacity constrained and still has idled or not yet completed capacity to come online.
Concern therefore for SolarCity has focused on such areas and the question of why it needs to enter the risky, low-margin manufacturing sector when a rival such as SunEdison could remain fabless in the future.
The mono factor
A small but very important aspect missing in this capacity analysis is that the majority of installed capacity and newly announced capacity through 2014 has primarily been related to multicrystalline. SolarCity has been specific in acquiring Silevo, which is focused on lower-cost but high-efficiency monocrystalline cells and modules, a segment dominated by SunPower and Panasonic.
Furthermore, high-efficiency, low-cost mono c-Si cell capacity in the merchant market is almost non-existent and dominated by small companies or some tier-one firms with limited in-house capacity.
SolarCity has clearly done its homework on this front and realises longer term that high-efficiency, low-cost mono modules will be the preferred choice for both the residential and commercial rooftop markets as the class of module will be key to post-subsidy markets around the world and the post-investment tax credit market in the US.
The final key concern for SolarCity has been the scale of the venture. Unlike any other c-Si PV producer before, SolarCity is starting not with 100MW or 200MW of integrated capacity but shooting straight to a 1GW facility.
The company has acknowledged that Silevo had a small production line and was transferring that with its R&D tools to the larger and former Solyndra facility, implying a strategy was in place that would be similar to copy exact or copy intelligently from lab to fab.
However it is the scale of leap that is the concern as debugging and fine tuning 200MW of equipment is not the same as trying to do that with an extra 800MW, regardless of copy exact or copy intelligent lab to fab practices.
Site selection incentives
Nevertheless, to achieve competitive production costs scale will be essential; simply put, the larger the scale of production the lower the manufacturing costs while low labour costs are not a strategic factor in low cost production.
In this regard, Solarcity has picked its location well as low-cost production through scale alone is not the way to go: an isolated 1GW production plant in the middle of the Mojave Desert doesn’t work; instead such scaled production plants need a host of supporting infrastructure and clustering of support services including trained workforce, local R&D facilities and future trained employees via local feeder Universities.
Having competitors locally can also be supportive rather than negative as this helps build a robust infrastructure around such large facilities and often supports further large capacity expansions better.
The decision to locate in Buffalo, New York meets most of these key aspects, though of course several other parts of the US would also meet key site selection criteria.
Therefore the most important site selection criterion is financial support, and in this New York State seems to have trumped other attractive locations as it has done in the past with the semiconductor industry.
Initially, the Buffalo fab was expected to require capital expenditures of US$400 million to US$450 million and as far as SolarCity’s contribution, that would be in the range of a couple of hundred million US dollars.
However, on the official announcement of the plant, New York State had agreed to put US$750 million into building the facility, supporting infrastructure and manufacturing equipment. This would be fed through the State University of New York's College of Nanoscale Science and Engineering, which would own the buildings and equipment.
It was understood at the time that SolarCity would lease the facility for 10 years and over that period invest US$5 billion into running the facility.
In recent financial filings from SolarCity further details of the agreement have come to light.
The Research Foundation for the State University of New York was to construct the facility that would be approximately 1,000,000 square feet, capable of producing 1GW of solar panels annually on a site of approximately 88.24 acres at a cost of around US$350 million.
The Research Foundation would also acquire certain manufacturing equipment designated by SolarCity for the facility, which would be completed and ready to commence operations in the first quarter of 2016.
The acquisition and commissioning of the manufacturing equipment was said to cost the Research Foundation around US$400 million.
Not only are the facility and equipment being completely subsidised, but SolarCity is also leasing the manufacturing facilities for US$1.0 per year plus cost of utilities for an initial 10-year period. There would be no charge for using the equipment through the period of the initial lease.
By any measure this is an incredible deal for SolarCity but the small print raises some issues.
The US$5 billion spend in the region remains in place but penalties exist on an annual basis of US$41.2 million to the Research Foundation under a ‘program payment’ should generally accepted annual spending not be met, which includes ‘job creation obligations’. The job creation obligations are intertwined with the successful manufacturing milestones.
After three months of the facility being completed and reaching full production output within three months thereafter, SolarCity needs to have 1,460 high-tech jobs at the site and retain that minimum figure for five years.
“However, SolarCity also has to employ at least 2,000 other personnel in the State of New York to support downstream PV module sales and installations for five years as well as work with the Research Foundation to attract 1,440 additional support contractor and supplier jobs in the State of New York following the plant reaching full production.”
It is also not just SolarCity that is under pressure to meet targets. Silevo has three key milestones to meet for its investors to receive full payment from SolarCity under the US$200 million all-stock transaction agreed in the acquisition and the further US$150 million promised when fixed milestones for the project are met.
The first US$50 million ‘earnout’, which is payable before 31 December 2015, requires Silevo to complete a new R&D production facility located in the US, which is capable of volume production with the targeted cell/module efficiency requirements.
Obviously, SolarCity has not publicly revealed production cost and cell/module efficiency requirements, yet the company will have to be producing competitively priced and competitive efficiencies for kWh produced to meet its overall total system installation cost reduction roadmap.
The first earnout looked a lot more achievable when SolarCity announced recently that it would be leasing Solyndra’s former manufacturing facility in Silicon Valley (Fremont), and that Silevo’s current R&D tools and small production line would be moved to the larger property.
The second earnout requires both volume production with target efficiency requirements from the new R&D production facility in Fremont and the completion and full equipment commissioning of the 1GW Buffalo manufacturing facility. The date for the earnout is before 31 December 2017, which seems doable within less than 12 months should the facility be built and the tool install go to schedule.
However the caveat here is that the third earnout, paid on the same date considerations of the second earnout, requires Silevo to meet volume production with target cost structures and cell/module efficiency requirements from the 1GW fab.
Those concerned with SolarCity’s production plans and milestones have drawn attention to the fact that there isn’t another company or project that has been so aggressive on all the key metrics and that it is therefore unprecedented.
SunPower has also been touted as a logical benchmark to SolarCity plans and observers have highlighted that it has taken SunPower significantly longer and without competitive manufacturing cost structures to achieve 1GW of production and beyond.
However, SunPower’s technology is not comparable to that of Silevo other than the use of monocrystalline wafers and then it’s likely both will have different wafer sizes, with Silevo opting for conventional 156mm x 156mm square format.
Indeed so little is really in the public domain about Silevo’s low-cost, high-efficiency process that attempts at benchmarking are meaningless. But as tools get ordered and materials suppliers contracted a better picture is expected to emerge and then perhaps some level of benchmarking may be possible.
The bottom line for SolarCity and Silevo is that the risks of not meeting the required milestones will remain in place until they are met. There is also a level of burden that comes with those target milestones, which typically do not exist for the majority of PV manufacturers.
However, one thing is for sure and that is SolarCity’s downstream business and targeted growth. Unlike most other wannabe manufacturers, SolarCity has itself as the customer and the downstream pipeline to support a 1GW plant in 2015 and 2016.
After that, all bets are off.
SolarCity, the largest residential solar PV installer in the US, has started down a path to become the largest domestic producer of solar cells and modules after construction of its ambitious 1GW Buffalo Riverbend Manufacturing Facility. Image: SolarCity
From groundbreaking in September 2014 to completion is a tough task.
SolarCity has enough demand to fully utilise the 1GW fab.